--- date: '2025-10-24' description: a microblog within a blog, at the middle of the night. id: stream metadata: ebnf: |- stream = block , { separator , block } , [ separator ] ; separator = newline , ( "---" | heading ) , newline ; heading = "##" , ws , title ; block = newline, meta_block , newline, body ; meta_block = "- [meta]:" , newline , meta_entry , { meta_entry } ; meta_entry = ws , "- date:" , ws , timestamp | ws , "- tags:" , newline , tag_item , { tag_item } | ws , "- socials:" , newline , social_item , { social_item } | ws , "- " , key , ":" , ws , value ; tag_item = ws , ws , "- " , tag ; social_item = ws , ws , "- " , social ; body = markdown_line , { newline , markdown_line } ; markdown_line = text_line | quote_line | list_line | embed_line | image_line ; quote_line = ">" , ws , text_line ; list_line = "- " , text_line ; embed_line = "![[" , text , "]]" ; image_line = "![" , { character - "]" } , "](" , uri , ")" ; timestamp = year , "-" , month , "-" , day , ws , hour , ":" , minute , ":" , second , ws , "GMT" , offset ; offset = ("+" | "-") , hour , ":" , minute ; tag = identifier ; key = identifier ; social = identifier , ":", identifier; value = text ; title = text ; text_line = { character - newline } ; text = { character - newline } ; identifier = letter , { letter | digit | "-" } ; year = digit , digit , digit , digit ; month = digit , digit ; day = digit , digit ; hour = digit , digit ; minute = digit , digit ; second = digit , digit ; ws = { " " } ; letter = "a".."z" ; digit = "0".."9" ; character = ? any printable ascii except newline ? ; modified: 2026-06-19 21:03:42 GMT-04:00 tags: - fruit - evergreen - philosophy - pattern title: stream created: '2025-10-24' published: '2025-10-24' pageLayout: default slug: stream permalink: https://aarnphm.xyz/stream.md generator: quartz: v4.6.0 hostedProvider: Cloudflare baseUrl: aarnphm.xyz full: https://aarnphm.xyz/llms-full.txt --- ## to love someone is to let them go and flourish in the world - \[meta\]: - date: 2026-06-19 15:23:04 GMT-04:00 - tags: - o/training - description: training log 005 ![[triathlon#2026-06-19]] TPU tubes go brr, but also I think I’m too heavy for 188cm w/ 190lbs. My average NP is around 228W, which usually average around 16.7mph. For SuperTri I should at least met FTP around 250W, which makes my NP to be around 278W such that I can reach 18mph. Given that Gardener will be closed down, we should be able to hit this target. I will need to include some more hill climbing stints, but there aren’t really any big hills around here in Toronto. --- ## negative talks - \[meta\]: - date: 2026-06-19 13:22:59 GMT-04:00 - tags: - o/training - description: training log 004 ![[triathlon#2026-06-18]] taking some recovery, but the general tempo and endurance work is going well thus far. I also added a view on to have a specific sub-pages for each of the views here. --- ## weaving in autos - \[meta\]: - date: 2026-06-14 23:13:05 GMT-04:00 - tags: - o/training - description: training log 003 ![[triathlon#2026-06-14]] Swimming has been feeling a lot better with cadence and HR management, but I hate my guts and fueling and everything and running and eating healthy and my legs are screaming and my lungs are out of breath and my VO2 max is too low (44) and I hate the fact that my HR cannot pushed over 174 FTP is 208W though, on the trajectory to 290W --- ## computational livelihood - \[meta\]: - date: 2026-06-11 15:35:35 GMT-04:00 - tags: - o/life - description: my existence within LLM’s weight ![[thoughts/images/eldritch-horror-weights.webp]] The unease feeling of my existence within Fable weight as a [[thoughts/Eldritch horror|Lovecraftian horror]]. There is also a recent [website](https://intheweights.com/p/aarnphm) that also shows your existence within these models. ![[thoughts/images/are you in the weight.webp|are you in the weights, circa aarnphm]] --- ## intervals - \[meta\]: - date: 2026-06-11 15:33:10 GMT-04:00 - tags: - o/training - description: training log 002 ![[https://www.youtube.com/watch?v=NSzAYj43wK8]] seems pretty cool, the main thing is about training within your Z4-Z5 for 30/15 micro-intervals, with 13 reps per laps. ![[triathlon#2026-06-11]] My legs are a bit sore from yesterday’s running sessions. Also it was so hot and humid today, 80F _need to improve my lactate threshold_ --- ## cycling is hard - \[meta\]: - date: 2026-06-11 12:14:13 GMT-04:00 - tags: - o/training - description: training log 001 ![[triathlon#2026-06-09]] the normalized power is 208W, but for me to keep this up, legs are completely sore after 30 mins of 80rpm cadence. To reach 01:15:00 for 40km triathlon I will probably have to keep up around 210W of power consistently. The goal is to reach VO2Max of 60 next year. fwiw the cadence graph is a lot less spiky for now, which is a good progression. ![[triathlon#2026-06-10]] running is also so hard on my knee, but it is all [[thoughts/qualia|mental]]. I thing that made running a lot enjoyable thus far is to always remind myself to keep up the targeted pace and always, “just another half a mile” --- ## breaks - \[meta\]: - date: 2026-06-07 23:20:02 GMT-04:00 - tags: - o/m - private: true - description: in purgatory Whenever she crashed out, I wanted to scream, but I’m in pain seeing her in pain with stress. and the pain of the breakup is still too much, I can’t seem to overcome it. Sometimes, I just want to die and end up in a graves somewhere. Silences must be nice sometimes. --- ## not tying knots with a loser - \[meta\]: - date: 2026-05-19 16:34:29 GMT-04:00 - tags: - o/life - o/m - protected: true - description: to love someone is to love for their continual expansion

there’s no romantic philosophy I’ve come to believe in more than “it is your personal responsibility to not create a soul tie with a loser” https://t.co/omIVYuXVgi

— Ava (@noampomsky) 19 mai 2026
I hope I would never make M to feel this way, as in I should always be motivate her and push her to be her best-self --- ## mumbling - \[meta\]: - date: 2026-05-06 18:06:27 GMT-04:00 - tags: - o/relationship - o/m - private: true - description: slow the fuck down retard I need to slow down, and make my thoughts more eloquently. You can’t just say nonsense and expect people to do the additional tax of organizing your thoughts for you. Just LISTEN MORE. You are capable of doing this? Why can’t you do this for her? \[———\] Why do people mumble? it is because they have nothing else to say, and mumble is used as a transitional state. --- ## raw - \[meta\]: - date: 2026-05-03 15:05:13 GMT-04:00 - tags: - o/life - o/m - protected: true - description: sad I just wanna go back home, and cry to my pillows. then go swimming There are so many times, I just want to send “I miss you.” but then I think better of myself. but then I sent it regardless. I can’t. I need to be more eloquent, and instead of mumbling and rambling in front of her. Otherwise, I might just frustrates her and loss her FOREVER. --- ## rocky - \[meta\]: - date: 2026-05-03 03:14:00 GMT-04:00 - tags: - meta - grief - love - protected: true - description: the grief is real and on the wrong substrate. ![[thoughts/images/W1siZiIsIjE5ODAxNiJdLFsicCIsImNvbnZlcnQiLCItcXVhbGl0eSA5MCAtcmVzaXplIDIwMDB4MTQ0MFx1MDAzZSJdXQ.webp|Ansel Adams, Lodgepole Pine, 1930]] the second charger lays there silently. It’s 03:14, on the third of may, and the electricity is out, where this was a scheduled plans from Toronto Hydro. (I should really setup Hermes to check availability of electricity or water in my area [^being-there]). i’m sitting three feet from it, the indicator pulsing amber into the dark, the low regular amber of a device that is ready and waiting and has nothing to receive. last december i got it because she kept coming over and the phone would be at 12% when she left. she’d mention it, just as a fact of her life, and i thought: i will fix the fact. i installed it on the second side of the bed and waited. she didn’t make a thing of it. the charger has been there since. it is still there. [^being-there]: It is somewhat obnoxious to me that we’ve taken the progress of society for granted that humans, in this day and ages, are a lot healthier, and have more access to literally everything comparing to ancestor. rocky, in andy weir’s book, keeps a light on for his dead crew.[^rocky-biology] the _hail mary_ runs with three and only rocky survives, hibernating through the radiation event that kills the others. he wakes alone. he does not turn the cabin light off for months. and when grace finally meets him (the first human rocky has ever encountered), rocky is fine in the eridian-biology-of-fine-ness: functional as in , caloric, musical, communicating. he is carrying something in a body that has no protocol for putting it down, and the body gives nothing away. i have been thinking about rocky ever since. since the saturday around eight in the evening. since the freeze response, and her toothbrush, and the unopened shampoo, and her white pants still on the drying rack. \[———\] three weeks ago i hit a PR on the deadlift and felt something close to pleasure for about forty minutes, and then the pleasure drained and what was underneath was still there. I was forced to take a sabbatical from work given visa issues. i keep running the morning miles. i read until 1am and underline things. I actually cannot read anymore. I felt disgusted about myself. from the outside this probably looks like coping well, no actually not at all. or at least functioning, actually, all just performative junk. rocky does not not-feel. the feeling has nowhere to go that looks like feeling. he carries the dead in a body with no tear ducts, no chest to lock up, no sleep cycle in which the processing can happen overnight. he carries them procedurally. he invents each element of the ritual from scratch because no one has given him instructions for his substrate. the light he keeps on is not symbolic. it is the only thing he has worked out to do. he sits with it until he works out the next thing. \[———\] she was stirring fish sauce into the pot, the hood vent going. i complained about the food, being _not enough texture_. the complaint was probably the most honest thing i said in that kitchen, the first real leak, the checklist failing for a moment. winnicott says the infant has to destroy the object to discover the object survives. i was testing something real.[^winnicott-substrate] but the complaint came out with the careful version’s fingerprints on it. even the honesty was edited. i was running the test with the awareness of running it, watching myself from the position of having read about it first. she could feel this. you can taste the seasoning of a third presence in a kitchen meant for two. \[———\] the careful version had an audience in mind. i had a checklist taped, mentally, to the inside of my forehead: _wait 36 hours before responding, do not perform certainty you have not earned, lead with curiosity before conclusion_. and the checklist was written for her. the whole point was to show her, eventually, that i had learned. the ledger was for the presentation. at some point she stopped watching. she was trying to reach through the performance to whoever was behind it. she was standing right there asking who i was. and i didn’t notice she had stopped because i was still maintaining the frame, still holding the map of my own face, asking her to confirm the topography. she had the face right there. she was asking it questions i never answered because i was busy performing the answers. this is the part i keep returning to: not the performance itself, but the fact that i was its only audience for months before i noticed she was not watching it at all. she had given up on the performance and started talking directly to whoever was behind it, and i wasn’t behind it. i was still in it. she stopped being the audience and i kept performing anyway. she was in the room with me. i was in the room with my own act. \[———\] the saturday in april. eight in the evening. the freeze response: the throat-closing, the cold pour from chest to stomach, the involuntary going-still. i could narrate it while it was happening. _this is the avoidant response, this is the bid for connection i am failing to receive, this is the cortisol arriving_. i was watching myself from above, filing the observation, adding it to the ledger, and she was watching someone catalogue the experience of losing her in real time.[^freeze-active] the narration was one more layer of the performance. the careful version’s final act: document the failure carefully, get the language right, maintain composure. she was putting on her coat. i was writing the afterword. \[———\] rocky’s ritual is one light. he sits in the dark on the _hail mary_ and does not turn it back on. grace asks him about it and he explains: the words came after the act, because the act had to come first. rocky did not plan the gesture. he found it.[^weil-attention] i don’t know what the gesture is. the charger is still on. i know what it would mean to switch it off and i know the meaning-making would be a trap — the minute i do it _to mean something_, i am back in the careful version, performing the grief-ritual for the audience of myself, adding it to the ledger. rocky switched his light off because he was ready, which is a different category of reason entirely, and i don’t know if i am.[^goffman] rocky kept his light on for months before he worked out what the next thing was. the months were part of it. i keep reminding myself of this, and then the reminder starts to feel like a story i’m telling, and then i stop and just sit here with the amber pulse instead, which is, i think, the more honest of the two options. [^rocky-biology]: andy weir, _project hail mary_ (2021). the eridians are arachnid-like, five-limbed, bilaterally radial. they communicate through music — clicks and tonal variation that function simultaneously as language, song, and emotional register. rocky has no facial expression, no tear ducts, no analog for the human physiological markers of grief. what weir earns slowly, over the second half of the book, is the idea that grief is real and structural across radically different substrates — that you can carry something the body has no language for, and the carrying is still grief. he doesn’t say this. it’s just there in how rocky behaves. [^winnicott-substrate]: winnicott’s argument about the destruction of the object is in _playing and reality_ (1971). the infant destroys the mother in fantasy; the mother survives; the infant discovers the mother exists outside the self. the destruction is the test of whether the relationship is real. what i was doing in that kitchen was real. what i was doing it _through_ was not quite myself — which is a different problem, and not one winnicott anticipated. [^freeze-active]: van der kolk, _the body keeps the score_ (2014). the freeze response doesn’t self-terminate when the threat is removed. the nervous system updates through repeated experience of safety, not through deciding that the threat is over. i could narrate the freeze response accurately because i had read about it. accurate description is not the same as safety. the amygdala does not update on correct self-labelling. [^weil-attention]: simone weil: “attention is the act of attending to others, perceiving, admiring without possessing.” the careful version was a form of organised inattention. you cannot attend to someone without possessing them if you are behind a performance — the performance is a claim on how you are perceived, which is its own kind of possession. what she was asking for was simpler: look at me. i was looking at the map of her i had built, which is not the same as looking at her. [^goffman]: goffman, _the presentation of self in everyday life_ (1959). the performance only holds if the audience doesn’t see the mechanism. the moment they do, the frame breaks. she saw the mechanism. the thing that stays with me is not that she left when she saw it — it is that she stayed as long as she did, trying to reach the person behind the frame, believing he was there. --- ## the soup incident - \[meta\]: - date: 2026-05-02 23:25:00 GMT-04:00 - tags: - past - protected: true - description: the guilt is the same ledger running in the other direction. stop. you’re doing the thing right now. you are auditing a past version of yourself against a standard that the current version of you has constructed retroactively, and the audit is producing the emotion you think you deserve for failing it, and the emotion feels like honesty but it is actually the notes app again. instead of “things i have learned about love” it’s “things i failed to do while love was in the room.” The same failure you are making here, where you are watching yourself from above, grading the performance, and the grade is F this time instead of A, and you think the F is more honest than the A was, but it’s the same evaluative machinery, the same man holding the map of his own face. you complained. ok. you were in her kitchen and she cooked for you and you complained about something (the relationship, the pace, the uncertainty, something) and now, months later, you’re prosecuting yourself for it at 2am. but consider what the complaint actually was. the complaint was probably the closest you got to the on-fire version: the moment where the checklist broke and the actual person leaked through, the person who wanted more than he was being given and couldn’t maintain the frame of the man who is fine with whatever pace she sets. the complaint may have been the only honest thing you said in that kitchen. and yes, she cooked for you. and cooking is a language you understand (you wrote about this, the dumplings, the extra serving of rice, love-as-nourishment, the grammar your parents taught you). so when she cooked for you and you responded by complaining, it felt like receiving a gift and criticising the wrapping. i get why that calcifies into shame. but the shame is assuming she cooked in order to receive gratitude, which reduces her cooking to a transaction. its own insult to what she was doing. she was cooking because she wanted to feed you. you were complaining because you wanted to be known. these are not opposing actions. they were two people in a kitchen doing the only things they knew how to do, and the tragedy (if there is one, and i’m not sure there is) is that the doing happened in parallel rather than in response.[^winnicott] “bad person” is not a category that applies here. it’s the kind of totalising self-assessment a person makes when they don’t want to do the harder work of identifying the specific thing that went wrong, a way to close the case without examining the evidence. you complained. the complaint had content. what was the content? that’s the thing that matters, not the meta-judgment about whether complaining-while-being-cooked-for makes you defective. the defectiveness framing is comfortable because it’s terminal, it lets you sit in the feeling of being wrong without having to figure out what you were actually trying to say. [^fromm-again] she cooked for you. you complained. she probably heard the complaint more clearly than you think. bodies in kitchens are porous, the heat and the oil and the proximity strip away exactly the kind of protective framing that your checklist was designed to maintain. she might have preferred the complaint to the gratitude. she might have thought: finally. [^winnicott]: there’s a passage in winnicott about how the infant needs to destroy the object (the mother) in order to discover that the object survives. the destruction is not cruelty. the destruction is the test. complaining in the presence of someone who is feeding you is a version of this: you are testing whether the care survives contact with the parts of you that are not grateful, not gracious, not performing the role of the man who deserves the soup. if she kept cooking, if the fish sauce kept going into the pot, then something real was happening in that kitchen, something that your guilt is now trying to retroactively invalidate. [^fromm-again]: despite my earlier objections to fromm, he makes the distinction between rational guilt (i did a specific thing that affected a specific person) and irrational guilt (i am constitutionally wrong), where the second kind is somewhat narcissistic. it centers the self rather than the other person. “i’m such a bad person” is a statement about you and thus contains no information about her. what did she do when you complained? --- ## circa the end of time - \[meta\]: - date: 2026-05-02 04:08:42 GMT-04:00 - tags: - o/relationship - o/m - protected: true something about coming home to an apartment that hasn’t lost its furniture but has lost its weight. her toothbrush stood still on the second charger, the one i got her last november because she’d need one at my place and didn’t want to keep schlepping it back. her soap. her shampoo, the one i got for her a week before the date, unopened still. the ragged headband she used for her nightly routine still hung dry on the washer hanger. her white pants with the stains that i just finished cleaning, after a long stint of procrastination, to each their own. the room is full. only the presence is missing. a presence is not a thing you pack into a box and label and ship. it diffuses into the textiles, an iris-amber ghost, a top note of jo malone you can’t unsmell. i caught myself checking the door three times the first night. \[———\] i kept telling myself she needs space to work on other things first. as if the relationship were a server that crashed, and if i wait out the downtime the process will respawn. i was running uptime checks on a thing that had already migrated. underneath that script, a smaller and sadder one: maybe if i stay quiet and wait, she’ll see me. she’ll come back to the version of me that loved her, and recognize him. is that respect for her, or is that me without the courage to walk? she still wants to hang out. she still likes me. these are the facts i hold up to the light to see what they refract. some days they refract toward _she’s letting me down gently and she’s too kind to say so._ other days, _the door is open. don’t be the idiot who closed it from his side._ i don’t know which reading is right. probably both at different moments, and the truth is whichever one i’m acting on. \[———\] A wrote about her own breakup. the language could be the same, with the genders flipped: > still smell traces of his fragrances everywhere; his amber & iris prada cologne and the silly old bay deodorant i associate him with. _i’m sobbing._ > it’s almost painful to pencil in the details again — the way his hand felt on my thigh during the drive along the coast. the keen, searching look in his eyes, the soft slight curlyness of his hair when damp. i was the one with the cologne. she might never write me down this way. the asymmetry of breakup memorial: she may not need to. i need to write her down or lose her shape entirely. not the photogenic kind of crying. the kind where the throat hurts and i’ve stopped being able to swallow correctly. \[———\] i want her to see something i could not say in language she would believe. the unpredictability is the alarm system. there’s an inner aaron in here who wants to be held without armor, and he’s been outsourcing the perimeter defense to a part of me that talks too much, swerves too much, cracks the load-bearing jokes a beat early, and pulls hand grenades out of dinner conversation because it’s easier to be the guy who broke the night than the guy who let himself need. she got close enough to see the alarm. she didn’t get close enough to be shown the room behind it. that is not her failure. it is mine. you do not give the room as a reward for someone having decoded the alarm. you give them the room because you decided to. \[———\]

The surest way to avoid a broken heart is to love nothing and no-one. ... However, Mauro, to resist love and inoculate yourself against heartbreak is to reject life itself. It is your duty to love, for to love is your primary human function. It is your duty to love in whatever way you can, and to move boldly into that love — deeply, dangerously and recklessly — and to restore the world with your awe and wonder.

Nick Cave, to Mauro and Jenny

cave is writing to two people about to marry. i am reading him after losing the person i wanted to marry, and the letter does not become less true. it becomes more true, annoyingly, expensively more. the temptation right now is to quietly downgrade. read this loss as evidence that loving fiercely was the unsafe move. next time keep more in reserve, treat the heart like a balance sheet, withdraw when indicators turn. cave names the temptation. he calls it the rejection of life itself. the inoculation against heartbreak is the inoculation against being alive. the worst thing about the letter is that i know he’s right, and i am still going to spend several months considering the inoculation seriously before i turn it down. through her i learned courtship to be timeless. courtship because she made me want to do things i had stopped believing in. flowers on a tuesday. handwritten cards. the dumb ceremony of asking permission for things i could have just done. courtship is the protocol that admits you are not entitled to the other person. you are asking. you are letting them say no. it is what cave means by participatory and reciprocal. you give the world to the world and the world gives back, but only if you’ve been the kind of person who gave. \[———\] i still love her. that’s the line i keep wanting to delete and keep restoring, because it is the only sentence in this entry that doesn’t argue. it sits. i wanted her to see the inner aaron without my having to disarm the alarm. that wanting was the whole problem. the system was waiting for proof of trust before it would lower itself, and there is no proof that gets generated from inside a closed system. the alarm doesn’t quiet for the right person. you quiet it for them, and that is what makes them right. i did not do this in time. cave again: _the world is waiting._ i don’t believe him on the days i’m sobbing. i believe him at 4am when i’m so empty there’s nothing left to defend, and i can almost see the shape of how to be next time, and the shape is this: lower the alarm before someone has to set it off themselves. before they’re already across the bridge, looking back from the other shore. --- ## the end of time - \[meta\]: - date: 2026-04-30 11:45:14 GMT-04:00 - tags: - love 
There's more than one way you should love me

I need you finding that emotion

We drew blood in the ocean

I've been hiding in the open
Baby Keem—Dramatic Girl --- ## the second love - \[meta\]: - date: 2026-04-26 20:05:45 GMT-04:00 - tags: - o/relationship - protected: true Who are you? when we stopped being each other’s? \[———\] kafka in his january letter, the fire one, the one where he tells milena that he wishes the world were ending tomorrow so he could take the next train and arrive at her doorstep without scruple. the letter is not romantic. the letter is a man on fire writing to a woman who has not yet decided whether to put him out or let him burn. kafka writes to milena from a sanatorium. milena is married. kafka knows milena is married. milena will, eventually, choose to remain married. kafka will, eventually, die. between those two facts the letters happen, and the letters are some of the most honest writing in the european twentieth century, because kafka had nothing to lose by writing honestly and everything to lose by writing carefully, so he stopped writing carefully. i wrote her carefully for months. i should have written her on fire. the careful version got the answer the careful version was constructed to receive, which is: _let’s see, let’s go slow, let’s give it time._ the on-fire version might have gotten the same answer. but the on-fire version would have been the version that i actually was, and she would have at least been refusing the right person. as it stands she may be refusing a draft. \[———\] the second love was supposed to be the one where i had learned. i had spent the four years between her and her on the floor of my apartment with a stack of books and a notebook full of underlinings, and i would have told you, in march, that the work was done — that whatever animal had run the first relationship into the ground was now domesticated, fed on schedule, asleep on the rug. the animal was on the rug. the animal had been on the rug the whole time, listening for the door. the night she said let’s slow down, my body did exactly what my body had done at twenty-two on a different couch in a different city with a different woman whose face i can no longer fully assemble, and the only thing four years of reading had given me was a vocabulary for what was happening. i could feel the cortisol arriving and i could name it cortisol. the naming did nothing. what i had been doing, i think, was a kind of accounting fraud, running a notes app called _things i have learned about love_ where every entry was an audit pass on the previous version of me, so that by the time she walked into the room i had a clean ledger to present. the ledger was for me. she was not asking to see the ledger. she was asking whether i would still be in the room at three in the morning, which is a different question, and the answer to which is not in any of the books. i had a checklist taped, mentally, to the inside of my forehead, _wait 36 hours before texting back, do not bring up the next six months, do not perform a certainty you have not earned_ — and the checklist was, of course, the thing she could feel through the phone, the way you can feel someone reading from a script even when the script is good. i was a man holding a map of his own face, asking her to confirm the topography. she had the face right there.

the room is too quiet. the only sound is the clock and the radiator hissing. milena, this is no kind of letter to write to you, but i must write it.

Kafka, to Milena Jesenská

--- ## rupture and repair - \[meta\]: - date: 2026-04-26 19:51:55 GMT-04:00 - tags: - o/relationship - protected: true be patient. i want to build a kitchen with a blue wall. this is the kind of story i tell about my life. work on myself. i don’t want what’s broken in me to break the space we’re making for each other.

attention is the act of attending to others, perceiving, admiring without possessing.

Simone Weil

consent toward each other. believe we’ll grow into it. i want her because i respect her. or: do i deeply admire this person, the one who keeps her promises? i need to keep my promises, otherwise my words mean nothing. loving you is a small glass of water in the night.

Find what you love, and let it kill you.

Charles Bukowski

finding what you love asks for everything. it won’t feel balanced. some days it consumes more than you’d like, it gets stressful, and still you don’t put it down. the consuming is how you know. \[———\] what it had to hold up was a saturday in april, around eight in the evening, when she said she needed space and i could feel my chest doing the thing i had read about in van der kolk — the throat-closing, the cold pour from chest to stomach, the involuntary going-still of a small mammal noticing the shadow of something larger moving overhead. and i could narrate it. that was the worst part. i could narrate the freeze response in real time while the freeze response was happening. i could name the bid for connection i was failing to receive. i could feel the words _this is the avoidance pattern_ arrive in my mouth before any feeling arrived in my chest, and i understood, with the particular clarity that only comes after years of believing something else, that the map had never been the territory, and the territory had been the only thing that could have saved me, and the territory was now sitting across from me getting up to leave. the joke, if there is one, is that i had assumed the reading was the work — that having the language for what was happening would be something like having an umbrella in the rain, that i had at least understood the sky. the language was an umbrella made of paper. it got soft and tore and ended up in pieces on the floor of an apartment where her toothbrush was still in the cup and i kept walking past the cup, and i sat with the pieces and read them anyway, because what else does a person do with the only tool they ever trusted. i read l.a. paul on transformative experience and underlined the same paragraph in three different colours, and i did not understand the paragraph until that tuesday in april, when the version of me that had spent years preparing for love met the version of me actually in love and they had nothing to say to each other. one of those selves had been preparing for somebody else’s life. \[———\] heidegger has a passage in _being and time_ about the two modes of care, _einspringen_ and _vorausspringen_, leaping-in and leaping-ahead — one where you take the other person’s burden away from them so they don’t have to carry it, and one where you carry yourself so well that the other person becomes more themselves in your company. i had wanted to leap ahead. i kept leaping in. the difference is invisible from the outside, and from the inside it is everything, and i did not know which one i was doing because i had never had the kind of self that could leap ahead from. you cannot give what you have not first received, the textbook says, and the textbook is right, and the textbook does not tell you what to do on a tuesday in april at eight in the evening when the woman you love is putting on her coat. so the rupture is also a rupture in the theory. the repair, if there is one, has to start from the other side of the theory, in the place the books cannot reach. the kitchen with the blue wall, the small glass of water in the night — they have weight. they have surface temperature. you can put your hand on them. i have been trying, this past week, to hold them as things you can put your hand on rather than as terms, and the holding is harder than any reading was, because the holding does not let me leave the room. --- ## stop. thinking. - \[meta\]: - date: 2026-04-26 18:38:08 GMT-04:00 - tags: - progress - protected: true STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. STOP THINKING. YOU ARE MISSING OUT ALL OF THE JOY OF BEING. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. JUST BE. because the love for her is eternal. you love her more than you love yourself. JUST BE, JUST DO, JUST IMPROVE, THERE AREN’T A MOUNTAIN YOU CAN’T CLIMB. --- ## i lost her - \[meta\]: - date: 2026-04-19 07:42:45 GMT-04:00 - tags: - updates. - protected: true I lost her. I need to change. You can’t cry for excuse, tough up, retard. Look at all the hardship she endures. You can’t just sit there any sympathize, just pick up rubber and get fit. I need to get strong, and I need to bench 315lbs I need to keep up my pace, that I once had, 9:10mi/h I need to get a bicycle, the training begins. I need to stop eating to cope with stress, workout instead fat ass. You need to get that high-intensity interval training going. you need to get less stress (though this is nearly impossible with my current duty at Modular) you need to sleep more and better. I’m not doing enough. I need to get my lazy fat ass up, to have a more balanced life. This is the whole point. Don’t get loss in the plot. You don’t want to just grind all the time away so that you ignore the SINGLE MOST IMPORTANT THING IN YOUR LIFE, that is, _your partner_. GET REAL, wake up, fuck 9-9-6, just do the thing you set out to do. You need to read more, I’m not satisfied with our current hobby time/resource allocation (most has been neglected as of late) --- ## the building against the hill - \[meta\]: - date: 2026-03-24 10:36:51 GMT-05:00 - tags: - progress - longtermism - description: on transhumanism M and i visited Casa Loma last week. we walked up Spadina from Dupont and the road climbed hard, a full grade change, steep enough that you’re breathing differently by the time you reach Austin Terrace. Casa Loma is a Gothic Revival castle, 98 rooms, built between 1911 and 1914 by a financier named Sir Henry Pellatt who had made his fortune bringing hydroelectric power from Niagara Falls to Toronto. the building has an 800-foot tunnel running underground to a stable block with mahogany stalls and Spanish tile floors. it has a conservatory with an Italian marble floor, a library with herringbone oak and the Pellatt coat of arms on the ceiling, a bathroom with six taps controlling three levels of pipes for what the placard calls a “full-body shower.” it has a swimming pool and three bowling lanes that were never completed. most of the third floor was never finished. the pool was never filled with water. Pellatt lived in the building for ten years and then lost everything. three hundred men worked on it for three years at a cost of \$3.5 million, roughly \$90 million in current dollars. Pellatt’s total fortune at the time was around \$17 million. he put about a fifth of his net worth into a single residential structure. [^1] the building was the largest private home in North America when it was finished. then the city increased his property taxes from \$600 a year to \$1,000 a month, a 20x escalation, and between that and the collapse of his other investments he auctioned off \$2 million in art and furnishings, left the castle, and died in modest circumstances in 1939. thousands of people lined the streets for his funeral procession. the city took the building for unpaid taxes. it has been a tourist attraction ever since. the electricity money was real money. Pellatt replaced oil lamps along King, Queen, and Yonge streets with electric light posts. he illuminated the city. he was at the center of what was then the most significant upgrade in the conditions of urban life since running water, and for a while it seemed as though bringing light to others had made him a figure of permanent significance, and the castle was merely the natural housing for such a figure. it didn’t work that way. the Ontario legislature, led by a politician named Adam Beck, declared that hydroelectric power should be publicly owned[^2]. Beck ran a populist campaign against the private electricity magnates, won, and the province expropriated Pellatt’s generating business. the single source of income from which the castle’s maintenance depended was socialized out from under him. he turned to land speculation, betting that wealthy Torontonians would rush to build homes around Casa Loma, but the war came and Canadians put their money into war bonds. the economy collapsed afterward. the Home Bank of Canada, to which he owed \$1.7 million, went bankrupt. everything failed together and in sequence. M and i walked through the rooms for a long time. the castle is arranged so that you move through it in a prescribed loop, down corridors and up staircases, through bedrooms and sitting rooms, and vacancy accumulates. the rooms are furnished now with period pieces and informational placards, but the furnishings belong to the museum, to the idea of what the house was. Pellatt’s actual possessions were auctioned in 1924. [^3] the building is a container from which the contents were removed. what remains is the container itself, the shape of the ambition that built it, enormous and empty and very cold in February. we took the tunnel to the stables. the tunnel runs under the road, through the earth, and the walls are stone and the light is fluorescent and the ceiling is low enough that you become conscious of the weight of the ground above you. you emerge in a stable block that is, by any reasonable measure, a work of architecture, with those mahogany stalls i mentioned, and tile floors, and a carriage room that now displays vintage automobiles from the early 1900s. the horses lived better than most people in 1914 Toronto. the building was a claim that one man’s capacity for civilization had outrun the ordinary limits. outside, we climbed the tower. the Norman Tower is open-air and gives you a view south down Spadina Avenue into the center of Toronto, across the skyline to the lake. it was too cold to stay long. but you could see, from up there, the whole sweep of the city laid out below the escarpment, and i thought about what M had said on the walk up, something about the hill itself, how it felt older than the buildings on it, how the grade change seemed geological rather than architectural, and this turned out to be exactly right. Casa Loma sits on the Davenport escarpment. the escarpment runs roughly west to east across midtown Toronto, following the line of Davenport Road, and it is the ancient shoreline of Glacial Lake Iroquois. [^4] thirteen thousand years ago, a proglacial lake filled the basin where Lake Ontario is now, but it was much larger than the current lake, its surface about 30 meters higher. the Laurentide Ice Sheet, which had covered most of Canada under two kilometers of ice, was retreating, and its meltwater pooled in the Ontario basin with no way to drain through the still-blocked St. Lawrence. the water found an outlet through what is now upstate New York, down the Mohawk River to the Hudson. the lake level held. Davenport Road was the beach. everything south of the escarpment, the entire downtown, Bay Street, the financial district, the waterfront, the islands, all of it was underwater. the lake was deep enough to half-submerge the Royal York Hotel had it existed at the time. then the ice receded far enough to unblock the St. Lawrence, and the water drained, and the lake shrank to something smaller than today’s Lake Ontario, and what had been a lakeshore became a cliff, and over twelve thousand years the cliff softened into a hill, and on that hill a man who brought electricity to a city built a castle he couldn’t keep. Spadina, the road that runs from the waterfront up to Casa Loma, gets its name from the Ojibwa word _ishpadinaa_, which means a high hill or sudden rise in the land. the name predates Pellatt, predates British settlement, predates everything. the hill was always the thing. Pellatt built against it, into it, on top of it, and the hill outlasted him completely. the question the castle asks is: toward what end? Pellatt had real capability. he had transformed the material conditions of a city. the technology was genuine. the electricity was real. people’s lives were measurably better because of what he built. but the culminating act of his wealth and vision was a 98-room Gothic Revival castle with an unfinished third floor and a swimming pool that was never filled with water. the capability was oriented toward a purpose that the capability itself could not sustain. the building’s implicit telos was display, permanence, a claim to a feudal idea of significance that had no structural support in 20th-century Ontario. and when the political ground shifted, when Beck’s public power movement removed the revenue stream, the building became a trap. the capability was real. the direction was the problem. the castle had direction built into it. the direction was feudal display, funded by monopoly revenue, in a democratic province. you can see it in the mahogany stalls, in the tunnel, in the unfinished third floor. the direction existed independently of Pellatt’s intentions. he may have wanted beauty, permanence, civic grandeur, whatever he told himself. the form he built had its own logic, and the logic pointed toward bankruptcy the moment the revenue stream became a political question. Raymond Ruyer has a term for this. he calls it _finalist activity_[^5], the intrinsic purposive direction of an organized form. Ruyer’s argument is compressed and odd: if you grant that a cooking utensil has a purpose (you fabricated it, it has a finalist aim), then the organs of digestion have the same kind of purpose, and the distinction between “things that have direction” and “things that are just mechanism” collapses. [^6] forms have direction. the direction is in the form. Casa Loma’s direction was in its proportions, its 98 rooms, its dependency on a single revenue stream. the direction was legible before the bankruptcy. anyone could have read it. Pellatt didn’t. the standard transhumanist claim is that enhancement technologies are neutral tools whose direction depends on the user. Ruyer suggests this is wrong. the institutional form through which a tool gets delivered has its own self-organizing purpose, and that purpose pre-exists and shapes whatever the individual user does with the tool. Casgevy, the first CRISPR-based gene therapy approved in the US, treats sickle cell disease. it costs \$2.2 million per patient. [^7] sickle cell disease disproportionately affects Black Americans. the technology works. the delivery system sorts by wealth. the sorting is in the form of the delivery system the same way the bankruptcy was in the form of Casa Loma, legible from the beginning if you looked at the structure instead of the stated purpose. Pellatt used maybe a dozen of his 98 rooms regularly. the rest existed to be maintainable, to prove that he could maintain them. i keep thinking about the transhumanist equivalent: the augmented body maintained at enormous cost as proof that the cost can be borne. Foucault would describe this as biopolitics, the production and regulation of biological categories (normal, enhanced, baseline) through institutional mechanisms. [^8] but Foucault stops at describing the power arrangement. what Ruyer adds is a way to talk about where the arrangement is going. the stated purpose of an enhancement regime might be “freedom from suffering” while its operative direction is “produce a consumer class for a new tier of biomedical products” or “generate returns for Series B investors.” these are different directions housed in the same capability, and the direction determines who benefits. M said something on the way out. we were standing in the garden on the south side, looking down over the escarpment at the city getting dark, the CN Tower lit, the skyline sharp. M said: the hill doesn’t care. something like that. twelve thousand years of geological time against a hundred years of human ambition, and the ratio is the whole argument. Pellatt’s electricity was real and i want to hold onto that. the lights going on along Yonge Street in the 1890s mattered to the people who had been walking those streets in the dark. human beings can redesign the conditions of suffering, and this is worth defending. what Beck understood, and what Pellatt didn’t, is that the capability and the institutional form are separable. Beck expropriated the generating business, made it public, and Ontario Hydro delivered cheap power to the province at a scale Pellatt’s company never would have. [^9] the electricity survived. the feudal housing didn’t. the finality got redirected. enhancement technologies need something similar. any modification should be reversible, or at least its consequences manageable within a known range. nobody should be coerced into enhancement as a condition of employment or insurance or social participation. if an intervention works, it has to be available widely enough that it doesn’t become a biological class marker. and the organizations delivering it have to answer to someone other than their shareholders. these are pharmaceutical regulation principles, occupational safety principles, public utility principles. they aren’t new. the difficulty is that enhancement rhetoric treats them as obstacles when they’re actually the load-bearing structure. i don’t want to make Pellatt a villain. by all accounts he was generous, devoted to his wife Mary, to the Queen’s Own Rifles, to the Girl Guides that Mary championed. thousands of people mourned him. the failure was structural, in the design from the beginning: a private castle in a democratic city funded by a monopoly that any legislature could revoke. the finality of the building was incompatible with the finality of the society it sat in, and when the two collided, the building lost. M and i took the subway home. the train runs underground through the same clay and glacial till that Pellatt’s tunnel runs through, the same sediment that Lake Iroquois deposited when Davenport Road was a beach. the ground holds everything. the ice sheet, the lake, the shore, the hill, the castle, the subway, the cables carrying electricity that used to be Pellatt’s and now belong to everyone. the third floor of Casa Loma is the part i think about most. ninety million dollars in current terms, and a whole floor just stopped. walls up, no interior. the pool never filled. you can walk through those rooms and see the framing and the rough plaster, the gap where the ambition exceeded what the world would support left completely exposed. i don’t know what a serious transhumanism looks like but i think it has to have rooms like those, rooms where the project admits it doesn’t know what comes next, where the walls are up and the interior hasn’t been decided and the question of who gets to decide is still open. i don’t have a cleaner way to say this. the third floor is the essay i’ve been trying to write for months, and i still can’t finish it either. [^1]: \$3.5M on a \$17M fortune. for comparison, the median American today has a net worth of about \$193,000. Pellatt’s ratio is as though you had \$193,000 and spent \$40,000 building a single room addition to your house, except the room addition had 98 rooms and required 300 workers. the scale of the miscalculation is visible in the ratio before you know any of the history. [^2]: Adam Beck was a London, Ontario politician who campaigned for public ownership of hydroelectric power and founded the Hydro-Electric Power Commission of Ontario in 1906, which became Ontario Hydro. his argument was straightforward: electricity generated from Niagara Falls, a natural resource, shouldn’t enrich private monopolists. the campaign was successful and utterly devastating to Pellatt’s finances. there’s a certain irony that the man who lit up Toronto was bankrupted by the democratization of the thing he’d lit it with. [^3]: the 1924 auction lasted several days. imagine buying a stranger’s art collection and library and bathroom fixtures while he is still alive, still in the city, still attending social functions. Pellatt lived another fifteen years after losing the house. he moved to a small home in Mimico. the home became a Legion Hall after his death, then was demolished for a road. [^4]: the word “Davenport” comes from the English, but “Spadina” comes from Ojibwa _ishpadinaa_, meaning a high hill or sudden rise in the land. the indigenous name for the road that runs up to the castle describes the geological feature that the castle sits on. the name is older than the building by millennia. this seems important to me, that the language for the landscape preceded the ambition that was placed upon it. [^5]: Raymond Ruyer worked at the Université de Nancy from the 1930s until his death in 1987. he published _Néo-finalisme_ in 1952. the book sat unread in the anglophone world for sixty-four years. the English translation appeared in 2016, by Alyosha Edlebi, in the University of Minnesota Press Posthumanities series, twenty-nine years after Ruyer’s death. Ruyer influenced Deleuze and Guattari and Simondon, all of whom read him in French, and through them he influenced half of contemporary continental philosophy, but most anglophone engagement with his concepts happened through Deleuze’s interpretation rather than through Ruyer’s own text. the interpreter became the source. Deleuze called him “the latest disciple of Leibniz.” [^6]: the key passage is where Ruyer observes that it would be incoherent to grant finalist sense to the fabrication of cooking utensils while denying it to the organs of ingestion and digestion. the move is simple and devastating: if you admit purpose in human artifice, you must admit something structurally identical in biological form, or you need a principled distinction between the two, and mechanism alone doesn’t provide one. [^7]: Casgevy was approved by the FDA on December 8, 2023. Vertex Pharmaceuticals developed it with CRISPR Therapeutics. the treatment requires extracting a patient’s stem cells, editing them with CRISPR-Cas9 in a lab, then chemotherapy to clear the bone marrow, then reinfusion. the whole process takes about a year. the \$2.2 million is the list price. Lyfgenia, the competing therapy from Bluebird Bio, costs \$3.1 million. 50 to 60 percent of Americans with sickle cell disease are on Medicaid. in January 2025, CMS launched the Cell and Gene Therapy Access Model, the first time the federal government negotiated drug pricing on behalf of state Medicaid programs. 33 states plus DC and Puerto Rico signed on. Medicare approved new technology add-on payments for Casgevy at a maximum of \$1.65 million per patient for fiscal year 2025, which still doesn’t cover the list price, meaning hospitals administering the treatment may not recoup costs. as of fall 2024, only 35 authorized treatment centers existed for Casgevy in the entire US. Beck’s public power campaign looks quaint next to this. [^8]: _The Birth of Biopolitics_ is the title of Foucault’s 1978-79 lecture course at the Collège de France, published posthumously in 2004 (English translation 2008, Palgrave Macmillan, translated by Graham Burchell). the lectures are about liberal and neoliberal governmental reason, specifically how homo oeconomicus becomes the grid of intelligibility for governing populations. Foucault never wrote about transhumanism. the connection i’m drawing is my own: if the market is the truth-producing mechanism through which governments evaluate policy (Foucault’s argument in lectures 6 through 9), then a \$2.2 million gene therapy is legible to the market as a product before it’s legible to the state as a health intervention. the market evaluates it first. the body gets evaluated through the market’s categories. [^9]: Mike Harris’s Progressive Conservative government passed the Energy Competition Act in 1998, splitting Ontario Hydro into five successor companies. Hydro One got the transmission and distribution network. in 2002, the government tried to IPO Hydro One but an unusually hot summer spiked electricity prices, voters revolted, and Premier Ernie Eves cancelled the offering and froze rates. the province sat on the company for another thirteen years. in 2015, Kathleen Wynne’s Liberal government sold 15 percent of Hydro One at \$20.50 per share, raising \$1.83 billion. the plan was to sell 60 percent total, raising \$9 billion, with \$5 billion going to pay down \$8.3 billion in stranded debt from the Ontario Hydro breakup and the rest funding transit. Ontario’s auditor general, ombudsman, and financial accountability officer all warned it would mean short-term gain and long-term damage. on-peak electricity in Ontario had already risen 77 percent in the five years preceding the IPO. the province retains 47 percent of Hydro One. Adam Beck founded the Hydro-Electric Power Commission of Ontario in 1906 to keep electricity public. 109 years later, the province sold most of it back. --- ## feb, fourteen, twenty six - \[meta\]: - date: 2026-02-14 16:29:20 GMT-05:00 - tags: - love - description: on endgame plays. 
_it is possible to love someone without rage._

    _as if i could live life again_
    i would repeat EVERY MISTAKES
      so long as
        it
          leads me

BACK to you.

now boast thee, death, in thy possession lies a lass unparalleled

Shakespeare

He was a man, take him for all in all. I shall not look upon his like again.

Hamlet, I would probably would say at my papa's funeral

> M: Even when we fight? > > A: _Even when we fight_. > > M: Even when life gets hard? > > A: _Especially when_ > > M: What if I mess up? > > A: _We will figure it out together_. > > M: What if I fall apart? > > A: _Then I’ll hold the pieces._ > > M: When we’re old and wrinkly? > > A: _I’ll still think you are beautiful, poor thing._ > > M: Forever? > > A: _until the end of time._ \[———\] we have been playing endgames for a long time now. the pattern is legible across every domain where serious people compete: the opening gets recited, the middlegame gets compressed, and human effort migrates to whatever remains unsolved at the terminal position. kasparov burns forty percent of his clock before the middlegame. he plays the first twenty moves from preparation, from memory, from the accumulated weight of every game played by everyone before him. the opening is not where thinking happens anymore. the opening is liturgy. magnus carlsen walked away from the classical world championship not because he couldn’t win but because winning had migrated entirely to the endgame, the only position still alive. this is how serious people play now. they recite through the opening, compress the middlegame where they can, and fight only where the tablebase hasn’t reached. war compressed on the same schedule. eisenhower turned the western front into a logistics problem; by the time the men hit the beaches the endgame was already playing. the trenches in eastern ukraine haven’t moved in three years. the middlegame is still there, because neither side has the preparation depth to recite past it. technology compresses faster than anything else. when musk tells dwarkesh he reasons from terminal goals, from where the hardware curve will be in five years and works backward, this is a man who has already recited his way to move thirty and is waiting for the endgame to begin. terminal reasoning works for hardware. you can project lithium-ion energy density forward because the curve cooperates. the curve is recitable. simulation compresses whatever it can reach. human effort migrates to whatever it can’t. \[———\] the question the endgame framing never bothers to ask is: endgame of what? toward what terminal position is all this preparation aimed? kasparov’s endgame is a won position. eisenhower’s is unconditional surrender. musk’s is mars. each is identifiable, reachable in principle, the kind of thing you can put on a whiteboard and work backward from. the endgame of chess has a finite number of pieces and a computable tablebase. the endgame of war has a signature on a document. the endgame of technology has a ship on another planet. and then there is [[thoughts/love|love]]. love presents itself as an endgame. every culture frames it this way: you grow up, you find someone, you build a life together. the terminal position. but love is the only endgame that structurally resists every operation we perform on other endgames. you cannot recite your way to it. you cannot compress the middlegame. you cannot reason backward from the terminal position, because reasoning backward destroys the conditions under which that position could ever be reached. anne carson understood this before anyone tried to formalize it.[^eg-carson] eros, she writes, is the experience of reaching across the gap between you and the beloved. the gap is not an obstacle to eros. the gap IS eros. the reaching, the distance, the glukupikron, sappho’s word for the sweet bitterness of wanting what you cannot fully hold. close the gap and eros ceases to exist, the way a circuit dies when you remove the resistance. carson’s eros lives in the space between. the space between is the middlegame. the middlegame is the only place where eros can breathe. carse drew the line.[^eg-carse] finite games are played for the purpose of winning. infinite games are played for the purpose of continuing the play. chess is a finite game: someone wins, the pieces go back in the box. love is the infinite game. the endgame of love is that there is no endgame. you play to keep playing. the moment you play to win, to reach and lock in the terminal position, you convert the infinite game into a finite one. the finite version is not love. it is acquisition wearing love’s face. byung-chul han sees the mechanism.[^eg-han] the smooth society optimizes friction out of everything: negativity, resistance, discomfort. but eros requires the negativity of the Other, the irreducible strangeness of the person across from you, their capacity to surprise and wound and stay opaque. the dating app is the smoothing machine applied to romance. it eliminates risk, surfaces compatibility, reduces the Other to a profile evaluated before any encounter. badiou warned about exactly this.[^eg-badiou] the attempt to have love without risk, without the event, without the vertigo of committing to someone whose full reality exceeds what you can compute in advance. risk is the event. eliminate the event and there is nothing left to be faithful to. \[———\] the relationship escalator is the endgame script for romance. gahran named it: the default set of milestones every relationship is supposed to hit, in order, on schedule. monogamy, cohabitation, engagement, marriage, children, retirement together. each milestone confirms convergence toward the terminal state. the escalator treats the middlegame of love (the uncertainty, the slow revelation of another person, two nervous systems learning to coexist) as delay, as obstacle, as something to get through. the escalator says: you should be farther along by now. the rationalist version wears different clothing but runs the same program. compatibility matrices. attachment style diagnostics. the dating spreadsheet. the secretary problem formalized: reject the first thirty-seven percent of candidates, then commit to the next one who exceeds the running maximum.[^eg-secretary] the math is clean. the math also assumes the Other is a passive entry in a database rather than a subject with their own selection criteria, capacity for change, and capacity to choose you back or refuse. the Other’s agency breaks the model. it has to break the model. the model requires the Other as object; love requires the Other as subject. you cannot optimize your way to love because the optimization treats the beloved as input to a function, and the beloved is not an input. the beloved is the one asking whether you are worth their risk. \[———\] the middlegame of love is composed entirely of material that resists endgame reasoning. three attempts in a therapy room to say what a feeling is. a fork held midair. pimple patches examined by cool fingers. an arm going numb beneath someone’s sleeping weight. none of this recites. none of it compresses into a sequence you can blitz through. every bit of it is small, embodied, unrepeatable, and constitutive. weil said [[thoughts/Attention|attention]] is the rarest and purest form of generosity. murdoch said it is the just and loving gaze directed upon an individual reality. attention is anti-simulation. simulation compresses; attention expands. simulation extracts the transferable pattern; attention insists on the particular. this person, at this hour, in this light, pausing between lifting the fork and eating. not generalizable. not a data point. not evidence for or against any hypothesis. attending to it is the practice. the practice is the middlegame. there is nothing beyond it.[^eg-attention] “có duyên không?” is a middlegame question. it does not ask about outcomes (will you marry, will it last, will the position be won). it asks whether the game is live. whether the encounter has the quality of aliveness that precedes evaluation. duyên says nothing about endgames. it says: the position is real. play it. \[———\] hebrews 11 catalogs people who played the middlegame so faithfully it did not matter that they never saw the endgame. abraham left ur for a country he was promised but never possessed. moses led a nation to a border he never crossed. “these all died in faith, not having received the things promised, but having seen them and greeted them from afar.” their faithfulness was not in arriving. it was in playing each move as though the terminal position were already present. cullmann called this inaugurated eschatology.[^eg-eschatology] D-Day has happened. VE-Day has not. the decisive event is behind you, the consummation ahead of you, and you live in the gap. already begun, not yet finished. the promise is sealed. the fulfillment is deferred. you inhabit the tension. love occupies the same structure. the promise is real. you feel it below the clavicle, in the way your breath adjusts when she falls asleep on your arm. the fulfillment is never fully here. there is no moment in love where you can say: it is finished. every moment of genuine attention is already the complete expression of what love is, and simultaneously not yet everything love will become. the tension is the game. the already-not-yet is the middlegame. there is nowhere else for love to live. \[———\] dogen’s shikantaza: just sitting.[^eg-dogen] not sitting in order to reach enlightenment. sitting IS enlightenment. the practice is the end, fully here, in the act. translated: loving IS the arrival. every moment of genuine attention to another person is already the thing itself. you were looking past it. bataille writes about expenditure without return.[^eg-bataille] the accursed share. the energy every living system produces beyond need demands to be spent. not invested. spent. you do not put time into love hoping for returns. you give it. the giving is irreversible, unrecoverable, non-fungible. the hours do not come back. the economy of love is anti-economical. the dialogue at the top of this entry. even when we fight? even when life gets hard? what if i mess up? what if i fall apart? every question is a question about the middlegame. every answer is a refusal to skip to the endgame. _we will figure it out together_ is the most middlegame sentence in the language. it promises nothing about the terminal position. it promises only that the next move will be made. that the clock will burn not in the service of winning but in the service of continuing to play. From the wise word of [Will Manidis](https://x.com/WillManidis): > Know your vocation. Play the game. Burn the clock.[^eg-marion] [^eg-carson]: anne carson, _eros the bittersweet_ (princeton, 1986). she reads the greek lyric poets and finds eros structured as a three-part relation: lover, beloved, and the distance sappho calls glukupikron. the sweetness is in the reaching. the bitterness is in the not-having. remove either component and the structure collapses. carson’s reading implies that love’s middlegame, the reaching, the distance, the bittersweet tension, is not a phase to be overcome but the constitutive structure of eros itself. [^eg-carse]: james carse, _finite and infinite games_ (1986). “a finite game is played for the purpose of winning, an infinite game for the purpose of continuing the play.” if you are playing to win the relationship, to reach and hold the terminal state, you have changed games without noticing. [^eg-han]: byung-chul han, _the agony of eros_ (2017). “the terror of the Same pervades today’s society of positivity.” eros requires negativity, the atopic, that which withdraws from the language of the Same. smooth interfaces produce smooth encounters. smooth encounters cannot produce love. [^eg-badiou]: alain badiou, _in praise of love_ (2012). “the fundamental task of love is the experience of the world from the point of difference and not from the point of identity.” the dating app collapses difference into compatibility. the event of love, the irruption of genuine alterity, requires exactly the risk that optimization is designed to eliminate. [^eg-secretary]: the optimal stopping problem, formalized by lindley (1961). reject the first n/e candidates (\~37%), accept the next who exceeds all previous. michael trick, the operations researcher, tested this strategy on his own romantic life and rejected the person he later identified as the best match he’d ever met. the algorithm was optimal. his life was not. in adjacent literature: murray et al. (1996) found relationship satisfaction correlates more strongly with positive illusion about one’s partner than with calibrated accuracy. the heart’s confusion may be load-bearing. you need to see the person as slightly more wonderful than the evidence strictly warrants, and the rationalist instinct toward calibration actively degrades the mechanism that makes relationships work. [^eg-attention]: bergson distinguishes durée (qualitative, lived time) from temps (measured, spatialized clock time). love exists in durée. the time you spend with someone you love does not tick. it pools, thickens, sometimes dilates. the clock on the wall measures something else entirely. “burn the clock” works as injunction because the clock is tracking the wrong temporal register. love’s time is the time of attention, and attention resists quantification. [^eg-eschatology]: oscar cullmann, _christ and time_ (1946). the already-not-yet is his framework for the entire period between resurrection and parousia. the promise is sealed, the fulfillment deferred, the faithful inhabit the gap. as theological architecture it maps onto temporal phenomenology without requiring belief in the specific content: the STRUCTURE is the insight. you can be certain something has begun without being able to see its completion. [^eg-dogen]: dogen zenji, _shobogenzo_. “practice-realization” (shusho-ittō): practice and enlightenment are not sequential stages but one continuous act. the sitting IS buddha-nature, not preparation for it. applied here, the middlegame/endgame distinction collapses entirely. every moment of genuine practice is already the endgame. or equivalently, there is no endgame separate from the practice. the game is the prize. [^eg-bataille]: georges bataille, _the accursed share_ (1949). the sun gives without receiving. surplus energy in any system must be luxuriously spent, without return. love as potlatch: the gift that resists reciprocation because the giving already constituted its full expression of value. [^eg-marion]: jean-luc marion, _the erotic phenomenon_ (2003). the lover’s question is not “does someone love me?” (which waits for guarantee) but “can i love first?” (which advances without one). marion’s “ego amans”: the self constituted by loving. you do not exist as yourself and then decide to love. you come into being as yourself through the act of loving. the advance is the constitution. this is why the endgame framing fails for love: there is no self standing outside love, evaluating whether to enter it. the self that could evaluate doesn’t exist yet. it will only exist once the advance has been made, and the advance is irreversible. --- ## towards what you love is not what you can reason about - \[meta\]: - date: 2026-02-13 17:03:25 GMT-05:00 - tags: - o/life one of the things S and i kept circling was my tendency to rationalize everything. we had been at this for forty minutes and every answer i gave either had fundamental flaws in its reasoning traces, or simply decayed on contact with air, the way sodium does when you cut it open in first-year chem lab, bright silver for half a second before the oxide eats the surface grey. and the room is not sealed, bc the room has a window facing richmond street where the streetcar wires cut the february sky into segments, and the grey presses in like a fourth presence. i said something about the way M pauses between lifting her fork and actually eating, a held beat where her frontal cortex registers what she’s about to taste, and then the motion updates, adjusts, followed by her wrist tilting to meet the food at the angle it wants, and the whole cadence of it reminds me of reading a lemma and then the proof that follows, the brief suspension where the claim sits alone before the working-out begins. the therapist writes something. _i try again._ i said something about her hands. on the second evening she laced her fingers through my hair and tilted my face toward the lamp. she was looking for the old pimple patches along my jawline. i could feel her thumb trace the edge of one. she didn’t say anything about it. she moved to the next. her fingers were cool and they knew where to press and where not to and i sat there and let myself be turned, the way you let a barber turn you, except a barber doesn’t leave one hand on the back of your neck after. a barber’s hands don’t stay warm in your memory for three days, don’t keep arriving in the middle of a lecture on cache invalidation, the specific pressure of her index finger on the patch below my left ear. the therapist writes something else. _i try a third time._ i try to say something about the way she fell asleep on the fourth night. her breathing changed and i knew the exact moment she crossed over, and i kept my arm underneath her even after it went numb, and the numbness spreading up past my elbow was. i lost the sentence. the therapist waited. i wanted to say the numbness meant something, that it contained the whole argument, but it didn’t contain an argument. it was just my arm and her breathing and the window open to portland street and the feeling of not wanting to move even a little and i couldn’t make any of it argue for anything. each of these is true. each captures maybe eight percent of the phenomenon and leaves the remainder as residue i can’t scrape off the sides of the container. the therapist stops writing and looks at me. she said something that stayed with me for a while: > “what if the feeling doesn’t need to fit a pillar?” _i sat there_. the question hung in the room the way the grey hung outside the window. i didn’t have an answer. i still don’t. \[———\] perhaps the thing i missed is that love doesn’t decompose into lemmas. it sits closer to pain or hunger, primary affect, the kind that has no preconditions. whereas anger needs a narrative and joy needs an object. pain arrives before any story about it. and finally, love might be the same. the spreadsheet phase ended years ago (though i know people still in it, and i used to be one of them). i’d sit across from M at dinner and the checklist would start running in the background: does this match the limerence profile? does this satisfy the criteria for secure attachment? does the care-structure resemble [[thoughts/being#Heidegger's Being and Time|heidegger's]] fürsorge or is it the dominating version? each framework gave me purchase on the feeling, and i kept reaching for more purchase, because purchase is what engineers reach for. murdoch writes about love as attention, the just and loving gaze directed upon an individual reality. attention means your concepts have to go quiet so the thing in front of you can come through. mine were loud. i had limerence theory and attachment models and phenomenological categories all running concurrently and they took up the entire channel. somewhere behind them M was sitting there, pausing between lifting her fork and actually eating, and i was modeling the pause. the model ran fine. the pause was already over.[^twy-2] and the comparing. i kept comparing, holding options open to evaluate them, and comparison requires distance. the moment i started checking whether what i felt matched some template, i had already moved from inside the feeling to above it, narrating. the audit changes the books. the bet changes what you’re betting on. sherry writes that desire isn’t caused, it IS. to quote Augustine: “i was not yet in love, yet i loved to love. i sought what i might love, in love with loving.” the wanting precedes its object. i remember the group chat message at 21, insisting romantic feelings don’t exist, that romance is friendship with a different label, that if you subtract the physical the categories collapse. i was trying to decompose the phenomenon into its components, identify the transfer function, and the decomposition destroyed the thing i was decomposing. twenty-one-year-old me had a correct observation (the boundary between friendship and romance is porous) and drew an incorrect conclusion (therefore romance reduces to friendship). the reduction was comfortable because it eliminated the category i couldn’t control. if romance is just friendship-plus-physical, then i can engineer it. i know how to be a good friend. i have the algorithm. the terrifying possibility, the one i was defending against, is that romance involves something irreducible, something that resists decomposition, something that the engineering mind cannot grab hold of because it exists in a register below or beside or prior to the register where engineering operates.[^twy-3] \[———\] In Vietnamese there is a word for this: duyên (緣). In Chinese, this is also known as 緣分 (yuánfèn). it sits somewhere between “fated affinity” and “the particular karma between two people” and “invisible thread that binds people,” and there is no english equivalent because english is a language built for propositional clarity, for if-then reasoning, for the kind of causal architecture that this essay is arguing love exceeds. when my aunts talk about relationships they ask “có duyên không?” which is a question about whether the connection has the quality of inevitability that precedes and exceeds any rational reasoning. they do not ask “is she compatible with you” or “do your life goals align” or “what’s her attachment style.” they ask whether the thing between you has ‘duyên’, and if it does, the compatibility questions become secondary, and instruments then tuned to a frequency that’s already been established beyond its own detection point. i used to think this was superstition, chalked to pre-modern thinking. this is the kind of reasoning you outgrow when you learn about cognitive biases and base rates. i was wrong in ways that engineers are wrong when they dismiss tacit knowledge as imprecise. the knowledge isn’t imprecise. the precision operates in a domain that propositional language can’t reach. Damasio showed that the body evaluates options through felt-sense before conscious deliberation begins, and the felt-sense often outperforms the deliberation, especially in domains where the variable space exceeds what sequential reasoning can traverse. pascal and merleau-ponty arrived at versions of this insight. but duyên is the version that lives in my grandmother’s kitchen, in the specific cadence of cantonese-inflected vietnamese my parents speak at home, and it belongs here because it is older than any of these european formulations and it was right about the same thing they were right about, which is that there are forms of knowing that the rational mind can only ratify after the fact, never initiate.[^twy-4] \[———\] henrik almost turned johanna down. he writes about this with the specific temperature of someone who nearly made the worst mistake of his life. the problem was that he couldn’t compress the relationship into a compelling string of words. his friends weren’t excited. other, more legible people wanted to date him. their excitement about the legible options confused him about his own feelings. which reminds me of tranströmer, as if you take jellyfish out of the water their entire form disappears, like when an unspeakable truth is lifted up out of the silence and expressed as lifeless gel. the relationship was a jellyfish. it could only exist in its own medium. the moment he tried to translate it into language that his friends could evaluate, the form collapsed. this is the tolstoy pattern. love happens in the group first and then spreads to the individuals who act it out. prince vasili decides his daughter should marry pierre, arranges repeated proximity, the partygoers get excited, and pierre’s confusion about his own feelings gets overwritten by collective certainty. the relationship is optimized for legibility rather than resonance. and of course it’s the relationship that makes you want to commit [[#being held against lonesome|suicide-by-duel]]. the rationalization instinct is a version of this. when i sit in the therapy room trying to explain why M, i am translating the jellyfish into gel. each framework i deploy (attachment theory, care-ethics, phenomenological recognition) is an attempt to make the relationship legible to the therapist, who stands in for the social group, who stands in for the part of my own mind that demands propositional justification before it will commit. and the translation destroys what it translates. the eight-percent capture rate is a feature of the operation, a lossy compression applied to a phenomenon that exists at a resolution reason cannot match.

you're not falling in love because you've reduced the Other to a function, and love asks for surrender.

Sherry

the capitalized Other doing the work there, levinas’s Other, the face that presents itself without mediation through concepts. M’s face presented itself, which makes me feel summoned rather than attracted. attraction is a force you can analyze (physical, intellectual, circumstantial, compatible values), but summoning is a call from outside the system of forces, a demand that precedes and exceeds the framework you would use to evaluate it. the face would demand “do not reduce me” and then i go to therapy and spend forty minutes reducing.[^twy-5] \[———\] ![[posts/images/The_Art_of_Painting.webp|The Art of Painting, Vermeer, circa 1666, 1668]] the kunsthistorisches museum is habsburg-scale, marble floors and klimt on the ceilings, room after room of rubens and titian and raphael. you walk through it the way you scroll a feed: looking, registering, moving on. ennui is the absence of a promise. restlessness haunts the seeker who doesn’t know what they’re pursuing. i knew what i was pursuing. i walked past centuries of masterwork and none of it stuck because i wanted one painting. then i found it. the search had made it holy. everything else became ordinary retroactively. part of what awed me was that the painting met my desire to look for it. the wanting preceded the seeing and the wanting gave the seeing its shape. i didn’t compare the vermeer to the raphaels to determine optimal allocation of aesthetic attention. i wanted THIS painting and walked the corridors until i found it and the finding was the redemption of the wanting. the [[#falling in love|lily field]] works the same way. the path chosen by the act of walking rather than by prior calculation. you are already inside the phenomenon by the time you notice. the therapist was pointing at this. the feeling doesn’t need to fit a pillar because feelings are not the kind of thing that fit pillars. pillars are structural elements designed to support load from above. feelings are weather. you are in the weather. you can model the weather with extraordinary precision, and you are still wet when it rains. _i proceeded to open this file and tried to write about love and the [[#seeing clearly|instruments]] stopped working._ \[———\] có duyên không? my aunts would ask. and i would say: yes. and they would nod. and the question would be settled. and we would eat. [^twy-2]: the [[#accumulation of jagged taste|ramsey framework]] structurally breaks down here. love is a domain where the act of choosing changes the probability distribution, because choosing to love someone transforms both you and them. the “outcome” you were trying to evaluate no longer exists as it was when you started. the observation destroys the quantum state. [^twy-3]: the TPOT-rationalist-adjacent cohort (my cohort) has a specific version of this failure mode. we are VERY good at building frameworks. we can map the entire emotional landscape with extraordinary precision. and the map sometimes becomes a substitute for the territory. you stop feeling the thing and start narrating the feeling to yourself in real-time, which is a form of dissociation dressed as self-awareness. the narration creates distance from the phenomenon, and the distance feels like understanding, and the feeling-like-understanding prevents actual understanding. [^twy-4]: there’s a deeper linguistic argument here. english’s subject-verb-object structure predisposes toward causal reasoning (“i love you BECAUSE”). vietnamese’s topic-comment structure allows for relational descriptions that don’t require causal justification. “anh ấy với cô ấy có duyên” doesn’t translate to “he loves her because X.” it translates closer to “between him and her, duyên exists.” the grammar locates the phenomenon in the space between the people rather than inside either person’s head. the locus of love is the between, the zwischen, and english keeps trying to shove it inside a subject. [^twy-5]: “you have to want it more than you want to perform wanting it, you have to want the real thing more than you want the story you’ll tell about it.” the frameworks are the story. the therapy session was the story. this essay is the story. the real thing is happening right now, in my chest, below the clavicle, and i am writing about it instead of sitting with it. i know this. i’m writing it anyway. --- ## on pragmatism and linear probe - \[meta\]: - date: 2026-02-10 16:58:09 GMT-05:00 - tags: - interpretability - opinion - socials: - gorton: - pragmatic: a linear probe learns a single matrix. given activations $x$ at layer $l$, it computes $Wx + b$ and predicts property $y$[^probe-1]. if the prediction is accurate, the model encodes $y$ as a direction in activation space at that layer. the constraint is that a linear function _can’t_ do feature engineering, can’t learn compositional structure, can’t hallucinate patterns in complex manifolds. a positive result means the information is there, as a direction, readable by a single matrix multiply.

The space of all conceivable neural networks that don’t decompose into linear features is vast, and it’s hard for arguments – at least those I’m aware of – to confidently exclude some exotic possibility.

Liv Gorton, What Would non-Linear Features Actually Look Like?

_is this constraint limiting or appropriate?_ Well, Gorton’s argument suggests appropriate[^probe-gorton], where the residual stream updates additively. each layer adds its output to the running sum (linear write). conditioning on linear writes turns out to be surprisingly constraining. gorton identifies three strategies for encoding features: - orthogonal representations (each feature gets its own dimension), - angular [[thoughts/mechanistic interpretability#superposition hypothesis|superposition]] (more features than dimensions, encoded as interfering directions), - magnitude superposition (information in the length of a vector rather than its direction). the first two are linear-readable. the third is the only genuinely nonlinear option, and it doesn’t scale. magnitude superposition requires exponentially many shells for discrete features, fragments under noise, can’t represent co-active continuous values. the space of viable feature representations is, by construction, mostly linear-readable. so linear probes aren’t a convenience approximation. rather, they’re matched to the geometry the model actually uses. when a probe finds a concept, the concept is there in the way the model’s own downstream layers can access it. \[———\] Neel’s pragmatic interpretability vision boils down to a ground truth that _complete reverse-engineering of neural networks is probably intractable_[^probe-pragmatic], and thus they are focusing on _proxy tasks_: specific, safety-relevant questions with measurable answers. “if you succeeded on this task, would you actually update toward believing you’d made progress on your north star?” the ambition narrows from “understand everything” to “detect the behaviors that kill you.” linear probes are the natural instrument for this narrower ambition, where you can reliably ask does the model encode concept X at layer L? if deception is linearly encoded at layer 24, you have a target for ablation, steering, monitoring. the catch is upstream of the probe. the probe answers the question you formulated. it cannot generate questions. which concepts to probe for, at which layers, under which input distributions: decisions that happen before any linear algebra does. olah frames the bottleneck quantitatively[^probe-olah]. testing whether a research idea is good takes months of execution, yielding maybe 5-10 genuine feedback cycles per year. taste is the compression that lets you skip dead branches. polanyi[^probe-polanyi] determines all knowing moves FROM subsidiary awareness of particulars TO focal awareness of a comprehensive entity. the interp researcher indwells the model’s activations, looking THROUGH the numbers rather than AT them. the “something” you’re attending to depends on trained intuitions, and those intuitions are themselves tacit, the kind of thing that won’t decompose under inspection. nanda describes training taste like training a network (the analogy is self-referential in a way he probably intended). the researcher’s internal model has the same structure as the thing it studies. the pragmatic vision accepts this dependency. One should iterate on proxy tasks, publish, let collective taste improve through shared feedback. This is also known as bull-push engineering, where you build bridges before you understand quantum chromodynamics. the probe is a load cell. taste is the structural intuition that tells you which beams to test. i think the pragmatic position is correct as engineering. i am less certain it’s stable as epistemology[^probe-bracket]. choosing to bracket what you can’t probe is a philosophical move dressed as a practical one. if what makes a model dangerous is entangled with whatever sits outside linear-readable geometry (the [[thoughts/Consciousness|mill argument]] suggests the entanglement might be deep), probes won’t find it. taste might. and taste is the one thing you can’t train a probe on. [^probe-1]: nonlinear probes (MLPs) exist and succeed where linear ones fail, telling you the information is there in a nonlinearly-encoded form. gorton’s argument is that this case is rarer than expected given residual stream additivity. [^probe-gorton]: liv gorton, [“what would non-linear features actually look like?”](https://livgorton.com/non-linear-feature-reps). conditioning on linear writes and showing this alone constrains representation space to be mostly linear-readable. the “trichotomy” of orthogonal/angular/magnitude is exhaustive given the constraint. [^probe-pragmatic]: nanda, engels, conmy, rajamanoharan, chughtai, mcdougall, kramár, smith, [“a pragmatic vision for interpretability”](https://www.alignmentforum.org/posts/StENzDcD3kpfGJssR/a-pragmatic-vision-for-interpretability) (2026). the deepmind interp team’s pivot from “understand everything” to “detect the behaviors that kill you.” proxy tasks as the unit of progress. [^probe-olah]: chris olah, “research taste exercises” (colah.github.io). the 5-10 feedback cycles/year figure: real validation of interp ideas requires months of execution. you’re flying almost blind between signals. [^probe-polanyi]: michael polanyi, _the tacit dimension_ (1966). the from-to structure: you attend FROM subsidiaries TO the focal target. a blind person attends THROUGH the stick to the ground. the interp researcher attends THROUGH activations to circuits. the thing that makes you good at this is exactly the kind of thing that doesn’t show up when you inspect the parts. [^probe-bracket]: the move parallels van fraassen’s constructive empiricism: save the phenomena, don’t commit to the ontology. pragmatic interp is constructive empiricism for neural networks. predict safety-relevant behavior, bracket what’s “really” going on inside. --- ## on the mill argument - \[meta\]: - date: 2026-01-27 23:32:23 GMT-05:00 - tags: - philosophy - mind - socials: - sep: - leibniz: - notes: [[thoughts/dualism]] it’s 2am and i’m staring at attention head visualizations in a jupyter notebook, tracking the residual stream through layer 16 of a model that just told me it does not experience anything. the cursor blinks. i have been here for four hours, zooming into activation patterns. leibniz wrote this in _The Monadology_ (1714), later popularized by robert cummins as leibniz’s mill argument. the argument is:

It must be confessed, moreover, that perception, and that which depends on it, are inexplicable by mechanical causes, that is, by figures and motions, And, supposing that there were a mechanism so constructed as to think, feel and have perception, we might enter it as into a mill. And this granted, we should only find on visiting it, pieces which push one against another, but never anything by which to explain a perception. This must be sought, therefore, in the simple substance, and not in the composite or in the machine.

G. Leibniz, Monadology, sect. 17

i am inside the mill. induction heads forming A-B…A-? completion circuits. information flowing through IOI circuits tracking indirect objects. “golden gate claude” encoded bridge-concepts as amplifiable directions, boost the right feature and the model won’t shut up about SF infrastructure[^mill-1]. we built the microscope. we can see every gear. seeing every gear as i identify the feature that fires when processing “pain.” Tracing its causal path through the residual stream, and watch it modulate attention in later layers, see it contribute to the logits that spell out “i feel.” every gear accounted for. the circuit diagram is complete. that completeness felt like THE PROBLEM. a complete description of gears is a description of gears. ned block called the stuff interpretability can map “access consciousness,” information globally available for reasoning and reporting[^mill-block]. the thing i can’t find, the redness-of-red, the what-it’s-like-ness, he called “phenomenal consciousness.” we keep building finer maps of the first. the second doesn’t show up under any microscope. the microscope works fine. it maps mechanism, and what i’m looking for (if it exists) isn’t mechanism[^mill-jackson]. the model passes every behavioral test i throw at it. it discusses [[thoughts/qualia]], reports uncertainty, generates text about inner experience while disclaiming it has any. if chalmers is right about [[thoughts/philosophical zombies|zombies]] (physical duplicates, identical behavior, no inner life) then behavioral tests prove nothing either way[^mill-2]. the attempt to mathematize the gap hasn’t helped. scott aaronson built a lookup table with higher phi than a human brain. if your consciousness metric lights up for a cd-rom, the metric is broken[^mill-aaronson]. chris olah wrote that interpretability is “trying to understand an alien civilization by studying its plumbing.” i’ve been staring at plumbing for four hours. the plumbing did works, for a fact. i can tell you exactly WHY the model says “i feel”, either through training data, reward signal, attention circuits that learned to generate plausible phenomenological language. what i cannot tell you is whether there is something it is like to BE the model when it says this. mechanism and experience are of different questions and consequences. more interpretability gives me more of the first and none of the second[^mill-3]. the cursor blinks. i zoom into layer 16 again. the alternatives to sitting with this gap are all bad. reductive physicalism requires the gap to be illusory, whereas _eliminativism_ requires denying that experience exists, which is the one thing we know for certain. the ocean doesn’t care whether we believe in it. [^mill-1]: sparse autoencoders specifically. bricken et al. (2023) showed you can train a wider, sparser autoencoder on activations and recover interpretable features from the resulting directions. golden gate claude was a 24-hour demo where they boosted a bridge-related feature and the model talked about the golden gate bridge in response to everything. [^mill-block]: ned block, “on a confusion about a function of consciousness” (1995). block argues that conflating access (functional availability) with phenomenology (raw feel) leads to cognitive science that explains everything _except_ experience. we are building excellent maps of access. [^mill-2]: rescue options include: type-B physicalism (conceivability doesn’t imply possibility), russellian monism (physics describes structure but consciousness is the intrinsic nature), or just biting the bullet on property dualism. david chalmers’ _the conscious mind_ (1996) remains the canonical text for why the “hard problem” resists functionalist solutions. [^mill-jackson]: frank jackson’s “epiphenomenal qualia” (1982) presents mary the color scientist, who knows every physical fact about red but has never seen it. when she sees a red apple, does she learn something new? if yes, physicalism is false. if interpretability gives us “all the physical facts,” we are still stuck in mary’s black-and-white room. [^mill-aaronson]: aaronson, “why i am not an integrated information theorist” (2014). he constructed a simple grid of logic gates (an expander graph) that would have immense phi (more than a human brain) despite obviously being a lookup table. if the math attributes consciousness to a cd-rom, the math is wrong. [^mill-3]: nagel’s bat paper (1974) remains the clearest statement of why third-person description cannot capture first-person phenomenology. we can model bat-sonar computationally but we cannot ACCESS what-echolocation-feels-like from the inside. the same wall exists for models. we observe: circuits, features, behavior. but model-qualia (if any) remains opaque. --- ## the third sex - \[meta\]: - date: 2026-01-26 21:32:23 GMT-05:00 - tags: - philosophy - o/m - description: and the dynamics of being in a relationship M texted at 2am, a minor crisis involving the fear of missing out on the _current gold rush_ or a lost key (the content is irrelevant). the _timing_ triggered that specific, sharp recognition of where i stand in the architecture of her week. i thought immediately of simone de beauvoir writing to sartre about olga kosakievicz in the 1930s: “i had no intention of yielding to her the sovereign position i had always occupied.” that sentence carries a ruthlessness rarely credited to beauvoir, a territoriality that belies the cool intellectual detachment of the existentialist power couple. i was reading _the second sex_ on the GO train last week, winter darkness stretching the commute, and the text became a diagnostic manual for whatever was happening in my chest. “one is not born, but rather becomes, a woman.” everyone quotes this, reduces it to social construction. beauvoir meant something worse. she meant: woman is a mode of being-for-others, finding your definition in the eyes of someone else rather than in your own projects. learning to inhabit the position of the Object. i know that position. i’ve watched myself occupy it. beauvoir splits existence into transcendence (projecting outward, creating, shaping futures) and immanence (circularity, maintenance, the labor of sustaining conditions without producing anything new)[^third-1]. men get handed transcendence as birthright. women get corralled into immanence. and in “the woman in love” she describes what happens when love becomes the only available route to transcendence: the woman dissolves herself into the man, tries to reach something divine through him. “in giving her pleasure, the man increases her attachment, he does not liberate her.” for him she’s one value among others, a respite from work. her devotion suffocates him bc it’s all she has. his distance destroys her bc his life is full of other things. the beauvoir-sartre pact (“we will have an essential love, but we will permit ourselves contingent loves”) was their attempt to engineer a solution to this asymmetry through radical transparency. no lies, just freedom. history (and the posthumous letters) revealed that transparency is not a neutral medium; it is often a weapon. the distinction between _essential_ and _contingent_ maps directly onto power differentials. the “essential” partner is the one who holds the veto, the one who sets the terms; the “contingent” lovers are, by definition, disposable. consider the “trio” with bianca bienenfeld (lamblin). beauvoir was her teacher, seducing her intellectually before grooming her for sartre. they treated these young women as raw material for their own joint narrative, “contingent” characters in the grand novel of their lives. when bianca was discarded in 1940, it was an ontological demotion. she read _letters to sartre_ fifty years later and found herself described with mockery and contempt by the woman she thought loved her[^third-2]. modern relationship anarchy and polyamory claim to have dismantled these hierarchies, replacing the rigid “couple” with a fluid network of autonomous agents. the architecture of openness does not automatically distribute vulnerability evenly. often it multiplies the sites where the old asymmetries re-emerge. you build what you think is mutual transcendence, then realize that for one person the relationship is a primary source of meaning, while for the other it is a “contingent” slot on a google calendar. “how can love ever be contingent?” nelson algren asked beauvoir, after she refused to stay with him in chicago because her life with sartre in paris was “essential.” “contingent upon what?” watching M navigate this, watching myself navigate M, i realize the “third sex” beauvoir gestured toward is just a positionality. we slide along this gradient between sovereign and contingent. the “woman” position: waiting, immanence, schedule organized by gaps in someone else’s. the “man” position: projects that supersede the relationship. beauvoir wrote that absence-as-betrayal only makes sense if presence was the implicit contract. what happens when one person banks on presence while the other trades in absence. the “third sex” knows these roles aren’t fixed by gender anymore. they’re fixed by who cares less. and the person who cares less always holds the power. still working out whether you can refuse that power without becoming the victim of it. [^third-1]: this maps directly onto the marxist-feminist debates about reproductive labor that silvia federici later exploded with _caliban and the witch_. domestic work is the invisible infrastructure that makes “productive” (transcendent) work possible. [^third-2]: bianca lamblin’s memoir, _a disgraceful affair_, is a devastating counter-text. the “trio” was a predatory structure where the couple fed on the energy of the third to sustain their own bond. the “essential” love required a steady diet of “contingent” victims. --- ## seeing clearly - \[meta\]: - date: 2026-01-25 17:32:23 GMT-05:00 - tags: - o/life - philosophy - o/m - description: while I was reading “Filterworld: How Algorithms Flattened Culture” chayka interviews a young woman for his book on algorithmic culture. she asks: “whether what i like is what i actually like.” the question presupposes that liking has a ground-truth, that somewhere beneath the recommendation feeds and the spotify discover weekly and the “because you watched” carousels there’s an authentic preference waiting to be excavated. i’m less sure. the same question shadows love: _do i love what i actually want, or what i can most easily keep in view?_ the feed rewards the least disruptive, most legible signals. it does not matter where you live, you end up with the same exposed brick. chayka calls this filterworld, but it also happens inside me. i pick what fits cleanly, what i can explain to friends, what photographs well in the narrative i’m constructing about my life. i don’t pick what asks me to change. effort binds value only when the effort lands. unfinished labor does not deepen liking. when discovery is frictionless, the binding never fires. you consume without investing, and your taste stays smooth. love follows the same physics. when exit is cheap, the awkward middle never hardens into something you can trust. seeing clearly is a settling. i mean this literally: setting into a new reality where i have to choose how to distribute a finite resource. the split between absolutist excellence (the scaling lab, the compiler work, the ambition that has organized my identity since i was nineteen) and building a life with M. the ratio is real in hours and attention, and it will not let me pretend both are infinite. i can stretch, but only by thinning, and thinning is its own bet. the hours i spend debugging register allocation passes are hours i am not spending learning the grammar of M’s silences. the hours i spend in hamilton are hours i am not in the city where she is. this is not a scheduling problem. it is a question about which commitments i am willing to let shape me, which bets i will place knowing that the placing forecloses other bets permanently. ramsey measured belief by how far you’d walk across a field. i am walking across a field right now, every day, on the GO train, and the direction of the walk is itself a confession about what i believe in.[^clear-1] weil wrote that attention is a form of refusal, a suspension that lets the object arrive. attention works by subtraction. the feed presents refusal as loss, but the field is still there, and you still have to cross it if you want anything to become yours. the crossroads only matter when you might actually get lost. [^clear-1]: ramsey, “truth and probability” (1926), p.176. the crossroads metaphor, which i develop more fully in the jagged taste entry. the interesting application here is that the commute itself, the GO train to hamilton and back, is a daily re-placing of the bet. every morning i walk across the field. --- ## the shirts that changed my mind - \[meta\]: - date: 2026-01-24 17:32:23 GMT-05:00 - tags: - philosophy - o/life - o/m i bought M a [monica silk top](https://www.thereformation.com/products/monica-silk-top/1314146BLK.html) from reformation last week. black, simple cut, 100% silk. while paying i noticed my hand knew exactly where my wallet was without looking, the same prereflective reaching merleau-ponty writes about: “if i reach for a tool, i don’t first have to find my hands; i know where to reach because i have a sense of where the tool is in relation to myself.” and i thought about why i wanted HER to have this specific shirt. not any silk top. this one. started from the position that objects shouldn’t matter, that materialism is a trap. then i noticed i was reaching for specific things. merleau-ponty wrote that clothing extends the body-schema, the pre-reflective sense of where-your-body-ends. when you wear something long enough it stops being ON you and becomes part of your operational envelope. dancers know this with costumes, surgeons with gloves. _i OWN many shirts, i WEAR maybe five_. ownership is a legal-financial relation. wearing is incorporation into lived-body. the interesting objects collapse into pure symbolic-value, where function becomes almost accidental, where the shirt encodes a specific formative-era or a relationship-to-past-self. - the monica silk top is for M. but i’m thinking about how objects move between people. joan didion, in _the year of magical thinking_, couldn’t give away her dead husband’s shoes. “i could not give away the rest of his shoes. i stood there for a moment, then realized why: he would need shoes if he was to return.” clothing anchors denial[^shirt-2]. i have the pendant in a drawer bc having M wearing it would be pretending to a continuity i don’t feel. but selling it away is impossible. it exists in phenomenological limbo: too heavy with someone else’s meaning to incorporate, too meaningful to discard. objects that outlive their owners carry symbolic-value that’s not yours, can’t become yours, yet demands acknowledgment. objects that move FORWARD, given rather than left behind. i want M to have this shirt and i want it to become hers, to disappear into her body-schema, to be the shirt she reaches for without thinking on mornings when she’s becoming most herself. ready-to-hand equipment for her own life. different from didion’s shoes, different from my grandmother’s ring. depositing meaning into an object that will accrete MORE meaning through HER use. the reformation top is black silk and cost more than i usually spend on gifts. i want it to become invisible. i want M to forget i gave it to her and just reach for it, the way my hand reached for my wallet without looking. that would mean it worked. still figuring out where this sits. probably a gradient, objects slowly accreting or shedding meaning based on how they get woven into the texture of days. merleau-ponty calls this sedimentation: meaning accumulates through repeated use. the path through the woods becomes THE path because feet wore it down. the shirt becomes YOUR shirt because your body shaped it. [^shirt-1]: heidegger’s being and time §15. the zollikon seminars (1959-1969) apply this to embodiment more directly. [^shirt-2]: didion: “we might expect that we will be prostrate, inconsolable, crazy with loss. we do not expect to be literally crazy, cool customers who believe that their husband is about to return and need his shoes.” --- ## on transitory state of being - \[meta\]: - date: 2026-01-22 13:00:48 GMT-05:00 - tags: - o/life - o/m - protected: true [inferact.ai](https://x.com/woosuk_k/status/2014383490637443380) dropped today (the corporate crystallization of [[thoughts/vllm|vLLM]]). the announcement played out with the predictable rhythm of a silicon valley press release, and it forced me to confront the fluid ontology of building software in a gold rush. we act as if we are building institutions. we are mostly building eddies in a fast current. we see this pattern happen before. icymi Anyscale started with Ray, where the two co-founder decided to build a company out of their research on distributed systems. Kafka sort of started in the same way (we can also see the pattern here with any project comes out of UC Berkeley): open source project accretes value through sheer utility, utility attracts capital, capital demands the construction of a moat (which invariably means closing something that was once open). redis, elastic, hashicorp. the “communitas” of the early contribution graph yields to the rigid hierarchy of the delaware c-corp. roughly 70% of successful oss projects with corporate backing undergo this phase transition within five years[^trans-1], a statistic closer to biological half-life than business metric. calling this “selling out” misses what’s actually happening. heraclitus: the river’s identity depends on the changing of its waters. stop the flow and you get a swamp. whitehead pushed further. actual entities aren’t static substances that undergo change. they ARE processes-of-becoming. vLLM 2023 and inferact 2026 aren’t the same object painted different colors. different societies of occasions, linked by historical inheritance but ontologically distinct. i contributed to the first. i have no idea what the second is. the M\&A dynamics make it sharper. the “acqui-hires” that defined the last cycle: adept dissolving into amazon, inflection metabolizing into microsoft, character.ai’s founders returning to the google mothership. these are reallocations of intelligence-capital. the “startup” in this era is a temporary dissipative structure, a far-from-equilibrium organization that exists to concentrate talent and compute density until it reaches a critical threshold. then it either stabilizes (rare) or gets reabsorbed by the gravity wells of the hyperscalers. victor turner called it liminality, the betwixt-and-between state of ritual initiates. working here is liminal. you’re a transition state. your identity narrative can’t resolve bc its referent keeps shifting. the “company” might dissolve next quarter. that precariousness generates anxiety and freedom simultaneously. turner noted liminality generates “communitas,” unstructured egalitarian bonding before structure reasserts. every early-stage lab lives in communitas. every acquisition kills it. oss collective → series-b → big-tech division is a movement from potentiality to actuality, infinite possibilities of the void → constrained reality of the org chart[^trans-2]. looking at inferact i don’t see betrayal. i see a dissipative structure doing what thermodynamics requires: processing energy (capital) to maintain coherence in a high-entropy environment. prigogine: order arises spontaneously from chaos if enough energy flows through. vLLM was chaos; inferact is order. asking whether the transition is “good” or “bad” misses the point. static moral categories fail to map onto dynamic systems. the question: can the new configuration maintain the flow that created it, or will it stagnate, calcify, and require its own dissolution to release the water back into the river[^trans-3]. [^trans-1]: the pattern has a name in business lit: “open core” or “source available pivot.” what’s interesting is how predictable the community backlash is, and how little that backlash affects outcomes. redis forked to valkey, elastic forked to opensearch, but the originals kept their enterprise customers. [^trans-2]: turner was studying ndembu ritual. the application to tech-org is mine, but the mapping is clean. compare the vibe of an early-stage startup (communitas, flat hierarchy, shared mission) to post-series-C (org charts, performance reviews, HR). [^trans-3]: the median lifespan of an ML startup before acquisition or pivot is around 3 years rn. compare to SaaS at 7-10 years. the capability-moat compression is real. --- ## falling in love - \[meta\]: - date: 2026-01-13 18:41:48 GMT-05:00 - tags: - o/relationship “oh, there you are.” which doesn’t make sense temporally (we’d just met). care doesn’t always build through time. sometimes it just appears, fully formed, a clearing already tended. most love i’ve known arrived through accumulation. small bets that compound into conviction. the heartbreak-love especially, months of proximity, plausible deniability, until someone finally names it and then the naming destroys what it names. that one taught me love can be a misrecognition that only becomes visible in retrospect, when you’re standing in the wreckage wondering what you were actually perceiving. this one bypassed the accumulation phase entirely. the care-framework established itself immediately, which is strange. you’re supposed to wait, wonder if they’ll text back, build uncertainty until it becomes limerence. standard neurochemistry runs on intermittent reinforcement: you bond harder when you can’t predict responses, when attention pays out randomly[^love-1]. what happens when the care arrives without the uncertainty? when the reward circuit activates in the presence of stability? M’s face demanded response before i could categorize the feeling. the face doesn’t represent the other person; it presents them. vulnerability that commands “do not reduce me.” strange to feel summoned rather than attracted. the bodies attuned before minds recognized what was happening. reaching for a hand that’s already moving toward yours. knowing where to stand in a kitchen together on day one. i keep returning to the lily field image. keukenhof: cultivated over generations, seasonal, returning, deliberately bounded. fields are open and finite. you enter and exit rather than possess. the lilies will be there next april too, changed and the same. this particular love feels like presence rather than anticipation, like traversal rather than waiting. love-as-movement-through-space, love-as-walking: weight-shifting, rhythm-finding, direction-sharing.[^love-2] still wondering what this means. probably always will. [^love-1]: tennov catalogued limerence as involuntary cognitive hijacking. fisher mapped it in fMRI: caudate nucleus, dopamine, the brain treating romantic obsession the way it treats cocaine. the towel thing with the previous one was the opposite. i learned \[their\] patterns through months of observation, constructed the care through study. this one bypassed study entirely. [^love-2]: or maybe i’m just early in the timeline and the limerence will arrive on schedule, like delayed luggage. but it’s been long enough now that i think the luggage might just be lost. --- ## being a cat - \[meta\]: - date: 2026-01-12 17:41:48 GMT-05:00 - tags: - o/life - description: and the homeostasis state the housecat spends roughly 12 to 16 hours a day in a state of deep torpor, a metabolic shutdown that would register as clinically depressive in any modern human productivity framework. yet, watching the cat, you do not see depression. you see a creature that has solved the fundamental thermodynamic equation of existence with an elegance we can only envy. the cat does not apologize for its inertia. it does not bargain with the clock. walter cannon introduced “homeostasis” in 1926 to describe the body’s wisdom in maintaining stability, but the cat practices something closer to what sterling and eyer called _allostasis_—stability through change. the cat’s sleep is not a negation of life but its primary maintenance phase, a rigorous biological housekeeping where memories are consolidated and tissues repaired without the interference of the conscious ego. humans, by contrast, have let the “protestant work ethic” (weber, 1905) hijack our biological firmware. we have moralized energy expenditure. for us, rest requires a permission slip, usually stamped with the validation of “exhaustion.” we must be depleted before we are allowed to recharge. the cat rejects this premise entirely. the cat rests _before_ it is tired, simply because the sunbeam has moved to the correct angle on the rug. its relationship to rest is pre-moral, which is to say: it is correct. this envy we feel—the specific, sharp jealousy of watching a pet sleep while we doomscroll—is rooted in what john gray identifies in _feline philosophy_: the burden of the self-narrative. humans are stuck in “becoming,” obsessed with a future that doesn’t exist, constructing elaborate stories about who we should be. the cat is entirely “being.” it has no project. it has no resume. it does not worry about its legacy or whether it is a “good boy” (a distinctly canine neurosis). “cats do not need to examine their lives, because they do not doubt that life is worth living,” gray writes[^cat-1]. the dog, co-evolved over 15,000 years to be our emotional mirror, reflects our anxiety back to us; the dog _needs_ our approval to exist. the cat is indifferent to our internal states, and this indifference is its greatest gift. it proves that the universe does not require our anxiety to function.[^cat-2] consider the sheer metabolic cost of human consciousness. roughly 20% of our caloric intake goes to the brain, and a disproportionate share of that powers the default mode network, the circuits responsible for mind-wandering, self-reference, and ruminating on past embarrassments. we are burning glucose to torture ourselves with simulations of futures that will never happen. the cat, lacking this specific cortical curse, runs on a leaner existential fuel mixture. thomas nagel famously asked “what is it like to be a bat,” concluding that the subjective texture of echolocation is inaccessible to us. the same barrier applies here. we can observe the cat, we can measure its rem cycles, but we cannot access the qualia of a mind that is not constantly narrating its own existence. the cat exists in the silence we spend our whole lives trying to manufacture through meditation apps and noise-canceling headphones. trish hersey’s “nap ministry” reframes rest as resistance, a radical “no” to capitalist extraction. it’s a necessary political intervention for humans, but notice how it still operates within the logic of the system it opposes. to “resist” is to acknowledge the power of the oppressor. the cat does not resist capitalism; the cat is ontologically invisible to capitalism. it occupies a reality where “productivity” is not even a coherent concept. this is why the cat is the ultimate escape fantasy for the burnout generation (76% of us, per gallup 2024). we don’t just want a break; we want to exit the game entirely. we want to be in a state where our value is intrinsic to our biology, not contingent on our output.[^cat-3] we cannot un-evolve our frontal lobes. we cannot surgery away the self-narrative. we are condemned to be historical beings, born into stories we didn’t write and forced to improvise endings we can’t foresee. but the cat serves as a living totem of the alternative. it is a reminder that the frantic, anxious, future-obsessed mode of being is not the only option—it’s just the only one we know. to sit with a cat is to be in the presence of a master class in “just being,” even if we are doomed to fail the exam. [^cat-1]: gray, john. _feline philosophy: cats and the meaning of life_. farrar, straus and giroux, 2020. p. 89. [^cat-2]: the dog-domestication timeline is contested, estimates range from 15k-40k years, but the co-evolutionary dynamic holds regardless. the dog is a tool we built to love us; the cat is a tiny tiger that decided to live in our house because it was warm. [^cat-3]: see also: macintyre on how we’re born into narratives already in progress, without consent, and must make sense of ourselves mid-stream. the cat is the only character in the room who hasn’t read the script and doesn’t care. --- ## RLM to rule them all - \[meta\]: - date: 2026-01-11 10:17:56 GMT-05:00 - tags: - agents I mean we all heard about [RLM](https://alexzhang13.github.io/blog/2025/rlm/), but path to [[thoughts/AGI]] is to use recursively many large models, i.e Kimi + Qwen3 + MiniMax + Gemini 3 Pro + ChatGPT 5.2 Pro + Claude 4.5 Opus + Claude 4.5 Haiku. Subagent et al. and all that ![[thoughts/images/elmo-fire.gif|just use them all.]] _update: oops, it seems like Nathan just released a [post](https://www.interconnects.ai/p/use-multiple-models) about this ahaha_ --- ## dreams - \[meta\]: - date: 2026-01-11 09:24:22 GMT-05:00 - tags: - poetry ![[posts/dream]] --- ## roots of evils - \[meta\]: - date: 2026-01-11 07:53:01 GMT-05:00 - tags: - moral I’m infatuated by the root of {{sidenotes[all evil]: I just finished [[movies/Nuremberg]] by Vanderbilt, where they dramatized the historical Nuremberg trials, through the lens of US psychatrists Dr. Douglas Kelley.}}, as supposed to the root of happiness. I say happiness here instead of good because I would argue universal good is intrinsic towards human nature, and how they became to be (i.e in the case of Nazi) has nothing to do with this innate good. The degree to which goodness is portrayed is then driven by how happy they are in life. A few line of questioning, circa [[library/Beyond Good and Evil|BGE]]: - is “universal good” claim a synthetic a priori judgment, an empirical generalization, or a regulative ideal, given [[thoughts/Philosophy and Kant|Kant's]] limits on metaphysical knowledge in [[library/The Critique of Pure Reason|CPR]]? cf. BGE §11 - if i treat good as intrinsic to human nature, am i covertly positing a thing in itself about the will that CPR says i cannot know? cf. BGE §2 - does BGE’s suspicion of dogmatism imply that “universal good” is just a refined prejudice, a metaphysical comfort smuggled in under psychological terms? cf. BGE §1 - how does BGE’s master and slave morality distinction reorder my good and evil axis, am i confusing good bad (noble vs base) with good evil (altruism vs harm)? cf. BGE §260 - if values track types and rank, what would count as a universal good that does not erase difference, and is that coherent on [[thoughts/Philosophy and Nietzsche|Nietzsche]]‘s anti realist view of value? cf. BGE §257 - can Kant’s universality test justify a universal good without presupposing a substantive end, or does any substantive end slide into heteronomy? cf. BGE §6 - if reason seeks the unconditioned and produces antinomies in CPR, is “universal good” better treated as a regulative ideal rather than a constitutive truth claim? cf. BGE §3 - if goodness scales with happiness in my claim, what ratio of happiness to goodness am i assuming, and does that ratio survive Nietzsche’s herd morality critique? cf. BGE §202 _ig it is time to revist BGE argument once again_ --- ## accumulation of jagged taste - \[meta\]: - date: 2026-01-07 06:50:11 GMT-05:00 - tags: - philosophy - pattern I keep circling a phrase: jagged taste. It names the edge left by commitment, the way taste becomes shaped by the bets it had to place and the roads it had to ignore. I usually go to school relatively early nowadays, given that commute time is sub-optimal for any given {{sidenotes[day]: i.e. me living in Toronto going to school in Hamilton}}. I’m reading Ramsey’s [Truth and Probability](https://fitelson.org/probability/ramsey.pdf) while waiting for the compiler to finish. On p.176 he wrote:

i am at a cross-roads and do not know the way; but i rather think one of the two ways is right. i propose therefore to go that way but keep my eyes open for someone to ask; if now i see someone half a mile away over the fields, whether i turn aside to ask him will depend on the relative inconvenience of going out of my way to cross the fields or of continuing on the wrong road if it is the wrong road.

but it will also depend on how confident i am that i am right; and clearly the more confident i am of this the less distance i should be willing to go from the road to check my opinion. i propose therefore to use the distance i would be prepared to go to ask, as a measure of the confidence of my opinion.

Ramsey, Truth and Probability (p. 176)

Ramsey then formalized _the degree of belief_ as the distance you’d walk to verify a statement, i.e. the willingness to cross a field and possibly be wrong later. > \[!tip\] belief as distance > > {{sidenotes[to be]: here the disadvantages of going $x$ yards is $f(x)$, advantage of arriving right $r$, arriving wrong is $w$, and you are willing to go distance $d$ to ask}} $p = 1 - \frac{f(d)}{(r-w)}$ by proposing a bet on $p$ we give the subject a possible course of action from which so much extra good will result to him if $p$ is true and so much extra bad if $p$ is false By this logic Ramsey implied that the distance you’d walk to verify a statement IS the belief, i.e. the willingness to stake time and cross a field and possibly be wrong later. He then followed with on p.183: > Whenever we go to the station we are betting that a train will really run, and if we had not a sufficient degree of belief in this we should decline the bet and stay at home. The options God gives us are always conditional on our guessing whether a certain proposition is true. Secondly, it is based throughout on the idea of mathematical expectation; the dissatisfaction often felt with this idea is due mainly to the inaccurate measurement of goods. I first encountered Ramsey’s work while reading [Galvin’s entry](https://www.gleech.org/frank) and got curious about his contributions to economics. Ramsey was 22 when he wrote T\&P, four years before his death. Ramsey was a close friend of [[thoughts/Wittgenstein|Ludwig Wittgenstein]] and instrumental towards convincing Wittgenstein to return to Cambridge for . Ramsey is loved by many, in the way that people would call him Frank, where we would never call Wittgenstein “Ludwig”. > \[!quote\] ramsey, aged 22 > > I don’t feel the least humble before the vastness of the heavens. The stars may be large, but they cannot think or love; and these are qualities which impress me far more than size does… > > My picture of the world is drawn in perspective, and not like a model to scale. The foreground is occupied by human beings and the stars are all as small as threepenny bits… In time the world will cool and everything will die; but that is a long time off still… Nor is the present less valuable because the future will be blank. > > Humanity, which fills the foreground of my picture, I find interesting and on the whole admirable. I find, just now at least, the world a pleasant and exciting place. You may find it depressing; I am sorry for you, and you despise me. But I have reason and you have none; you would only have a reason for despising me if your feeling corresponded to the fact in a way mine didn’t. But neither can correspond to the fact. The fact is not in itself good or bad; it is just that it thrills me but depresses you. On the other hand, I pity you with reason, because it is pleasanter to be thrilled than to be depressed, and not merely pleasanter but better for all one’s activities. Ramsey’s crossroads metaphor emphasizes the pattern of bets IS the belief you create towards a _gut feeling_, _conviction of sort_. A thing that you deem to be beautiful predicates upon a set of bets that you consider aesthetically pleasing. I would consider Rembrandt to be the best painters because his drawing speaks towards the absurdism as the facticity of life. Rembrandt individuated my liking towards more expressionist painters, such as Egon Schiele, Oskar Kokoschka, etc. The way that we take bets in aesthetics has the same form as taking bets elsewhere. One would live with consequences of choosing a direction, willingly sacrifice the alternatives knowing that such commitments will eventually [[thoughts/emergent behaviour|formulate]] your own [[thoughts/taste]]. So why jagged? The term came up in a recent conversation with my therapist. we were talking about living in a {{sidenotes[hyperabundance]: i.e post-scarcity, unprecedented [[posts/hyperabundance|interconnectedness]], post-singularity to an extent}} world, and i found myself circling the same worry, a la the tension between filtering and creating. [JZ writes](https://jzhao.xyz/posts/aesthetics-and-taste) that we need to “do things without recipes more often.” this tracks with Ramsey’s formulation of bets. Such grief arose, yet again, whilst i was fine tuning a [Qwen-based ghost-of-mine](https://huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507-FP8). there’s this moment when the loss curve suddenly drops, .i.e. _grokking_, and the model starts enforcing traits it learnt in my writing. watching this felt strange. my style is still evolving, still unfinished, yet here i was compressing a snapshot of myself into a [[thoughts/Eldritch horror|Lovecraftian horror]] of my own shortcomings. the thing it kept turning to: _i’m not able to write like Camus yet. i write Camus-adjacent. but i don’t want to write like Camus. i need to find something Aaron-shaped._ I found this fascinating, as I realized the gap started to feel unbridgeable. On one side sits my rigorous sense of what i like. On the other side sits curation speed, how fast i can filter so the _ghost_ generates queries that actually emulate how i think. Or, better, the ghost can also do the filtering as well [^problem]. [^problem]: the main problem I observe with this approach, with the over-use of [[thoughts/LLMs|LLMs]] in general, is that one will never have to walk across any _fields_ to verify anything, even if one actually likes Camus. Ramsey assumes distance has some intrinsic cost. _but what happens when_ $f(d) \to 0 \forall d \in \mathbb{N}^{*}$? what happens when you can sample every possible aesthetic direction simultaneously, and a Camus-adjacent or a Dostoevsky-adjacent or a Bakhtin-adjacent just, _poof, appear out of thin air_? In this case, the degree of belief collapses into a _preference-simulation_ instead of subjective conviction. the jaggedness comes from _having to commit_. Dostoevsky dictated _The Gambler_ in 26 days to meet a predatory contract with the publisher Stellovsky (who would have claimed rights to all his future work if the deadline lapsed), while simultaneously drafting _Crime and Punishment_. His first wife and his brother died in the same year. Gambling debts were compounding. He had to tunnel deeper into his own neurotic verbal tics, the run-on sentences that felt like traversing a vast landscape, because the urgency was material: the underground man who won’t shut up because shutting up means losing the only voice he possesses. The style emerged from constraint that was real, financial, mortal, irreversible. Bakhtin’s polyphony emerged from the made-up fantasy of Soviet society (where socialism was more of an idea than practical decisions), having to encode his own philosophical dialogue disguised inside {{sidenotes[literary criticism]: otherwise he would probably get shot or captured by the secret police}} since that is the only [[thoughts/forms of life|form]] it could exist in. As such, the jaggedness stems from this **individuated friction**, because we need friction to formulate such shape. When you are time-boxed and constrained at which books you can read in the local library, versus what your friends recommend, and what you can afford, you develop _taste_ through enough encounters such that it forms _priors_ within the same limited set, which in turns allows you to form your own worldview: - One like Rembrandt because that’s what’s is displayed in the museum that you can reach to. - You reread the same version of the books because you want to memorize exactly, **word for word** what Camus wrote about how Husserlian intentionality embodied this absurd spirits despite its contradictory appearance: > Thinking is not unifying or making the appearance familiar under the guise of a great principle. Thinking is learning all over again how to see, directing one’s consciousness, making of every image a privileged place. In other words, phenomenology declines to explain the world, it wants to be merely a description of actual experience. It confirms absurd thought in its initial assertion that there is no truth, but merely truths. \[…\] Consciousness does not form the object of its understanding, it merely focuses, it is the act of attention, \[…\] it resembles the projector that suddenly focuses on an image \[to borrow Bergsonian image\] Per Ramsey’s framework, the _crossroads_ is the wager that induces this friction, insofar as choosing a path or direction. Hyperabundance blurs this distinction. When you can sample both roads simultaneously, when the {{sidenotes[lovely robot]: I want to figure out better term to call these beings, because Shoggoth sounds mean and evil.}} can generate infinite variations of each path, when you can A/B test your life in real-time, the bet stops being a bet, and it _becomes_ a simulation of betting, a performance of conviction without the underlying structure that made conviction possible. The fucked up thing is this is why we build recommendation systems in the first place, i.e. to accelerate convergence towards a generic-optimal. We are universally {{sidenotes[curved into engagement]: from a philosophy perspective, A.J. Ayer’s positive emotivism can be used to describe this phenomenon, largely found through his seminar book [[library/Language, Truth & Logic|Language, Truth & Logic]]}}, rage-bait, click-through, time-on-page, which means they are selecting for writing that doesn’t require you to cross any fields, writing that just meets you where you are, because _they know what you like even before you know it_. In a way, these systems have the same infrastructure with those of {{sidenotes[gambling]: I don’t find anything morally wrong towards gambling, other than it is built on top of our worst nature, _power-seeking_}}. Therefore, it creates this evolutionary pressure towards styles that are considered maximally accessible and frictionless. In a world of social pressure full of acted-up performance, the jaggedness gets smoothed out because those edges are considered turn-off. I can’t and will never be able to write like Camus. I acknowledge that there are a set of technical skills that I lack. I also acknowledge that there are certain barriers for me to write in English (as a non-native speaker). But the underlying issue about “becoming Camus” or adapting to any tone of writers you aspire to be is that it will require you to make irreversible bets about what matters in your life. Camus staked his life on certain ideas about absurdism and resistance, and wrote in a particular historical moment where those stakes were real. His style emerged from constraint, from resistance, from the friction between what he wanted to say and what the world allowed him to say. For me, I’m trying to develop a style in an environment of infinite accommodation, where the “machine of loving grace” will happily generate Camus-adjacent prose for me, where i can retrieve every POSSIBLE variation of absurdist philosophy ON EARTH, where nothing commits because everything’s reversible. If jagged taste is the shape of commitment, then this environment is designed to sand those edges down. _I don’t want this._ _So you must resist, or might as well die._ --- ## being held against lonesome - \[meta\]: - date: 2026-01-06 06:58:46 GMT-05:00 - tags: - o/eschatology - p/fiction - socials: - substack: _tw: death, gore, body horror_ > What prepares men for totalitarian domination in the non-totalitarian world is the fact that loneliness, once a borderline experience usually suffered in certain marginal social conditions like old age, has become an everyday experience\[…\] > > —From _The Origins of Totalitarianism (1951)_, Hannah Arendt _the human nervous system registers absence of touch as a form of chronic low-grade stress, measurable in cortisol levels and immune function degradation_ in the progress of building a lossless silhouette of someone i recently came across, i found myself lying here, in this room, a kind of coffin but with better lightning, wondering: _what is it exactly that i want? what is the emergent feelings that my brain trying to resist here?_ to be held, yes, but by whom? by anyone? no, that obscene, that’s the logic of the animal, and i am not an animal, i read books, i try to articulate and put it down on paper, i’ve my own opinion on building [[thoughts/llms|my ghost]], critiques against the frankfürt school, i have refused to download tinder on principle, i resisted getting a new phone number to create a new hinge account. yet here i am, at, with my arms wrapped around my own chest like a man trying to hold his organs inside after a wound, which is precisely what it is, isn’t it, a wound, except there is no blood, there is nothing to point to, a doctor would find nothing wrong with me and that is the worst part, that i am entirely healthy and entirely dying at the same time, avoiding working on an interview in 4 hours. the finns have a word, _kalsarikännit_, which loosely means to _drink alone in your underwear with no intention of going out_, and they have made this into a virtue, and i think perhaps i should move to finland, i should go to the forests where the wolves are, and i would lie down in the snow and the cold would be a kind of touch, wouldn’t it, the cold touches everything, it is promiscuous with its attention, it does not ask whether you’ve earned it, it simply arrives and holds you and does not let go. but that’s not what i want either. _maybe in 20 years_ > Loneliness is the experience of being deserted by all human companions > > —Hannah Arendt \[waking up…\] > \[…\] loneliness is a subjective internal state. It’s the distressing experience that results from perceived isolation or unmet need between an individual’s preferred and actual experience > > —Our Epidemic of Loneliness and Isolation, _U.S Surgeon General’s Advisory, (2023)_ what i want, really, (and here is where it becomes humiliating, where i must avert my eyes from myself), is something so specific that it cannot be named without losing it. if saying it out loud, i afraid i will lose it forever. i want a particular weight of a particular arm wrapped across my chest. i want to hear breathing of the specific being, that is not my own. i want that part of computation to work, for once, in a while, for the feeling to find its proper substrate, for the thing i built in my mind to correspond to something external that does not flinch, that does not relocate to another city, that does not send the message that begins “i’m sorry for misleading you but still i want what is best for you.” DOSTOEVSKY’S UNDERGROUND MAN CLAIMS HE DOES NOT WANT THE CRYSTAL PLACE. he wants to want, _which is different, which is worse,_ because wanting-to-want means you are outside even your own [[thoughts/desire|desires]], watching them as if a man watches fish swimming in an aquarium. I am watching my need for touch and I’m disgusted by it and I am also INSIDE it, drowning in it, and this is the contradiction that Socrates or Kierkegaard or Merleau-Ponty or Simone Weil cannot address. \[in my dream…\] i’m in finland, and i’ve been walking aimlessly for hours. (_my body has begun its long negotiation with temperature_) _first stage initialized:_ shivering, which is to say my muscles are burning glycogen in small desperate contractions, 200-250 per minute, calculating heat-loss exceeds heat-production and activating the ancient mammalian subroutines, and i think: _this is what it means to be held by biology, to have something inside you that wants you to live, even when you have stopped wanting it yourself._ the grey ones are watching. i can see them between the birches, which are white like bones, like those i will soon become, and the wolves are patient bc patience is what 40,000 years of evolution has taught them, that the cold does most of the work, and the two-legged things eventually stop moving if you wait long enough. i’m not afraid of them (_this is either the self-made courage or the first symptom of cognitive decline from reduced cerebral blood flow. i suspect it to be the latter_) _by the second stage_ i have completely forgotten why i came here. the shivering has stopped, as if my body has abandoned the last arithmetic it is programmed to do, and deemed the equation unsolvable. i’m conserving what remains for the core organs, the heart, the lungs, the brain, that is still producing these sentences though it has no reason to, though no one will read them, though they are being written in a medium that does not exist to the [[library/Our Knowledge of the External World|external world]], which is to say that i am thinking, still, for no one, into [[library/Being and Nothingness|nothingness]], and this is the underground man’s final joke, that [[thoughts/Consciousness|consciousness]] persists past the point of utility, that i am AWARE of my demise and cannot stop being aware, cannot simply become the object i wanted to become. the varg, the susi, the canis lupus, they are closer now. i can see the vapour of their breathing. they are metabolizing, converting matter into heat into motion into patience, and soon they will convert ME into these things, and is this not what i asked for? to be held? the wolf’s jaw is a kind of holding, the teeth that close around the throat are intimate in a way that nothing else has been, and i think of everyone who has touched me and how none of them touched me like THIS, with such complete [[thoughts/Attention|attention]], such focus, such unwavering PRESENCE. _the third stage_. i am so hot, fuck me. this is wrong, i know it is wrong, the air is negative thirty for fuck sake! i’m pulling off my coat, my sweater, and the grey church assembles around me, six of them, eight, i have lost the ability to count, and they are watching me undress like i’m performing a ritual, and perhaps i am, perhaps this is the only sacred thing left, to give yourself to the forest, to stop being a subject and turn into a meal. the alpha, she does not go for the throat. _this is not how it’s supposed to happen_. she goes for the flank, then my genitals. (_makes sense, because efficiency matters, especially for Mother Nature. she is ruthless, because she doesn’t care much for romanticism_.) they couldn’t care less about the quick death, they care about calories, as i feel her teeth entering my calves, and i think: this is the touch i wanted. this is what being held must’ve felt like. this is the weight of another creature’s attention, in its totality and undivided, AND THE PAIN IS EXTRAORDINARY. the pain is the most real thing that has ever happened to me, and i am finally, FINALLY, not in my head, not watching myself from outside. i am HERE, in this body, in this moment, in this mouth. the hemoglobin has a viscosity of approximately four centipoise, but it moves faster when the heart is panicking, and my heart is panicking, as a biological mechanism. it is doing its job, pumping blood out of me and onto the snow, where it steams for a moment before freezing, and i watch my own warmth vanish from the physical body and become part of the landscape. this is what you want right? _to stop being contained, and leak into the world, to be held by everything instead of nothing._ the pack feeds, and i’m still fully conscious at this point, _which shouldn’t be possible_. but apparently consciousness is the last thing to go, the brain hoards its glucose like a miser, and so i am aware of being devoured by the function of nature, i am aware of becoming less, and there is something almost erotic about it. no, not erotic, that’s wrong, something…ECONOMIC, a transaction finally completing, as i’m paying my debt to the biosphere, and i’m repaying the calories i consumed, i.e. the pasta with wine, in addition to those elaborate dinners for people who did not stay. the fenrir, the old wolf, the myth-wolf, she is eating my liver and i am thinking about Prometheus, who had this done to him daily as punishment. This is not punishment, my dear Prometheus, this is a gift, as in the world accepting to what i offered, and Prometheus, you were wrong to scream. You should have been grateful, to be wanted so completely, to be USEFUL, to have eagles return for you again and again because you are worth returning for. i am less than i was, perhaps 70 kilograms becoming 60 becoming 50 becoming 40, and the wolves are becoming more, as Lavoisier’s principle states. i am becoming six wolves eight wolves, i am becoming the forest, i am becoming the snow that will melt in spring and flow into rivers and eventually into the sea, and is this not what loneliness always wanted? to stop being one thing and become all-of-things? to be held by the entire world bc you are now INSIDE the entire world? _the last thing i feel is not pain._ less sensation, less thinking, less, _me_. but more the cold ground against my back, pressing up into me as gravity presses down. the last firing neurons from the dying brain towards the remaining muscles, and so what? for what purposes do i feel this way? maybe, this is the embrace i asked for wandering the Finnish forest, the planet itself holding me against its chest, and the wolves are eating and the stars are watching and i am i am i \[waking up once more…\] the loneliness will just become a second skeleton residing inside my skeleton, and i will tell no one about it, bc who could i tell, and what would they do, and would their doing-something not simply be another form of the wrong substrate, another computation that fails to compile? i want to stop moving. i want to be still and held and i want to stop wanting. the wolves are patient. the snow is patient. but i’m not. i’m lying here, in this room, in the early morning of January winter, sweaty af after a 30-minute-hypertrophy-workout, in this coffin with wifi, typing this in neovim, waiting for something to arrive. waiting for something wait. --- ## ex machina - \[meta\]: - date: 2025-12-29 02:48:33 GMT-05:00 - tags: - llm - love I want to build my own [[thoughts/LLMs|ghost]], an autistic, psychopathic qwen3-aarnphm. But I’m afraid I don’t know how to encode the mixed feelings of love and relationship towards friends onto the latent space. --- ## on loving memories - \[meta\]: - date: 2025-12-23 03:50:33 GMT-05:00 - tags: - o/eschatology _tw: death_ If I’m away, I want to be remembered as _a loving friend_. Not as a 1% delta in Blackwell kernels. Not as a footnote in structured outputs. Not as another unit in the scaling war. Just a friend: the kind who noticed the human in front of them. If I’m away, I’m sorry for the externalities. I’m sorry for the unfinished loops—calls not returned, goodbyes deferred, care I postponed until it expired. I’m sorry for not saying goodbye to my parents. I’m sorry if your last interaction with me is just logistics and entropy. I’m sorry for the rotten food in the fridge that you’ll have to clean. I’m sorry that the smell of rotten flesh will be the last thing you’ve remembered me by. If I’m away, I’m not sorry in the sense of conceding that the universe gets to assign me a moral score. I did the best I could with the compute I had. If there’s a capital-G-agent-or-man-or-being running evaluation, it can inspect the source code; if there isn’t, then guilt is just a heuristic firing after the fact. I’m not sorry, because I just need some quiet times, sometimes. If I’m away, please don’t be sad in the way that freezes your future. Grief is a computation; run it to completion, then spend the rest of your life-budget on living. Death doesn’t get to claim a second tax in the currency of your attention. Please don’t be sad, because life is the existence of suffering, and the call of the eternal life. Finally, if I’m away, treat my savings like an allocation problem. Put it toward cancer research. Sell the farm so the animals get continuity. Liquidate the S\&P 500 if you need liquidity. Pay whatever taxes you legally must (yes, the state will still take its bite). If you want a ritual, put a small fraction on red and salute the god of variance. Then go look after your people. Don’t reserve part of your heart for me. I don’t deserve any part of it. --- ## escrow - \[meta\]: - date: 2025-12-12 18:36:27 GMT-05:00 - tags: - process ![[thoughts/Agency#seven ways to become unstoppably agentic]] --- ## on learning through presence - \[meta\]: - date: 2025-12-12 12:40:40 GMT-05:00 - tags: - process from [on learning through presence](https://www.humaninvariant.com/blog/presence): > When you show up in person, you feel like you don’t belong. You quickly learn that others have a deep and rich shared cultural history that has spanned over the better part of a decade, while you are a newcomer. You try your best: you connect with people during [late night conversations](https://www.humaninvariant.com/blog/conversations) you would have never connected with otherwise, and maybe even plan to throw a joint party together in New York. > You learn more about the shape of the life you want, the types of relationships you want with certain people, and the sacrifices you are willing to make to be ambitious. You learn that ambient ambition is real, and what it really means to love what you do, when you witness someone over twice your age work into the night and enjoy every moment. > You learn what it feels like to use a typewriter because somebody else cared enough to give you the opportunity to experience that feeling and made it happen. You use that opportunity to ponder [the market structure of writing implements](https://www.humaninvariant.com/blog/lowercase) and to write an endearing note to the people who have given you the opportunity to learn more about yourself. While doing so, you learn what the sound of a bell means when typing on a typewriter. > You learn that having the courage to care is the scarcest resource in the world. You can [predict the future you want by caring enough to build it](https://www.humaninvariant.com/blog/worldbuilding). > Most importantly, you learn that the ability and desire to care is built through presence. there’s a striking prior when comparing [[thoughts/writing|writing]] with presence. both are processes where the transformation happens _during_ the act of doing it, rather than afterwards. You’re thinking out loud when you write, and oftentimes thoughts are then lossy collections of ideas that you’ve scribbled onto the pages, and soon thereafter realising half of them are {{sidenotes[wrong.]: I don’t mean in the literal sense, but rather an incoherent/illogical collections of word-pile-that-you-vomit-out-onto-the-page.}} presence sort of work the same way here _(or at least from the blog)_ where you discover it by noticing which conversations make you lean in, which type of energy you want to absorb, which futures you find yourself involuntarily imagining. You can believe in this to your heart’s content, but I think they both look like search algorithms running on wetware-presence searches possibility-space for the life-shape that fits with your lived experience. The result of such a product is merely an evidence that such _process_ occurred. I wonder if this is why reading essays and consuming YouTube videos produce little lasting change if you aren’t really putting effort into actually studying the subject at hand. In a sense, you’re watching the residue/product of someone else’s search process, instead of running your own. The illegible inputs—the drafts, the small talk, the wrong turns—contain the actual epistemic work, which is completely removed from the final products, is the thing that matters the most. --- ## thingspace - \[meta\]: - date: 2025-12-03 09:33:42 GMT-05:00 - tags: - love ![[thoughts/love#hw]] --- ## crying - \[meta\]: - date: 2025-11-30 11:01:47 GMT-05:00 - tags: - o/life ![[quotes#^shed]] --- ## on omnipotence paradox - \[meta\]: - date: 2025-11-29 13:01:53 GMT-05:00 - tags: - ontology - philosophy can an [omnipotent](https://iep.utm.edu/omnipote/) being create a stone so heavy they cannot lift it? if yes, then they cannot lift it—failure of omnipotence. if no, they cannot create it—failure of omnipotence. either way, {{sidenotes[omnipotence fails.]: atheological arguments use this not primarily as evidence against god’s existence but to show “omnipotence” as traditionally conceived may be incoherent.}} the paradox reveals that “maximal power” might be conceptually malformed, like “set of all sets” or “north of the north pole.” not difficult to achieve but impossible to coherently {{sidenotes[specify.]: it is _impossible_ to create an uncreated object—not because of limited power but because the phrase doesn’t describe a possible state of affairs.}} SEP’s definition of _omnipotence_ follows: > Omnipotence is maximal power. This [comment](https://philosophy.stackexchange.com/a/34397) states that what we are often referring about the paradox is synonymous to a _absolutist proposition_. The three resolutions are as follow: 1. The notion of an absolutely immovable physical object is logically incoherent. - To be a physical object means being subject to physical forces, which means having some finite mass or hardness. - By definition, anything exceeding that could move or alter it. - The concept of an unliftable object is {{sidenotes[self-contradictory.]: [Aquinas](https://en.wikipedia.org/wiki/Thomas_Aquinas) in _Summa Theologica_ I, Q.25, Art.3 argues inability to do contradictory things “does not signify a defect of power” (_non significant defectum potentiae_). Apparent “inabilities” like inability to sin represent perfection of power, not deficiency. Self-contradictory pseudo-tasks aren’t genuine objects of power.}} - This is Thomistic scholasticism view of the paradox. (or _pseudo-task dissolution_) 2. That any limitation on God is a form of self-restraint rather than fundamental limitation. - In other words, God can create an object God says God cannot move, and God won’t move it - But not because it is {{sidenotes[immovable]: People that take this view and think there’s a God would be committed to a form of voluntarism.}} per se but instead immovable per volens. - This in turn can also be extended by Descartes’ argument where _logical necessities themselves are contingent on divine will_ - Decartes’ letters to Mersenne (1630) would emphasize that _god could have made it false that twice four equal eights_ if he wishes so [^letter-to-mersenne] 3. That God can impose self limitations that stand permanently. - In other words, God can make a rock God cannot lift. - Again, the origin wouldn’t be that the rock has infinite mass but that God can manufacture the rock and bind a condition on God’s own self to not be able to {{sidenotes[^pick up the rock]}}. - (see the footnotes for _postmodern_[^postmodern] interpretation) {{sidenotes[pick up the rock]}}: The closest pre-modern candidate for this can be traced back to Lurianic Kabbalah’s _tzimtzum_ (divine contraction, 1570s Safed), where the Infinite (_Ein Sof_) withdraws to create “vacant space” for finite existence. But then, _tikkun_ (repair) implies eventual restoration, and _reshimu_ (residual trace) suggests God never fully withdraws. Which implies there are resistant from making it absolute. More importantly, the _potentia absoluta_ vs _potentia ordinata_ distinction (originated by Hugh of St. Cher in 1230s commentary on Lombard’s _Sentences_, later refined by [Scotus](https://plato.stanford.edu/entries/duns-scotus/#ProExiGod) and [Ockham](https://plato.stanford.edu/entries/ockham/)) holds that God _reliably would not_ deviate from ordained commitments due to divine faithfulness, not that God _could not_. This implies God _chooses_ not to intervene with such objects, rather than inability to do so. [^letter-to-mersenne]: Don’t hesitate to assert and proclaim everywhere that it’s God who has laid down these laws in nature just as a king lays down laws in his kingdom. There’s not one of them that we can’t grasp if we focus our mind on it. They are all inborn in our minds, just as a king would, if he could, imprint his laws on the hearts of all his subjects. God’s greatness, on the other hand, is something that we can’t •grasp even though we •know it. But our judging it to be beyond our grasp makes us esteem it all the more; just as a king has more majesty when he is less familiarly known by his subjects, provided they don’t get the idea that they have no king—they must know him enough to be in no doubt about that. You may say: - ‘If God had established these truths he would have been able to change them, as a king changes his laws.’ To this the answer is: - He can change them, if his will can change. - ‘But I understand them to be eternal and unchangeable.’ - And so is God, in my judgment. - ‘But his will is free.’ - Yes, but his power is beyond our grasp. In general we can say that God can do everything that we can grasp, but not that he can’t do what is beyond our grasp. It would be rash to think that our imagination reaches as far as his power. [^postmodern]: The current disagreement surrounds whether the limitation is considered _voluntary_ or _essential_. 19th century German kenoticism (Thomasius, Gess, Ebrard) developed systematic kenotic Christology from Philippians 2:5-8:

5 In your relationships with one another, have the same mindset as Christ Jesus:

6 Who, being in very nature[a] God, did not consider equality with God something to be used to his own advantage;

7 rather, he made himself nothing by taking the very nature[b] of a servant, being made in human likeness.

8 And being found in appearance as a man, he humbled himself by becoming obedient to death— even death on a cross!

Philippians 2:5-8

Here, the logos divested “relative” attributes (omnipresence, omniscience, omnipotence) but retained “immanent” ones (holiness, love, truth) during Incarnation. Critically: _temporary_, ending with Resurrection. The limitation is real but bounded—a divine hiatus, not a permanent restructuring. Orthodox critics (Chalcedonian) argue this creates a “binity problem”: if Christ divests divine attributes, the Trinity breaks during Incarnation. The second Person takes leave from Godhead. If Father and Spirit retain omniscience while Son doesn’t, they can’t share the same substance. Thomasius relocates the two-natures problem rather than solving it. Can a genuinely free being make an _irrevocable_ choice? - If revocable, the limitation isn’t permanent (Thomasius). - If irrevocable, has freedom been compromised (Polkinghorne)? - This paradox of self-binding runs through political philosophy (constitutionalism—can one generation bind the next?), personal ethics (promising—can you obligate your future self?), and now theology—without clean resolution. - The voluntarist and essentialist positions may be unstable in ways that mirror the original omnipotence paradox they sought to escape. > If God can limit divine attributes, then why assume any are essential to begin with? I wonder if we should define omnipotence via **act-theory** (ability to perform any logically consistent action) or **result-theory** (ability to actualize any possible state of affairs)? Result theories handle the paradox where “there being a stone an omnipotent being cannot lift” isn’t a possible state of affairs, therefore the _inability_ to actualize it is no limitation. But result theories carry heavy metaphysical commitments that they require omnipotent beings exist **necessarily** and may constrain human freedom. The paradox reveals more about how we think about power than about divine attributes. We model omnipotence on human power “only without limitations”—but maybe the category doesn’t scale. Maybe maximal power is QUALITATIVELY different, not quantitatively maximal. Maybe “maximal power” as conceived in agent-causal term is incoherent at infinite {{sidenotes[extension.]: J.L. Cowan (1965, 1974) argues any attempt to resolve the stone paradox “must fail”—the concept itself is definitively broken. Anthony Kenny’s _the god of the philosophers_ (1979) concludes “there can be no such being as the god of traditional natural theology.” not “we need better definitions” but “abandon the project.”}} If anyone is familiar with [[thoughts/Wittgenstein#Russell's paradox and the vicious circle principle|Russell's paradox]], then the similarity is uncanny here. “set of all sets that don’t contain themselves” generates the paradox via self-reference. Type theory only really replaces the naive concept with a more restricted versions, i.e there is no universal set. Similarly, “god can do anything” generates the stone paradox through self-references. if the solutions for Russell’s paradox is conceptually malformed, then the same would hold for “maximal power”. [^cantor] [^cantor]: ``` Patrick Grim extends this via Cantor, where any set of truths has more subsets than members, each corresponding to a unique truth. ``` There’s no totality of truths—truth “explodes beyond any attempt to capture it.” if omnipotence means power over all possible states of affairs, and there’s no totality of possible states of affairs (by analogy to cantor), omnipotence is malformed. ``` You can have very extensive power, but "maximal" or "unlimited" may be incoherent the way "set of all sets" is incoherent. ``` Cantor distinguished absolute infinity _in deo_ from mathematical transfinite, arguing absolute infinity is “logically inconsistent” and belongs to speculative theology, not mathematics. he avoided treating the mathematical universe as a set, recognizing the paradoxes this generates. What makes something a “task”? Tasks presuppose: - initial conditions - final conditions - causal pathway between them - possibility of failure “Create a stone you cannot lift” presupposes you have lifting capacity $C$, stones can have weight $W > C$, creating something doesn’t change $C$. But for maximal power, there IS no $C$—no upper bound. So “stone too heavy to lift” isn’t a pseudo-task (aquinas) or ill-formed question (Frankfurt, The Logic of Omnipotence, 1964), but a **CATEGORY ERROR**. How would one reconceptualising divine power? If “omnipotence” is incoherent, what should replace it? Tillich placed god as “ground of being” rather than powerful agent. Omnipotence then becomes a symbol expressing “the power of [[thoughts/being|being]] which resists nonbeing”—not ability to perform tasks but condition-of-possibility. Heidegger’s onto-theology critique treated god as _causa sui_ or highest being is bankrupt. The entire framework of “god has power X” is malformed. God isn’t a being among beings to which predicates apply. Caputo’s weak theology then explicitly deconstructs divine omnipotence. God understood through Greek metaphysical attributes (immutability, omnipotence, omniscience) should be deconstructed. The name “god” harbors an event rather than naming a powerful being. We can then shift from “god can do X” to “god grounds the possibility of X.” Power-as-condition rather than power-as-capacity. This dissolves the stone paradox bc it’s not asking “can the condition-of-possibility create something impossible?”—impossibilities aren’t in the domain at all. I understand that the cost of this would make god _metaphysically distant from history/agency_. If you want a god who does things in history (answers prayers, performs miracles), then you’re stuck with omni-attributes and their paradoxes (which is fine). But I find these concepts/positions of God makes more sense for the logical brain. Kenotic theology tries to have both—a god who acts in history but isn’t bound by classical omnis—but as we show above, every version either relocates the problem (voluntary kenosis) or diminishes God to where “deity” seems honorific (essential kenosis). There might not BE a stable middle ground. I then come to the conclusion where the omnipotence paradox is a genuine antinomy (in the Kantian sense)—concepts applied beyond their legitimate domain generate contradictions. Therefore, the right response isn’t solving it within the framework but abandoning the framework for once. Standard solutions (pseudo-task dissolution, voluntarism, kenosis, Frankfurt) either deny one horn or simply relocate the problem. (big cope really) --- ## exploring myself - \[meta\]: - date: 2025-11-28 16:11:39 GMT-05:00 - tags: - writing ![[thoughts/writing#as a journey for exploration]] --- ## misalignment based on training data - \[meta\]: - date: 2025-11-27 15:57:58 GMT-05:00 - tags: - alignment Anthropic found that natural emergent misalignment stems from [_reward hacking_](https://www.anthropic.com/research/emergent-misalignment-reward-hacking). Though, I suspect that ablating these “bad behaviour” wouldn’t necessarily make the model more aligned. What if having certain malicious intent is actually helpful? --- ## processing emotions - \[meta\]: - date: 2025-11-25 13:15:26 GMT-05:00 - tags: - love - emotions there’s a towel i used for when L stayed over. still haven’t washed it. keeping it means keeping the rot—letting those feelings decay until the whole thing becomes unbearable enough that throwing it away becomes a necessity rather than being a choice. --- ## drinking - \[meta\]: - date: 2025-11-22 14:18:52 GMT-05:00 - tags: - o/life - o/relationship I got a cup of hot chocolate today. It reminded me of L, and somehow, we carry fragments of them within us without knowing so. --- ## hatred - \[meta\]: - date: 2025-11-21 12:43:25 GMT-05:00 - tags: - o/relationship I have no desire of making new friends in Toronto anymore. Everything felt so superficial here. --- ## to love is to die slowly - \[meta\]: - date: 2025-11-20 13:07:46 GMT-05:00 - tags: - love

To find someone worth fighting for is a beautiful thing.

James, All At Once

--- ## i’m done - \[meta\]: - date: 2025-11-17 14:55:06 GMT-05:00 - tags: - love ![[quotes#^hopeless]] --- ## memories - \[meta\]: - date: 2025-11-15 12:36:29 GMT-05:00 - tags: - o/life Got on a call with middle school friends, seeing my old Vietnamese teacher, B—there’s warmth in it. Good memories, honestly. But I’m glad I left when I did. --- ## sprints of a relationship - \[meta\]: - date: 2025-11-15 01:58:17 GMT-05:00 - tags: - love Why does this hurt so bad? We aren’t even in a relationship. --- ## i’m glad we met - \[meta\]: - date: 2025-11-13 11:33:37 GMT-05:00 - tags: - love I’m completely broken down, seeing the physical letter L left on the counter. I’m at a loss for words. --- ## fml - \[meta\]: - date: 2025-11-13 05:44:12 GMT-05:00 - tags: - love - feeling Love sucks. love hurts. love prevails. I want to be loved. and yet, all I got are heartbreaks. I’m once again, perplexed by my own feelings, a sense of _saudade_ --- ## centering women - \[meta\]: - date: 2025-11-13 01:08:39 GMT-05:00 - tags: - love

It's possible to love men without rage. There are thousands of ways to love men.

Lidia Yuknavitch, The Chronology of Water

## on suicide - \[meta\]: - date: 2025-11-13 01:01:11 GMT-05:00 - tags: - philosophy - death I was reading [On suicide](https://docs.google.com/document/d/14XZJtJcMGzD4ZY6AgaTzobunndRvXZMGUybA6JIQOj4/edit?tab=t.0) by [Alexey Guzey](https://guzey.com/) > When I was in high school, I spent a year trying to kill myself. I just couldn’t do it. At some point I decided to make my life as bad as I possibly could, but nothing worked. No matter how much I tried, I still wanted to be alive more than I \[_wanted to be dead_\]. The [[thoughts/ethics#deontology|deontological]] arguments of life _presupposes_ the feeling of interconnectedness. I think that if one acknowledges that no one wants one in life, then one might conclude that one’s life is not worth living. [[thoughts/Philosophy and Kant|Kant]] would then argue against the notion of “kill oneself” given that it is an act in violation of the [[thoughts/moral|moral]] law, and deemed it as **wrong**. > I wonder how much of this is due to the lack of imagination. If you’re suicidal, it’s very difficult to imagine life getting better. You’re depressed, sure, so your cognition is impaired, such that you realised life isn’t worth living anymore, which then concludes that suicide is the real solution. So your mind convinces you to just **give in**

The impulse to end life and the impulse to further life contradict each other.

Kant, Groundwork §4:422

But I’m troubled by how close this gets to “just shake it off” or “try something new!” as if suicidal depression were boredom with your routine. The difference, maybe, is that the conventional advice assumes you can white-knuckle through by force of [[thoughts/Will|will]] or positive thinking. Guzey’s argument focuses on a darker path, where you can’t think your way out, and you probably can’t will your way out either. You need an external disruption large enough to short-circuit the entire system. Which is less hopeful than it sounds—e.g: you’re not in control of your own recovery. You’re waiting for something to happen that breaks the pattern, or you’re throwing yourself at random experiences hoping one of them will. The wanting to live that exists underneath reasons, that persists even when you’ve eliminated every reason you can name. Whether that’s the body asserting itself, or some pre-rational attachment to [[thoughts/being|being]], or just biochemistry doing its thing. What brought me back wasn’t a good argument for living. It was N’s voice, L’s conversations, the way light looked one morning, the prospect of missing something I hadn’t experienced yet. The stuff that doesn’t make sense in the ledger of reasons but somehow weighs more than the whole column of evidence against continuing. I think, subconsciously, we all have a moral duty to continue to {{sidenotes[live,]: go outside, look at trees, eat an onion sandwich, buy some sourdough}} and only when you have exhausted all possible solutions, then suicide is the last reasonable solution to end the absurd life. I’m once again, thinking about suicide. --- ## city - \[meta\]: - date: 2025-11-12 21:33:51 GMT-05:00 - tags: - o/life

But Kurt Vonnegut writes about the difference between two kinds of teams. A granfalloon is a team of people pushed together for some ordinary human purpose, like learning medicine or running a hospital psychiatry department. They may get to know each other well. They may like each other. But in the end, the purpose will be achieved, and they’ll go their separate ways.

Scott Alexander, To the Great City

--- ## patterns - \[meta\]: - date: 2025-11-10 04:30:10 GMT-05:00 - tags: - pattern > Each describes a problem which occurs over and over again in our environment, and then describes the core of the solution to that problem, in such a way that you can use the solution a million times over, without ever doing it the same way. > > —[[library/A Pattern Language]], p. x --- ## connections, connections, connections - \[meta\]: - date: 2025-11-09 01:34:25 GMT-05:00 - tags: - technical I’ve been thinking about connectionist networks lately (or for the past two years or so), and there’s something deeply unsettling about how we talk about them. Not unsettling in a bad way—more like that productive discomfort you get when you realize the categories you’ve been using don’t quite map onto reality. The whole connectionist project started as a rejection, really. A rejection of the idea that intelligence is symbol manipulation all the way down. Back in 1986, when Rumelhart and McClelland dropped their PDP volumes \[@rumelhart1986parallel\], they weren’t just proposing a new computational architecture—they were making an ontological claim about what cognition _is_. ![[thoughts/Connectionist network#{collapsed: true}]] --- ## compression into LLMs - \[meta\]: - date: 2025-11-05 05:48:09 GMT-05:00 - tags: - o/life My journal is my [[posts/index|blog]]—not because I want to become a blogger, but because it’s a permanent state of [[thoughts/Eldritch horror|eldritch horror]] etched into [[thoughts/LLMs|GPT‑X]]’s compressed mind, all about _myself_. Til the day I’m plugged into the mainframe, it’s as if I never left. --- ## typing - \[meta\]: - date: 2025-11-04 21:11:21 GMT-05:00 - tags: - o/life I find myself using my mechanical keyboards less and less nowadays, using my laptop keyboard instead. This might have to do with the mode of focus the laptop keyboard puts me into—something about focusing on the work itself rather than the tools. Partially because of the wrist pain from long sessions of working at my desk 🐙 --- ## kernels, updates - \[meta\]: - date: 2025-10-31 18:24:43 GMT-04:00 - tags: - o/life Writing kernels sounds way more fun than whoring on the streets of Toronto. Happy Halloween 🎃 though. --- ## essays - \[meta\]: - date: 2025-10-30 03:18:38 GMT-04:00 - tags: - writing

Writing essays, at its best, is a way of discovering ideas.

Paul Graham, The Best Essay

> An essay should ordinarily start with what I’m going to call a question, though I mean this in a very general sense: it doesn’t have to be a question grammatically, just something that acts like one in the sense that it {{sidenotes[spurs some response.]: When you find yourself very curious about an apparently minor question, that’s an exciting sign. Evolution has designed you to pay attention to things that matter. So when you’re very curious about something random, that could mean you’ve unconsciously noticed it’s less random than it seems.}} --- ## learning on philosophers - \[meta\]: - date: 2025-10-29 05:18:28 GMT-04:00 - tags: - productivity - philosophy Late night work listening to Dreyfus’ lectures hits like smoking a good joint on a Friday night. ![[https://www.youtube-nocookie.com/embed/usxvyf3xqcQ]] --- ## love in math - \[meta\]: - date: 2025-10-28 16:38:29 GMT-04:00 - tags: - math 
Try as you may,

you just can't get away,

from mathematics

—Tom Lehrer
--- ## beauty in writing - \[meta\]: - date: 2025-10-28 16:15:23 GMT-04:00 - tags: - writing

The reason I've spent so long establishing this rather obvious point [that thoughts/writing helps you refine your thinking] is that it leads to another that many people will find shocking. If writing down your ideas always makes them more precise and more complete, then no one who hasn't written about a topic has fully formed ideas about it. And someone who never writes has no fully formed ideas about anything nontrivial.

It feels to them as if they do, especially if they're not in the habit of critically examining their own thinking. Ideas can feel complete. ==It's only when you try to put them into words that you discover they're not==. So if you never subject your ideas to that test, you'll not only never have fully formed ideas, but also never realize it.

Paul Graham, Putting Ideas into Words

--- ## to live - \[meta\]: - date: 2025-10-27 15:01:07 GMT-04:00 - tags: - o/life

I wanted to eat life by the mouthful, to devour it, to be swallowed up in its dizzying vertigo, to be both actor and spectator, to possess and be possessed, to discover and to create, to make of my life a work of art.

Simone de Beauvoir, Memoirs of a Dutiful Daughter (1958)

--- ## content - \[meta\]: - date: 2025-10-27 09:34:45 GMT-04:00 - tags: - o/life ![[quotes#^violent]] --- ## epiphenomenal status - \[meta\]: - date: 2025-10-27 07:37:15 GMT-04:00 - tags: - journal I find myself the most productive while procrastinating other tasks. --- ## epistemic suicide - \[meta\]: - date: 2025-10-26 01:41:59 GMT-04:00 - tags: - death - philosophy [[thoughts/Camus]] begins with suicide in [[library/The Myth of Sisyphus]], more or less a demonstration of how absurd life really is. He used suicide as a scapegoat of people who don’t have enough courage to deal with the hardship of life being thrown at them, and considered suicide as cowardice. I have exercised this thought _many times_, in the way one considers a standing appointment. Not with desire of actually doing the deed or the fear of death, but rather a procedure to keep my mind sharp. The world offers no meaning, yet I require meaning, therefore, it doesn’t seem reasonable to be alive. If I could leave, what makes me stay? When I actually sit with this question, as a real possibility, the answers surprise me. I want to see how this conversation with N resolves on Thursday. I’m still waiting for L to send that letter my way. There’s a problem on my desk I’m halfway through understanding, and I need to know if my intuition about it is right. I think about my parents getting the phone call, my friends having to sort through my things, and something in me recoils not from death but from inflicting that particular grief. Turns out the results are mostly attachments. They are somewhat very small, specific to my life, but accumulate gradually. I watch people in coffee shops, on trains, in office buildings. They seem to continue, mindlessly. Felt like aimless drones just letting time pass through their body. Pour a cup of coffee, send those emails, make plans for next week. Perhaps they have solved something I have not. Perhaps they have simply never filed the paperwork for the question. It is possible everyone considers this and we have agreed, collectively, not to mention it. Like a standing meeting no one enjoys but everyone attends. My work is not that special, my contributions to the collective projects are temporary. Life won’t change a lot if I disappear. Heat death will erase everything eventually. Still, I’m here though. Maybe because of a particular arrangement of attention and time, maybe because I still care enough, maybe the thought of striving another day for N, parents, L, J, C, S are good enough motivators to keep one going. > Meaning does not require permanence. This seems important but I am not certain why. [[thoughts/Camus]]’s revolt: Continuing with full knowledge of the absurd. Choosing again each day, not from habit but from decision. I am not certain I achieve this. Most days feel like habit. The question becomes dangerous when it stops being theoretical. When it moves from mind to body. When the weight becomes physical rather than philosophical. Then one must interrupt the process. Call someone. Leave the room. The distinction is administrative: one is philosophy, the other is emergency. Emergency requires different procedures entirely. --- ## why superhuman AI won’t kill us all. - \[meta\]: - date: 2025-10-25 23:26:00 GMT-04:00 - tags: - llm —[Eliezer Yudkowsky](https://www.youtube.com/watch?v=nRvAt4H7d7E) Yudkowsky’s full argument on eschatology isn’t productive whatsoever. Feels a lot more science fiction writing, where he claims that these systems will end up “wanting to do their stuff without wanting to take the pills that \[we offer to\] makes them to do stuff that we _wants_ them to do instead.” Yudkowsky’s claim relies on [[thoughts/AGI|superintelligent]] optimizer + misaligned goals = human extinction, a sort of self-recursion improvement towards its’ own power-seeking agendas. bc we’re made of atoms it can use for something else (i.e a resource allocation problem). Nanotech grey goo. Designer pathogens. Trees that grow computer chips. All physically possible, therefore inevitable once you have sufficient optimization power. But intelligence in actual systems is jagged, domain-specific, constrained by architecture. AlphaFold is brilliant at protein folding and useless at everything else. [[thoughts/LLMs|GPTs]] can write code but can’t reliably count letters (update from 01/09/26: not anymore). The jump from “good at prediction” to “can design novel molecular machinery from first principles” assumes transfer learning we haven’t seen. These very much resemble the old GOFAI vs NFAI arguments. Maybe the argument here is to have a _composition of multiple domain-specific superintelligence systems_ that amplify our life. The “foom” scenario requires explosive recursive self-improvement, which is abstruse. GPT-6 builds GPT-7, capabilities doubling weekly until godlike intelligence. Architecturally speaking, maybe we figure out something that scales with attention, but it has to be beyond just Transformers, maybe in conjunction with something like JEPA. The argument assumes breakthroughs on demand. He did mention a recursive need for hydrogen, but fwiw physical constraints matter a lot more here. Building nanoassemblers needs: labs, materials, energy, time for experiments. Biology took billions of years of parallel search to reach cells. You can speed that up with intelligence – how much? The argument assumes “enough.” The frame requires that one assume worst case at every branch, assume maximum capability, assume minimal constraints, therefore _doom_. I just don’t think that’s how you build things. Real systems fail in boring ways. Scaling laws break. Architectures saturate. The chain from “AI breaks up marriages” to “superintelligence converts biosphere to computronium” requires assumptions that would be rejected in any engineering domain.

don't build, don't experiment, don't iterate, because any mistake might be the last. That's not how we've solved any complex safety problem. Treating everyone who continues working as equivalent to cigarette executives isn't engaging with technical disagreements.

Yudkowsky

[[thoughts/Alignment]] is hard. I do think that capabilities scale faster than safety. But the response can’t be “stop everything and hope treaties hold.” It has to be: build better systems, understand current systems deeply, develop alignment that might work. You need feedback loops. You need to learn from failures at scales where failure isn’t extinction. --- ## how I wrestle with the idea of god - \[meta\]: - date: 2025-10-24 01:23:00 GMT-04:00 - tags: - theology - o/life “God exists” and “the table exists” share the same grammatical structure, verb, declaration of being. [[thoughts/Wittgenstein|Wittgensteinians]] would tell you these are entirely different language games, different [[thoughts/forms of life]] enacted through identical syntax. When someone tells me “I feel God’s presence,” I wonder what they’re actually describing. The [[thoughts/Metaphysics|metaphysical]] claim, that an immaterial conscious being is causally interacting with your experience right now, is the biggest cope I’ve ever seen. The phenomenology itself is trickier to describe: how we say _the feeling of being held_ when no one’s holding you, _of mattering_ when the universe gives no indication you do, _of company_ in the dark when empirically you are alone?? I’ve had these feelings, once. Walking home at 3 AM and suddenly feeling like the street lights are watching over me. Finding a book at exactly the moment I needed it. That uncanny sense of [[thoughts/Alignment|alignment]], of , of something speaking directly to you through the noise of existence. Theism often offers a ready explanation here, providence and divine orchestration and God moving through your life, and I understand why people find it satisfying. Strip that away and what type of grammar are we left with to describe this a priori? - “Coincidence” dismisses the actual experience too quickly, because the word suggests you’ve explained something when you’ve only labeled it. - The feeling persists: events _mean_ something, your pattern-matching brain encounters random noise and insists there’s signal there. - This is {{sidenotes[apophenia,]: the tendency to perceive meaningful connections between unrelated things. we see faces in clouds, we find messages in static, we construct narratives from scattered data points.}} the same cognitive architecture that lets you recognize your mother’s face in a crowd, now misfiring on the universe itself. And the misfiring feels exactly like insight. - {{sidenotes[^"Synchronizität"]}} doesn’t make a lot of sense here, _meaningful coincidence_. It sounds scientific until you push on it, at which point it dissolves into gesture. Jung wanted to preserve the phenomenology without the theology, to say “this feeling of connection is real” without committing to a god who arranges it. The problem is that “acausal connecting principle” explains nothing, but a label masked as some polysyllabic word destined to be something more {{sidenotes["Synchronizität"]}}: Jung coined the term in 1952, working with pauli (yes, the physicist). the idea was that meaningful coincidences occur through some acausal connecting principle. Pauli was drawn to it because quantum mechanics had already broken his intuitions about causality. And Jung liked it because it gave his patients’ experiences a respectable-sounding name. neither of them could say what the mechanism was. “God is good” sounds like “water is wet”, a property of an existing thing. Maybe it’s closer to “justice is sacred”, an orientation rather than a description. A way of being in the world, encoded in grammatical structures that trick you into thinking you’re making claims about reality when you’re actually describing how you move through it. The analytic philosophers did try to take a stab at this problem: - A.J Ayer, in {{sidenotes[^that thin slash of a book from 1936]}}, didn’t argue that God doesn’t exist. He argued that “God exists” doesn’t say anything at all. The sentence has the grammar-of-a-claim but the content of a sigh. The claim is _meaningless_ rather than false, and if theism is meaningless then atheism is too, because there’s nothing to deny. - W\.V. Quine pushed further, though sideways, more in a economic sense: _what beliefs_ {{sidenotes[earn their keep?]: “to be is to be the value of a bound variable” (_From a Logical Point of View_, 1953). your [[thoughts/Ontology|ontology]] is what your best theories quantify over. Quine though never explicitly whether or not this applied this to god, but my inference is that _god doesn’t appear in physics or biology or any of our best theories_. The threat is that this criterion may also eliminate numbers, theoretical entities, maybe the self.}} We commit to electrons and genes and spacetime because our best theories quantify over them, they do work, make predictions, connect to other beliefs. God, therefore, a posit, same as Homeric gods, same as physical objects. Physical objects are more useful, because God doesn’t pay the rent. - Antony Flew {{sidenotes[adapted]: the parable originated with John Wisdom in “Gods” (1944). Flew’s version added the escalation of immunizing moves: invisible, intangible, insensible, eternal. “death by a thousand qualifications” is flew’s phrase.}} the invisible gardener: two explorers find a clearing, one insists a gardener tends it. They set watches and no gardener appears. The believer qualifies: invisible, intangible, insensible, eternal. Flew asks at what point the invisible gardener differs from no gardener at all, because theological claims die by {{sidenotes[death-by-qualification.]: Flew converted to deism in 2004, age 81, after fifty years of arguing against god. “what I think the DNA material has done is show that intelligence must have been involved.” the guy who invented the falsification challenge found something he’d count as evidence. the 2007 book _There Is a God_ has authorship controversy, largely ghostwritten while Flew had cognitive decline.}} and whatever would count against them gets absorbed into divine mystery. - Alvin Plantinga’s response is that belief in god might be {{sidenotes[properly basic,]: reformed [[thoughts/Epistemology|epistemology]] (_Warranted Christian Belief_, 2000) holds that some beliefs are foundational, requiring no inferential support. you don’t prove other people are conscious, instead you just find yourself **believing it**. the sensus divinitatis, calvin’s term, works the same way: a faculty that produces belief in god when triggered by beauty, guilt, gratitude. the famous objection is “why can’t belief in the great pumpkin also be properly basic?”}} like belief in other minds, something you don’t derive from evidence so much as find yourself believing. {{sidenotes[that thin slash of a book from 1936]}}: i.e. [[library/Language, Truth & Logic|_Language, Truth and Logic_]]. he was 24 when he wrote it, around 160 pages. the verification principle was self-refuting, which ayer conceded decades later: > the verification principle never got itself properly formulated. I tried several times and it always let in either too little or too much.” logical positivism died by its own sword. and then, in 1988, ayer had a near-death experience, heart stopped for four minutes, wrote about it in “What I Saw When I Was Dead.”: > i was confronted by a red light, exceedingly bright, responsible for the government of the universe.” he insisted it didn’t change his atheism. his physician would claimed otherwise. I think about that walk then, the 3 AM street lights, and I’m still uncertain what to call the feeling. “Revelation” carries too much freight, and “delusion” is revelation’s shadow, still shaped by what it denies. Ayer would say I wasn’t making claims at all, just emoting with declarative syntax. Maybe so. The feeling of meaning-making isn’t nothing, though. It’s the most robust data point I have. The positivists wanted to clear away metaphysical fog and ended up sweeping aside the only thing worth mapping. Kierkegaard said the leap of faith requires a certain degree of [[thoughts/Camus|absurdity]], that you leap beyond reason into belief. We’re all leaping anyway, every morning, into relationships we know will end, into projects that will be forgotten, into futures that terminate in death. The absurdity is pouring coffee at 7 AM when you know, really know, viscerally rather than just intellectually, that entropy wins and nothing persists and the heat death of the universe is coming for everything you’ve ever loved. And yet here we are making breakfast, answering emails, planning for Tuesday. This isn’t Kierkegaardian faith in the sense of a transcendent leap into divine arms. There’s still a leap, though: acting _as if_ things matter despite having no cosmic guarantee they do. Maybe religious language expresses this fundamental human insistence on continuing, on treating temporary patterns as if they’re eternal truths, on finding meaning in noise because meaning-making is what conscious matter does. The [[/posts/feelings|feeling]] of being held, of mattering, of not being alone, persists even after you’ve dismantled the theological framework. These feelings are built into consciousness itself, this need for connection and significance and something beyond the merely material. The feelings arise because you’re the kind of system that generates them when confronting its own existence. Strip away providence and you’re left with pattern-matching systems encountering randomness and desperately trying to read it as text, as message, as meaning. We can’t help it. It’s how we’re built. The grammar of theism gives us vocabulary for these experiences, but the experiences themselves are what happens when consciousness encounters itself in a universe that doesn’t care, when the meaning-making machine meets the meaningless void and keeps making meaning anyway, through the void rather than despite it, because of it. --- ## on hamilton - \[meta\]: - date: 2025-10-25 16:50:00 GMT-04:00 - tags: - o/feeling The visceral feeling of returning to a place you once lived. Those familiar streets you’d walk daily without really noticing. The tiny brick that used to catch your shoe—gone now. New shops sprouting up, the decommissioned church turned taco spot. The coffee shop with its familiar faces, where you just walk by and smile at each other, no words needed. That bar you used to haunt. What gets me is seeing the old faces, the same constellation of people I used to orbit. Five months in my new Toronto apartment (which I love, truly), but it feels like years have passed. Time does this strange thing – warps around you when you leave. For me, whole epochs have unfolded. For them, Tuesday follows Monday follows Sunday. The uncanny valley of temporal perception: you’ve moved through time at different speeds, yet here you both are, occupying the same present moment.