Reposted by PatVincent Carchidi @vcarchidi.bsky.social · 8hI feel like the big question hanging over the math stuff is: if OpenAI put their LLM on a few thousand problems, and it solved "only" a few hundred of them, would Navier-Stokes-esque brute force have solved the rest? Or is the model for some reason just not capable of solving all of them? 1212
Pat @apatday.bsky.social · 9hThink it came out okay. Not my best lining, but long overdue haircut 270
Pat @apatday.bsky.social · 10hCut my hair, shaved, and now I'm gonna spend a few hours getting a website setup (apatday.com) - yapping to come I guessapatday.comComing Soon 190
Pat @apatday.bsky.social · 11hThere's a contradiction here If the world is filled with complicated apparatus beyond any human comprehension, tech's bottleneck would by definition not be human understanding / time to reduce to human understanding 000
Pat @apatday.bsky.social · 19hYou can't just reframe racism & what that means as some hamfisted analogy equivalent "but for" some small detail. Thats nuts 1. One's racial identity is 24/7, not 9-5 & online hrs. 2. It's not a choice (which is 1 way the Supreme Court delineates classes of identity). 3. Tech bro is not visible. 020
Pat @apatday.bsky.social · 19himo if one's actions contradict stated beliefs, it's simply that there are unacknowledged beliefs. For example, I eat meat bc I am unconvinced by a vegan's moral reasoning. Trying veganism in earnest let me see something about my reason for being vegan was uncompelling. Know thyself (thru action) 040
Pat @apatday.bsky.social · 20hLean; a talk on what it means to train an AI on Possble Worlds youtu.be/p-z4bhj7p-g?...youtu.be[Berkeley Seminar] Benjamin Brast McKie | The Construction of Possible WorldsYouTube video by Topos Institute 010
Pat @apatday.bsky.social · 20hGot some interesting questions around neutron/tokamak challenges. In short, I don't have a solution for that - but there are options being validated. When plasma fusion reactions can be maintained for more than ~20min, the structural integrity of materials will become the most obvious challenge. 010
Pat @apatday.bsky.social · 21hElegance is an interesting parts of this convo fo sho Arguably elegance in math is related to how concisely and clearly a system of ideas can be stated. (Like e=mc^2) So if elegance is irrelevant, communication is irrelevant (or less relevant). But does it still matter for inter-agent comms? Idk 110
Pat @apatday.bsky.social · 22hOne thing is their actual code base - it isn't a formally verified codebase. And from people I know who work there, it has spaghetti code that needs to be untangled as a full time job. Similar to how OOP is barely mathematical (if at all depending on language) - an enterprise codebase isn't math 030
Pat @apatday.bsky.social · 22hI read this book too early in my journey, and had real trouble with the notation - which meant reading other essays & books (like on gentzen notation) But if I were doing this today I'd probably take a pic of the page & ask ai questions about each unknown. Wonder how that'd have changed things..? 000
Pat @apatday.bsky.social · 22hI have some research experience in neuro, and happen to mathy-program these days - total coincidence of topic :) 010
Pat @apatday.bsky.social · 22hShould we consider potassium channels and electrochem dynamics in the model? How can we norm across brains? What are the implications of variation across the population? That can change how we see morality if we didn't factor in frontal and limbic system differences - and instead said 'pari passu' 110
Pat @apatday.bsky.social · 22hSome of it also has to do w/error rates For example, if you have an attention study (eg social cues related) where two subjects have varying degrees of neural activation on some task...that can impact the kinds of decisions made due to bounded rationality, working memory retrieval, inhibition, etc. 110
Pat @apatday.bsky.social · 22hThere have been some experiments in this; and in some areas of cognitive neuroscience, one could argue the ontological foundations have been built. In reality, this would mean breaking down a large number of situations via formal methods, and making many ambiguous judgement calls 120
Reposted by Patꭓle Ormsby 𓆙 @e-infinity.space · 11/10/2026A colleague and I are developing a proposal to host open-weight models and supporting infrastructure at our institution. If others are pursuing something similar, or are curious about the project, I’d be glad to compare notes and share ideas. 1223
Pat @apatday.bsky.social · 10/10/2026Do you have any writing/thinking at the fund (or fund of fund) level on space economics? 010
Pat @apatday.bsky.social · 10/10/2026It seems heavily dependent on what we actually do, rather than some sort of predestination 010
Pat @apatday.bsky.social · 10/10/2026The idea being that if we have a class of solutions that fit classes of problems we encounter, thats a more fruitful direction for the field than looking for 1 philosopher's stone that opens every door. Instead we should find the domains that map to solutions of a particular form,& collecting those 000
Pat @apatday.bsky.social · 10/10/2026Appreciate the evidence based approach - I don't have a specific solution that exists today, but it's more a supposition about principle. (Also thinking more optimization than cryptography.) Basically we should find the domains that map to solutions of a particular form, and collect those domains. 100
Pat @apatday.bsky.social · 10/10/2026P# gets overlooked in these convos but those solutions have maybe more practical value than the larger p vs np convo. 110
Pat @apatday.bsky.social · 09/10/2026Any way to annotate the stuff that's being passed or is that asking way too much? Feel like thats got to be a ton of data 010
Pat @apatday.bsky.social · 09/10/2026Would be cool if these could be astrophysics valid - like really moving through the galaxy 120
Pat @apatday.bsky.social · 09/10/2026Since the start of democracy only ppl who had wealth enough to go and participate have been senators and representatives. Democracy and bourgeois democracy are a tautology 010
Pat @apatday.bsky.social · 09/10/2026We live in a totally different reality. Much more ambient violence 040
Pat @apatday.bsky.social · 07/10/2026The safety net, imo, should be tied more closely to asset appreciation. So im directionally same/similar On job change: that seems like a thing that's been happening with humans since we stopped being hunter gatherers. Being a philosopher looks really different today, than in 1410, or 200BC 020
Pat @apatday.bsky.social · 07/10/2026Gm Find myself thinking about an old unpublished book of algorithms w/a chapter on one called O1G3M (O(1) Galois-3 Matroids). Thinking maybe i clean up the chapter & submit to a journal as a paper. The book's focus is geometric/number theoretic algos in control systems. Have forgotten some proofs 010
Pat @apatday.bsky.social · 07/10/2026That's some major gc to do How'd you go about flattening the codebase process wise? 110
Pat @apatday.bsky.social · 07/10/2026I've been thinking about this issue lately - like how information is stored determines a lot about the processes and algorithms used when querying, unwrapping, unfolding, etc. If there's an algebraic data model used during storage it can change retrieval and search costs later. 120
Pat @apatday.bsky.social · 07/10/2026The issue seems to be that without disclosing their intractable problems it weakens our collective belief that there is even a future math profession. 010
Pat @apatday.bsky.social · 07/10/2026That's valid I run a Delaware C-Corp, and the Dutch have a program for transitioning existing orgs into the dutch ecosystem (mostly tech companies). You're probably right about the deposit -probably a shorter way to get this figured. Sort of on hold lately -but awesome you're on the adventure! 010
Pat @apatday.bsky.social · 07/10/2026Oh gotcha; americans can stay for the same amount of time without (<=90days), but for long term emigration it requires either DAFT, the startup visa, or edu I believe. Plus residency permit, which has been my barrier to figuring out a transition. Most won't rent to founders is my understanding 110
Pat @apatday.bsky.social · 07/10/2026I've been interested in doing this as well - did you go through the startup visa? 100
Reposted by Patbrennan @brennan.computer · 28/09/2026Unnamed Simcluster Podcast #11! A chat with @hikikomorphism.bsky.social about coining new phrases, the origins of solarpunk, predicting the singularity, and Astra tries to make some 3D models! also available on media.simcluster.stream youtu.be/xt-V9PdxY9cyoutu.beUnnamed Simcluster Podcast -- 011 hikikomorphismYouTube video by Brennan Letkeman 0438
Pat @apatday.bsky.social · 05/10/2026Completely agree - that's the worst part of human reasoning. So pernicious 010
Pat @apatday.bsky.social · 05/10/2026Though not conservative myself, I feel obliged to point out that conservatism has a place in the human story. It shouldn't be cast as simply evil and prejudiced. Sometimes we humans need to pace ourselves since we get startled with too much change too quickly. For example, the 100 years war. 110
Pat @apatday.bsky.social · 05/10/2026Sounds like the cost of living an interesting life, but hopefully worth it! 020
