Like a really old RPG where they just plop you on the overworld to get into random battles until you find the first town
Posts by Wild pentachoron correspondent
Post OT IX Cryptography
From the online supplement to their book
I think part of it is just that devotion to fomo/"skating to where the puck will be"/disruption has become more fully pervasive in tech leadership than ever before
Rationalists (or at least Yudkowsky & Soares) think that superintelligent AI would probably not be sentient
Keanu Reeves IS ChatGPT
Yeah I've thought of it as a lawyer for hire or something, it will aggressively prosecute whatever case you give it whether it's good/makes sense or not.
Reorganizing my entire life around avoiding ever seeing a definition of "frisson" in either English or French so I can keep on enjoying not being quite sure whether I really know what it means or not
if you can figure out a way to heavenban your opponents instead
and yes with the CLI one, you can review one change at a time, and hit escape and complain at it as soon as you see something you don't like, it will course correct
GHCopilot CLI is way better than the GHCopilot agent mode in vs/vscode for some reason. Logically it seems like they could be the same thing just hosted in different UI rectangles, but afaict they are totally different things
Yeah it clearly falls under the category of rituals and those aren't real in general
Yes, doing the same thing as efficiently with less code is what "good code" means
Oh it's just very silly
This idea has been in Yudkowsky-adjacent AI safety/etc discourse for a long time, usually under the name of "wireheading". iirc they have arguments for why wireheading would be just as much of a danger as paperclip maximizing, their arguments never made sense to me however
I still like open.substack.com/pub/recapitu... , besides that the content is interesting I'm impressed that he (1) has invented a new totalizing Manichean intellectual history lens - Plato-v.-Aristotle and 儒法闘争史観 having grown stale - & (2) managed to devise an acronym far more unwieldy than TESCREAL
iirc the trailer like cold opens onto him correcting the interviewer wrt the distinctions between different kinds of extinction
Yeah saw him there on the big screen before the movie when we saw The Bride
think the backlash against & decline in specifically consumer AI is interesting in itself?
Wait was the kid in White Lotus season 3 named after him
We can build better interfaces to language models than chat if we're willing to put in the design and engineering work. This post showcases a proof-of-concept implementation of one such interface: a semantic navigator for an arbitrary repository of files
haskellforall.com/2026/02/brow...
New blog post: "A sufficiently detailed spec is code"
I wrote this because I was tired of people claiming that the future of agentic coding is thoughtful specification work. As I show in the post, the reality devolves into slop pseudocode
haskellforall.com/2026/03/a-su...
always felt like lots of, e.g., cultural appropriation talk was always this, like it served as a kind of social capital protectionism or something
... and I end up either throwing it out or rewriting almost all of it myself and spending more time than if I'd just written it myself from the start.
It's occurred to me that maybe this just shows that I'm better at writing code and/or worse at everything else than I'd realized.
But whenever I've asked them to generate any substantial amount of code, beyond a few lines to demonstrate usage of some API I'm unfamiliar with, the results have always been terrible ...
Including everything else in software engineering around code (debugging, reviewing, reading code, brainstorming architecture, etc.) as well as discussing various questions in the humanities and sciences where I'm a dilettante.
Reading discussions online and comparing them to my own experience makes me feel crazy, because I've found today's LLMs and agents to be seemingly good at everything *except* writing code.