@Ngnghm

The Swiftian World of Houyhnhnm Computing ("Hunam"). Call me "Ann"—🐎see my blog! More cooperation, less coordination. Alter ego @fare (pol) @phanaero (crypto)

 Lair of the French Resistance
ngnghm.github.io/
Joined August 2015
  • Tweets10.9K
  • Following327
  • Followers1.9K
75Media
Lambda, the Ultimate Object: Object Orientation Elucidated fare.tunes.org/files/cs/poof… The ultimate book on the semantics of OO, or Internal Modular Extensibility. OO kernel in two lines of FP. WIP.
5
17
3
71
9,485
💻🐴Ngnghm retweeted
David Barbour@awelonblue10h
Replying to @jameskjx @Ngnghm
That's the same property that makes pure FP great for macros: it's great for calculating programs. Which happens to be the domain macros are trying to address. Indeed, pure FP is arguably much better at calculating programs than at running them; it doesn't align with hardware.
1
2
79
Oh noes, the European Lisp Symposium 2026 videos seem to have gone private? What happened?
4
2
17
709
Unfriendly Persistence: all the things you care about don't persist, the things your enemies care about persist. Make sure to backup all the Youtube videos you care about.
14
78
Hans Huebnner's Genera on Bare Metal: genera.vaxbusters.org
1
5
14
559
Never underestimate the stupidity of intelligent people—they always find more intelligent ways to be stupid. I would know that very well, if only I weren't so damn intelligent.
2
21
412
💻🐴Ngnghm retweeted
💻🐴Ngnghm
@NgnghmOct 20
Replying to @Ngnghm @msimoni @jonathoda
From my paper: "Schema Upgrade requires high-level language support, or else users may find the hard way that they can't keep their data without having to badly reinvent all of the mechanisms of manual persistence just for the sake of upgrading their Schema."
1
4
9
257
Cool: the earliest surviving contemporary written evidence I could find of the term “object-oriented” being used at PARC archive.computerhistory.org/… There must have been much earlier traces, but they did not survive, or were not indexed well enough for me to find so far.
1
4
18
396
While writing ch10 on implementing OO, I find myself writing a lot about general-purpose semantic-driven software engineering—because that's what most of OO implementation actually is, after you understand the semantics. I can't wait to actually get to the MOP, where OO matters.
8
258
💻🐴Ngnghm retweeted
💻🐴Ngnghm
@NgnghmOct 20
Replying to @Ngnghm @drorlb @jonathoda
So far as I can tell, the term "Durable Execution" in its modern meaning can be traced to a 2013 patent from Microsoft, by Abbas, Krishnaprasad & Dar—same Abbas who later co-founded temporal. So Microsoft had the technology and shelved it. Of course, EUMEL had it in ~1980.
3
2
10
332
💻🐴Ngnghm retweeted
David Sancho
@davesnxOct 19
I was interviewing someone with 4 years of OCaml experience, and said: "The language solved problems I didn't know I had, but the tooling invented problems I didn't know were possible"
20
54
9
1,111
26,881
💻🐴Ngnghm retweeted
💻🐴Ngnghm
@NgnghmOct 19
Replying to @Ngnghm @jameskjx
Those who don't study macros are doomed to repeat a lot of code that follows "design patterns" (or fails to). Yet those who *do* study macros are doomed to stand by helplessly while everyone else repeats a lot of code that follows "design patterns" (or fails to).
1
2
8
169
Chapter 10 is really hard to write... and keeps going in directions I don't anticipate (and the AI only less so). Yet I'm only scratching the surface... well the surface is all "philosophy" and no code (but lots of citations). On the other hand, the AI do be fixing LaTeX issues.
1
267
LaTeX issues: are you hyperlinks "eating" footnotes, footers and headers next page with interrupted by a page break? Well, it's fixed in lualatex—if you save some definitions from \DocumentMetadata{} before Scribble garbles them.
1
162
420... it's a good page count to pause at.
85
Line-art slop allelomeme of Zach Beane's famous meme—suitable for printing in my book?
🤖 Made with AI
2
1
6
333
Looks like the book is getting illuslopstrations!
67
💻🐴Ngnghm retweeted
🐾Eneasz Brodski
@EneaszWritesOct 5
the "Hard Problem" in a nutshell: 1. atoms aren't conscious 2. humans are made of atoms 3. humans are conscious 4. so at least one of 1-3 must be false Panpsychists say #1 is false Dualists say #2 is false Illusionists say #3 is false Materialists say #4 is false
1,160
564
425
9,467
1,957,204
💻🐴Ngnghm retweeted
Eric S. Raymond
@esrtweetOct 16
This is correct. It's one of the many obvious problems with AI doomer theory. Quite parallel with the concept of revealed preferences in economics. In fact, now that I think about it, the impossibility of pre-constructing an "alignment" function is formally identical to the impossibility of pre-constructing a preference function. TIL
Jon Stokes
@jon_stokesOct 14
Replying to @jon_stokes
I think the simpler answer than this is that alignment is impossible because a utility function is a hindsight construction that an observer constructs about an agent, and not an actual thing that exists inside an agent that can be tweaked or copied or "aligned."
6
11
2
131
5,875
After meeting maxim from @temporalio yesterday, I made clean ups to my essay "Orthogonal Persistence the Model" so he may better enjoy it—and so may you. git.cons.io/mighty-gerbils/g…
180
My Newcomb predictor gives money to all Newcomb predictors that don't give themselves money. Also, anyone who takes Newcomb's Paradox seriously has their head so deep inside their own ass that all I can tell them is "go unfuck yourself".
167
Load more