faun @faun@merveilles.town

I'm sort of obsessed with non-destructive construction aesthetic. No nails, no glues, everything connected with bolts and clamps and joinery, no wire ever laid in a place you wont be able to reach later, nothing permanent, everything renegotiable.

Follow these principles and experimentation will have no material costs.

I don't know how practical it is, but I was raised on Lego and Data and an edict of Reduce, Reuse, Recycle, and so I do not know if I can live well any other way, I just need it

Make things that change people

Now trying to generate realistic models of endorsement networks for performance testing. Surprisingly deep work. The agent models should be simple because the finished system's UX should be simple. I have no excuses now, it is time to face all of the fundamental questions about I expect this to grow. What sorts of people issue invites, how people discovery other people, what incentives a popular curator has to endorse another, how networks improve over time or birth themselves from nothing.

#computer #tasteweb
Graph DB is mostly written. Most of the concurrency work will end up being done by the "left-right" crate, which does the thing I was going to do where there's a read copy and a write copy and every now and then it swaps them, only it turns out I didn't really know how to make that lock-free, but I do now. This is going to be *the fastest* (eventually consistent, for work with a high read:write ratio, single-machine) graphdb

I'm pretty sure it would be "good", given what I can see from here.

A field research engineer who would be subjected to a series of challenges.

Made for a prototype concept of a game about manipulating the structural features of an ancient complex that I have more or less finished getting out of my head for now

Hey @cancel, do you ever think about using your familiarity with Qt to make a Paradise MUD client (with a chat room in every vessel), or a browser for one of those minimal web protocols like Gemini (or a better one that you designed).

I've been thinking about doing something like that with Flutter, but 1) Flutter isn't quite baked yet 2) Flutter apps often runs worse than web pages on my desktop 🤢 maybe that will improve but right now the only nice new apps I'm seeing are Qt

Every time I look at my peacewager (game where peace is possible but difficult), the small number of players, their personal but fundamental conflict, the landscape marred by terrible holes, reminds me intensely of Ridley Scott's Raised by Wolves and damn I need season 2 right now.

(Was the game inspired by raised by wolves?? I don't *think* it was, although IIRC I don't think the hole biomes came along until after I'd watched it so maybe a bit o--o)

@csepp plz invite me to snikket

Kirsten Dirksen's family home is just what you might imagine it to be after seeing the kinds of homes she likes to document: A near perfect optimum of grounded, flexible simplicity https://www.youtube.com/watch?v=9jnEJDt9Sjc

I wasn't interested in neural networks until I heard this interview. Now I'm *really* interested. I'm considering applying. If the interest hasn't died off in a month I probably will reach out and at least describe my view of what it is the visualization spec has to create, I'm not sure it's exactly what they think it is, it is not like most visualization jobs

Anthropic are looking for a data visualization specialist, the one who will make the lamps and lenses by which these murky caves of wonder could for the first time be seen https://jobs.lever.co/Anthropic/056ad13a-2b0d-4482-ab63-dcdbe1c03aac?fbclid=IwAR3s27SML2F-TPZwDeBw0YBIFXNs-fRh56On19xY6ZxufTIrZznmuRsHlZ8

Anthropic is a new AI alignment research organization, a large part of their approach is the development of interpretability: trying to understand how neural nets structure themselves. Frankly, it's responsible as fuck, and, hearing them talk about it, fascinating https://80000hours.org/podcast/episodes/chris-olah-interpretability-research/

#theCinema I watched the new netflix Ghost in The Shell 2045 Full 3d CGI. It might be okay. You might want to see how the second season reviews before going in.

Visuals: *Interesting*. Not "last-gen" imo, *Retro*. Plot: Decent, poignant, the post-humans (whose vice seems to be extreme levels of self-actualization) are a mood, but could turn out to be really offensively goofy depending on where all this is going. Tachikomas: Not as heart achingly precious as in SAC but still pretty precious

@body hanson's theory of xenosociology https://pca.st/podcast/417fbfb0-df4e-0133-2ec0-6dc413d6d41d

Includes Boulder lore: sprites were first photographed there

That probably didn't take all that many resources to make though.

The other listed projects are search engines for various encyclopedias (stuff like Britannica, stanford philosophy, "citizendium" (which seems like a smaller chiller less persnickety wikipedia, cosy)) and seem nice enough.

Looking at one of their projects... appears to be microblogging but RSS and with no follows or retweets. Very little technical merit, considering that apparently mastodon already supports RSS https://mastodon.social/@brownpau/100523448408374430

What a strange thing to fund. And Larry mostly only uses it for tweeting conservative culture war takes. And its reply view is.. actually kind of worse than mastodon's (only shows one level at a time. I guess it's debatable.)

You can tell that the project is still active because it says "Yes, it’s active. There’s stuff going on behind the scenes." in the faq. Seriously though it probably is, it was only launched in 2019.

@jim have you heard of this

I would ask them for funding to build a fast web of trust query server if I has any idea what I would do with funding