kqr

25 ms·

541.
by kqr 1y ago
Yeah, but it also takes very few commands from the player to get from the cyberpunk opening to You ride east, the coastal cliffs of the Grey Havens and the figures of wood elves giving way to rolling green hills. The familiar scent of pipe
542.
by kqr 1y ago
LLMs are not as great for text adventures as they might seem. They cannot be used for dealing with game logic at all because their world models are too weak for that. They could maybe be used as glorified parsers (i.e. read human intent and
543.
by kqr 1y ago
I think the mistake is looking at Emacs as an editor. It's an Emacs Lisp VM, and the editor that comes bundled with it is good, but not great. Fortunately there are many ways to improve it, and make it go beyond what other editors can.
544.
by kqr 1y ago
Shere are so many levels this could be answered at. All light in a narrow cone extending from the camera gets recorded to one pixel, entirely independently from other pixels. There's no reason this would be blurry. Blur is an artifact
545.
by kqr 1y ago
And the reason it matters that this is a single pixel at two billion times per second is that we can hypothetically stack many of these assemblies on top of each other and get video of a single event that is not repeatable.
546.
by kqr 1y ago
It would look something like this[1] except with slower visual propagation. Note that this camera (like any camera) cannot observe photons as they pass through the slits -- it can only record photons once they've bounced off the main p
547.
by kqr 1y ago
I was convinced it was about trees, just not ... that kind of tree!
548.
by kqr 1y ago
Some of the images (at least one I saw of a tree) had section drawings both from the side and the top, so no!
549.
by kqr 1y ago
I did the same. I eyed it suspiciously and thought to myself, "This site is really well designed. Do you think if I..." and it did switch! The obvious interaction was implemented! Rare these days.
550.
by kqr 1y ago
At first I was confused by this because the Kármán line is less than a percent of Earth's circumference up, but then I realised we're probably talking a geostationary anchor or something, which is very nearly a circumference up.
551.
by kqr 1y ago
I would argue an interpreter that needs 1,000 lines for its core loop is probably a complex piece of software, comparable to other 1000-line projects I've made.
552.
by kqr 1y ago
> These special surfaces were called “overlays” because they appeared to overlay the desktop. I have some vague memory of programs whose windows had funky shapes (i.e. not rectangular) also using overlays of some kind. Maybe that's
553.
by kqr 1y ago
I forgot to say why it's a problem LLMs invent details: which details are implemented and which are left out is an important part of how the implementer of a text adventure communicates to the player what sort of interaction they shoul
554.
Haiku 4.5 Playing Text Adventures(entropicthoughts.com)
1 points by kqr 1y ago|0 comments
555.
by kqr 1y ago
It doesn't work as well as you think it would. There are no AIs today that model the world well enough to drive a text adventure. They can be used to embellish a more primitive world model, but then they tend to invent details that don
556.
by kqr 1y ago
Going by the sample rooms above, (27, 21, 24, 28, 27) words multiplied by 4000 rooms divided by 350–450 words per page yields 200–300 pages indeed. But if it's mainly exposition and not interactable I can see how it would become dull r
557.
by kqr 1y ago
It was the same for me, except I grew up slightly later and did it with HTML documents and hyperlinks! (And in one case, I even think I got some help with an imagemap. Clicking the little hole in the foundation to the house led to an easter
558.
by kqr 1y ago
That's what I also don't quite get. I would build it as the parsing step mapped directly over the query. I suppose their approach lets them build a tree of operations but I'm skeptical it provides all that much benefit.
559.
by kqr 1y ago
This is surprisingly competent. A couple of months ago I evaluated some leading models on a bunch of text adventures[1]. Typical regression coefficients would be +0.02 for top level models like Sonnet and Gemini 2.5 Pro, but notably also Ge
560.
by kqr 1y ago
If you're a great editor, why do you let multiple LLMs edit for you?
561.
by kqr 1y ago
The important part for me is that the experience is legitimate, and secondarily that it's well written. The problem for me with LLM-written texts are that they're rarely very well written, and sometimes unauthentic. If we had real
562.
by kqr 1y ago
This is what frustrated me when I worked in e-commerce. They are blessed with the highest-quality signal ever – the purchase – but it's generated very rarely for anyone who is not Alibaba or Amazon.
563.
by kqr 1y ago
> Root cause works better if you can come back next time the same thing happens and find a different root cause to fix. keep repeating until the problem doesn't happen enough to care anymore. This sounds like continuously firefighti
564.
by kqr 1y ago
> you really need as much breadth as you can get Sure, more is always better. Practically, though, we are trading depth for breadth. In my experience, many problems that look dissimilar after a shallow analysis turn out to be caused by t
565.
by kqr 1y ago
> mental model It should be more clear in the article that this term is used more broadly. The "mental model" is the function that converts feedback into control actions. In that sense, even simple automated controllers have &q
566.
by kqr 1y ago
This is the hidden cost of heroics[1]: it severs an important information channel. [1]: https://entropicthoughts.com/hidden-cost-of-heroics
567.
by kqr 1y ago
> “you manage risk, but risk cannot be managed.” Ie. there is no terminal state where it’s been “solved.” I often hear people misinterpret "risk management" as if it means "risk minimisation", but this is the first ti
568.
Why Nix Will Win (and What's Stopping It)(ryanrasti.com)
12 points by kqr 1y ago|1 comments
569.
by kqr 1y ago
Author here. Please note this is an early draft/stream-of-consciousness. Feel free to read and share anyway but my actual published articles hold a higher standard!
570.
by kqr 1y ago
His next big book, Noise, is possibly even better.

More ›