klodolph

Maybe it just reached its peak with “just according to keikaku (translator’s note: keikaku means plan)”

You can figure it out from context if don't know one or two words in a paragraph. If you are missing a word from most lines of dialogue then you are probably going to struggle.

People who can figure out words from radicals probably have >500 word vocabularies to begin with. I’d say if you started learning on some natural progression, you would start learning how to figure things out from radicals once you knew somewhere around 300 kanji (which would mean you have a vocabulary of 1,500 words, ish).

Before that point, you are probably going to have too much trouble generalizing.

It’s just a choice of which key to use. I don’t think it was some decision made with this intent, I think it is just the unexpected result of a decision.

The KHR_debug is also kind of a pain to use because it gets called asynchronously (which is what you want, yes, but the difficulty is still there).

Yeah, I've migrated away from tmux and back to using terminal windows.

Interesting. I’ve been using it heavily on macOS (ssh to Linux host) and over the web using xterm.js, and it’s been working about as reliably as tmux did. Sometimes it doesn’t resize properly, but that’s also true of tmux (when I see a big field of .).

I’m aware of “normalization of deviance”, but I think of it as a problem, and I should solve it, work around it, or decide that the status quo is fine.

Like, if I got shouted at for saying what the previous five consultants had said, I would take that as a lesson.

I've been using shpool these days (especially for things like Claude / Codex / etc). Shpool only does persistence, not multiplexing. It turns out this is exactly what I want. I want to have one keyboard shortcut, to detach the session, and that's it.

I can do terminal multiplexing better in the GUI—tabs, virtual desktops, or in the browser with xterm.js.

Yes, you can learn all these different tmux shortcuts, and I know them well enough. I like using the shortcuts in my terminal emulator better. All I really want is a way to continue my session after putting my laptop to sleep, and shpool does exactly that and no more.

Doesn’t this line of debate get kind of predictable?

I didn’t see your reply.

It should be the harness responsibility to create / destroy things like worktrees and scratch spaces, and communicate to the agent (e.g. via system prompt) where the scratch space is and its purge schedule. Harness should also set appropriate env vars to make this work in tools.

> With the same behavior showing up in multiple high-profile harnesses (Codex, Claude Code), it feels likely there must be some actual reason for doing it this way.

I think fixing it is just not high on the priority list.

Harnesses should be responsible IMO for managing lifecycle stuff, including worktrees and temp files.

It’s simple, deterministic stuff. At least most of the time.

Over eager use of /tmp is the worst. That’s RAM, my boy! Precious gold.

It was a literal factbook online. In high school, I remember using it to look up facts like the area or population of different countries.

from what I understand, there’s a high standard here, beyond “beyond reasonable doubt”, along the lines of “no appearance of impropriety”

Technically true but not informative

Finder has its whole set of keyboard navigation shortcuts, like command + up arrow to go to parent, command + down arrow to open, ⌘3 for column view, ⌘⇧D for desktop, ⌘⇧H for home, ⌘⇧G for "go to"… I am very fond of the Finder’s navigation and as a long-time Mac user it still feels pretty clunky to try and get around on Windows. Both Windows and Mac are getting worse but it seems like Microsoft is making it worse faster, with OneDrive.

A lot of the Finder shortcuts work in the save / open dialogs in apps, although they broke some of the ergonomics of those dialogs years back.

Special relativity is one of those unusual theories that requires very little in the way of math or concepts to understand. You take the data from the 1887 Michelson-Morley experiment (which measures the same speed of light, despite different reference frames). You take the idea that the laws of physics work in any reference frame. You take some high-school math, et voilà, special relativity.

What I found when I searched for "on the fly" is this: https://github.com/mthom/on-the-fly-gc — it looks like it has pauses, from the docs.

In general when you tune knobs for GC, you pay for benefits in one area with sacrifices in another. Two big knobs to turn are pause latency and throughput. You probably wouldn’t want to go full “optimize for latency” because you’d end up with poor throughput. Also vice versa. Java’s reputation for poor GC performance is partly due to historical defaults that tune it for throughput.

Go’s GC is already a “concurrent mark-sweep garbage collector” and already has “extremely low mutator pause times, on the order of tens of microseconds”. It sounds like on-the-fly is just a different flavor of what Go already has.

100% agree. Most of the things that make development better for humans also make development better for agents… and I think docs are even more important with agents, because of some kind of multiplicative effect. The agents are coding faster, and the benefits of documentation are somewhat more pronounced because of the speed.

You aren’t promised your money back with interest; that’s what an IRA is. Social Security is a different system which works by different principles. In Social Security, the money you pay goes to the people who are retired, and your retirement is funded by people who work in the future.

Your money --> older generation

Younger generation ---> you

The main thing that makes my description incorrect / incomplete is that baby boomers, such a large cohort, broke the system. This was predicted and so the baby boomers actually did get their money back: they were taxed to fund OASI, which then goes back to fund baby boomers’ retirements, until it runs out.

Baby boomers (only, sort of) <---> massive cushion in the OASI fund

Social Security is funded by the people paying in, plus the money in funds like OASI. Yes, OASI is treasuries, so it’s connected, but the reason OASI exists is to cover the changing demographics as baby boomers retire. It’s not designed to be, long-term, the way the Social Security system works. The idea is that you collect income on the boomers greater than payouts while their cohort is in the workforce, and the pay that money back out when they retire.

Yes, there are reasons why this isn’t working out as planned. But the idea that “Social Security is the biggest driver of spending” is not a cogent argument.

Currently reading: Tracks (re-read) by Erdrich. Just finished: The Bad Beginning by Daniel Handler. Before that, Lolita by Nabokov.

I have a long-ish commute at the moment with some opportunities to read on the train, so I’ve been reading more. Been knocking out some of the books on my shelf.

A stain is a kind of mark :)

Is this happening at any significant rate?

AFAICT this is a fear that some game developers have, that someone will steal their game and run with it, and the stories told are more repeated based on this shared fear than based on realistic scenarios. People repeat it because it’s a good story.

And yes, I’m aware of some situations like 0x10c and the like.

Ok, in order to make money? Let people make some money making shitty copies of my game. I think there are roughly two scenarios here.

Scenario one: they’re competing with my game and stealing my thunder. In this scenario my game wasn’t very good to begin with.

Scenario two: we’re going after different customers. They get the customers who want a cheaper game, I’m getting the customers who want a better game.

I’m sure somebody out there is pointing an LLM at things and getting a $200 average payout or better on $100 in tokens, but if your game is getting copied for $100 in tokens, I’m not sold on the idea that you were going to make money in the first place.

What I think of is all the HD remasters out there which look disastrously worse than the original, at least in my eyes. Why does this happen? Are the best artists working on new games instead? Are remasters pushed out with less care, shorter schedules, and less budget? Maybe some combination… and maybe there are some parallels with quick, mostly unsupervised LLM copies of a game.

From what I’ve seen, most of the flood of LLM-created games aren’t profitable, just like most of the games people make in devlogs. There’s very little thunder to be stolen in the first place.

Ok, so let’s get some data points. Are there games being recreated from devlogs? I’d like to see what the recreations look like.