atq2119

> Untangling the entire lot to make the game available upon closure would be a nightmare in some cases.

This is only true if the game wasn't architected with open-sourcing in mind. Which affects how this kind of law should be structured.

For example, it makes sense to require the server code to be submitted to an escrow service from the beginning. Part of that process would be a license evaluation, which acts as a forcing function already during the development process.

There's the adage that writing is thinking, but even more accurately at least for me, editing is thinking.

Neither typing speed nor dictation speed is a true bottleneck, but editing speech seems like it'd be harder than editing text.

Though there may be some hybrid approach that can work well.

> Teachers were never highly paid.

Teachers never got rich, that's fair, but there was a time, at least in my culture, when teachers were usually one of the most highly regarded and respected people around, on a similar level as priests and doctors.

> The idea is also that parents with lower intelligence will value education for their offspring less and care less about whether their kids learn to read well, etc.

Educated people can neglect their kids. And less educated people can still recognize when education is valued by society.

Now look around. Do you feel like we're living in a society that values education? Did the successful people that kids see in their formative years get there through education, and/or do they visibly value education beyond lip service?

Some of them, yes. But I'd argue that between influencers, teachers' pay, and the increasingly obvious nepotism and corruption by people in power, the situation is looking pretty dire. We don't truly value education as a society and are therefore teaching a new generation that education isn't to be valued. And that has nothing to do with genetics.

Yeah, Trump showed the world exactly who he is even before his first term. He has absolutely no principles other than feeding his own wealth and ego. Anybody who is surprised by what has happened is either an incredibly bad judge of character or hasn't been paying attention.

Plus a systematic way of keeping the Gini coefficient of wealth small in a sustainable way. I'm a fan of establishing sovereign wealth funds whose dividends are paid out equally per capita for this purpose.

Gamedev dev is a good example actually. True, handwritten assembly has gone out of style. But knowing how caches work, and how to lay out data to improve performance is important. And stuff like vector intrinsics also gets used.

Yes, that is indeed O(N^2). Which, by the way, is not exponential.

Also by the way, caching does not make LLM inference linear. It's still quadratic, but the constant in front of the quadratic term becomes a lot smaller.

Permanent public ownership of (very large stakes in) these companies doesn't seem like such a bad idea anymore, does it? It's what we used to have for most of the 20th century at least in Europe.

It does make sense. Police are absolutely not beyond reproach, and there's screwups all the time. They need to be held to a high standard.

It's also easy to imagine reasonable compromises, like a time delay where they only have to report after e.g. 48 hours, and allowing a system whereby a judge can issue a warrant to extend that delay.

People have been building stacked PR workflows on top of GitHub for a while now. It's great to see that the message seems to have finally landed at GitHub, but what is actually new here in GitHub itself (i.e., not counting the gh CLI tool)?

There seems to be a native stack navigation widget on the PR page, which is certainly a welcome addition.

The most important question though is whether they finally fixed or are going to fix the issues that prevent submitting stacked PRs from forks. I don't see any indication about that on the linked page.

It's a fairly weak level of control, though.

Compare with the governing structures of public universities in (most of?) Germany where there is a "senate" composed of elected representatives of professors, students, and administrative and academic staff. Now that is approaching democratic control.

It's sort of explained in the article, actually.

Upscaling and related applications are already supported by cooperate matrix ops. Those shaders/kernels run in workgroups that operate on a tile of pixels at a time, so you naturally get matrix-matrix operations. And yes, at least in Vulkan, there has been cross-vendor support for those for a while now.

Neural materials and similar applications often operate on a single datum at a time, like in a pixel shader or a raytracing shader.

Yes, the hardware still groups the threads of a pixel shader together, but there is inherently more divergence. And so the natural program representation tends to have matrix-vector operations instead of matrix-matrix operations.

That's where cooperative vectors come in.

The problem is in your premise: Software usually does not end up finished.

> if Hitler had grok, would it really get any worse for the Jew

Not grok specifically, but yes.

The holocaust in the Netherlands was remarkably bad in large part because the Dutch administration was so well-organized and had kept a registry of Jews.

Bad guys are going to use this technology to evil ends if given the chance.

BTW, there's a chilling alternate history novel called NSA by German author Andreas Eschbach about precisely that kind of idea. The premise is that computer science progressed a lot more quickly. The book opens with German data scientists in the 1930s combining census and financial transaction data (i.e., food purchases using electronic cash) to identify households that are hiding Jews and other "undesirables" .

AFAIUI, elections in the US are not run by the federal government but by the states.

Trying to cancel elections seems like it'd be a lot harder than in other countries.

I find it interesting that the S-curve is much steeper for AMD than it is for the others. AMD maintains perfect prediction for much larger sizes than the others, but it also reaches essentially random behaviour earlier.

Are they really keeping a branch history that's 30k deep? Or is there some kind of hashing going on, and AMD's hash just happens to be more attuned to the PRNG used here?

Would be interesting to see how robust these results are against the choice of PRNG and seed.

These providers do not run models on CPUs, x86 vs. Arm is irrelevant.

> DEI -- along with hiring quotas -- tended to bring new "officials" at companies and government orgs ("head of diversity") which is another great way of "creating" more elites.

If anything, it's a way of placating existing elites.

The elite overproduction idea is that there is a surplus of people who feel that they should have an elite status compared to reasonably available elite positions.

Creating additional managerial positions is a way to attempt to absorb this situation.

Looks like it, yes. It's encouraging given that so many discussions of these topics online are wrong. The explanation of constraints on bank lending in particular is something many people should read.

Comment threads certainly work better through email than on GitHub PRs, at least when you can use a good email client (i.e., running locally, and not Outlook).

The challenge is integration with CI and other automatic work flows.

There may be an is/ought confusion in your exchange.

It is probably true that California has no such law today. It's also true that regulation always takes a while to catch up to technological advances, and so there is a useful, separate conversation to be had about whether California (and anywhere else) ought to have such a law.

Don't forget the literal deus ex machina in the form of the computer that coordinates everything.

If you read it carefully, you'll notice that the blog post misrepresents the AMD response.

The blog post title is "AMD won't fix", but the actual response that is quoted in the post doesn't actually say that! It doesn't say anything about will or won't fix, it just says "out of scope", and it's pretty reasonable to interpret this as "out of scope for receiving a bug bounty".

It's pretty careless wording on the part of whoever wrote the response and just invites this kind of PR disaster, but on the substance of the vulnerability it doesn't suggest a problem.

The challenge is that this doesn't really work for community-developed software.

Let's say somebody uses this scheme for software they wrote. Would anybody else ever contribute significantly if the original author would benefit financially but they wouldn't?

Mediating the financial benefits through a non-profit might help, but (1) there's still a trust problem: who controls the non-profit? and (2) that's a lot of overhead to set up when starting out for a piece of software that may or may not become relevant.

It does. And it works best if you elect politicians who are willing to listen.

Was that Jane Street? I remember watching a presentation from someone there about such a system.

If not, any chance this tooling is openly available?

In the large, ideas can have a massive influence on what happens. This inevitability that you're expressing is itself one of those ideas.

Shifts of dominant ideas can only come about through discussions. And sure, individuals can't control what happens. That's unrealistic in a world of billions. But each of us is invariably putting a little but of pressure in some direction. Ironically, you are doing that with your comment even while expressing the supposed futility of it. And overall, all these little pressures do add up.

Having used Claude Code in anger for a while now, I agree that given the state of these agents, we can't stop writing code by hand. They're just not good enough.

But that also doesn't mean they're useless. Giving comparatively tedious background tasks to the agents that I check in on once or twice an hour does feel genuinely useful to me today.

There's a balance to be found that's probably going to shift slowly over time.

My personal reason for switching some years ago was the excellent remote session support via ssh.

I haven't reevaluated that choice in a while, but that plus LSP support (and to a lesser extent ML Auto-complete) are must-haves for me nowadays.