tsukikage

I’m finally doing the thing I always said one day I would, and writing a game! Early days yet, but I have proof of concept for much of the rendering tech now.

https://fluff.toothycat.net/@moonshadow

…this ALWAYS bothered me ever since I first saw Gremlins as a kid.

How early in the morning can I feed them? Dude never says!

Why the indirection? - why not have the jump table reference f0..fX directly rather than jumping to a jump?

...yup, looks like there's no way to get the optimisation benefit in "ignore" mode, which is the only mode that doesn't explicitly compile the runtime checks: https://godbolt.org/z/KTx1Pc3aa compare codegen for test3 (the pattern that would be useful) vs test4 (the best we actually have)

The article makes no mention of optimisation use cases. In case anyone already knows off the top of their heads and before I start digging for the answer, is part of the intent here for these to inform the optimiser stages, perhaps similarly to what the if (!(some_condition)) { __builtin_unreachable(); } idiom can achieve?

Between this, the social media stuff and whatever further lockdown attempts are in the pipeline, that's an incredible amount of incentive to find hacks and workarounds. We'll end up with a generation that actually gives two hoots about owning and controlling their hardware and media, and has the engineering skills to achieve that.

A somewhat roundabout way of accomplishing that goal, but I'll take it!

"One body, three behaviours"

Much of the text is generated as well.

If your benchmark score beats the current incumbent, you get to wipe and replace them. There can only be one!

It's just like cookie banners: they shouldn't exist, but people's bonuses rely on never admitting that, so here we all are.

> Nobody wants this

"Oh. Well, then stop."

Have you ever considered that "well, it works on MY machine!" is perhaps not the most helpful response to people voicing frustration at something that is not working as they expect for them?

Realise that you are talking to some mathematics in a box. You can't get a rise out of it. It cannot feel guilt or remorse. Whatever emotional payoff one might want from "I asked for ONE button" cannot be had here. Meanwhile, not only are you being charged by the word, but every word you send that is not directly on the path to getting what you want done is just noise in the maths getting in your way.

So leave emotion at the door and make your words count. Voice your frustrations at the stress doll next to your monitor, sure, but spending tokens on them us just wasting time and money. Explicitly typing out your exasperation will not get you closer to whatever it is you are trying to accomplish. What can you type that will?

Laser focus: clearly, concisely explain what is wrong right now, and how you want it solved. Then do that again until all the moles have been whacked. If the AI is stuck in a sycophancy loop, start a new clean session. Do this frequently anyway: every turn in the same session charges for all preceding conversation again /and/ degrades the LLM's performance.

Traditional forum/wiki spam is intended for search engine crawlers, not humans. How this new stuff relates to human users of the resource does not seem to differ in any way that is significant.

“You could have LLM, LLM, chips and LLM. There’s not much LLM in that.”

We don't understand what those are or how they come about, so how does that help us objectively reason about whether a particular black box does or does not have them, especially when it behaves as though it does, beyond arbitrary I-know-it-when-I-see-it "things I don't like don't have the property" style declarations?

Meanwhile, entirely too frequently I see unit tests that have atol / rtol values which permit errors of one part in 200 or even higher, where people bother with unit tests at all. At some point someone in the pytorch ecosystem used rtol=5e-03 in a unit test and crowds of people started passing that around unquestioned.

With attention matrix sizes being what they are, that's high enough that you can literally zero out a row or two and still have the test suite pass. Guess how I know!

Check your numbers, folks.

> Or are they just using it for training

You’re making a distinction here, but training is something you repeat for every new point release, so you need to keep the data if you want to use it for training.

> The origin of this sentiment is users entering a quiet IRC room

For me, the origin of this sentiment is the culture for certain departments at $LARGE_CORP I used to work for. People would ping me on the internal instant messaging tool to ask for help with things. Sometimes they'd take the nohello approach and just fold everything together in a single message: "Hello, hope you are having a nice day. I'm trying to do X, and I see Y instead of Z. What am I doing wrong?"

And sometimes, they... don't.

"Hello." <wait>

"How are you today?" <wait>

A conversation that could have been done in 30 seconds drags on for many minutes, and the message trickle feeds just fast enough that I can't do anything of any complexity in between the IM app's alert lighting up again. Extracting what they actually want from them feels like pulling teeth. It's a complete waste of their time and utterly disrepectful of mine.

They are there to ask a question. I know they are there to ask a question. They know they are there to ask a question. I'm happy to answer the question, when they finally get around to asking it. Just ask the bloody question, ffs.

Facebook already requires users to tell them their age, and already mandates minimum ages for users in their T&Cs. Recent and current lawsuits establish that /this is not enough/. Hence, mandatory age checks involving some combination of government ID and face recording, and a big database proving you've done that so you don't get fined when kids get on your platform.

We're literally about to require Meta to prohibit the creation of multiple accounts. Their T&Cs already prohibit that, so what we're saying is that just asking is not enough - we are demanding it is enforced. What we ask for, we get.

We should take a minute to understand that the current lawsuits are pushing, hard, for this. Whether or not Meta also want to do spying thing for their own purposes, we are forcing them to roll this shit out when we declare it's not enough for a user to tell a social media service their age, the social media site needs to be /certain/ and make a big database of their proof.

The world that results may be exactly what we want, IDK. It's just be nice if we could decide, collectively, one way or the other, then be deliberate about it.

You’re supposed to use an AI on your end to get the original few words back, so the providers get paid twice for the same email as is their right

Meta acquired Rivos last year.

He explains, at length: there is no sane way to determine what the hardware you are running on actually supports, and so there is no sane way to ship compiled code that is both compatible and performant.

We already had the mystery meat CPU wars several decades ago. We know how to make sane ISAs now and should be past that.

What is the sane default response to "we and our 242 partners value your privacy"?

I mean, it's even the literal truth: they value your privacy in the sense of having their software do a little internal auction to put a price on it.

Had no problems at all self-hosting for two decades; then a provider went bust and I lost my static IP block, and with it its two decades of reputation.

I still self-host, but now I cheat for deliverability: route the problematic destinations (I'm looking at you, Outlook) through Amazon SES. At a few cents for five-figure numbers of emails, it's as close to free for personal/vanity email quantities as it gets to have someone whose full time job it is to make sure outgoing mail stays deliverable on your side while still keeping everything else under your control.

One thing to note is that Amazon are not very good at keeping amazonses.com up to date, so despite including it you can't set DMARC to hard-fail SPF or you will have mail bounce days.

This will bite you.

It's how venture capital works: the theory is that you invest in a dozen pie-in-the-sky pipedreams, almost all of them fail, and the one that doesn't pays for the rest. If you're not failing most of the time, you're not taking on enough risk.

When first encountering these ecosystems and looking at the various pieces they contain, one may repeatedly ask: "why didn't they just use <off-the-shelf solution> for this problem instead of writing this component/subsystem from scratch"?

The answer is often that the battle-hardened mature off-the-shelf solution did not exist at the time the code was written. You're doing software archaeology.

For the love of all that is good, clean up the AI slop before you parade it in front of people. Can't you see the misspellings, misattributions and duplicated labels in the images? If you couldn't be bothered to spare it even a passing glance, why should I bother? I, too, can prompt an LLM; so what, exactly, is the point of you?

Note that those are total comp numbers. Something like 60% of that is dependent on performance review, which is fed in large part by automatically collected metrics like number of commits, number of code reviews ticked off, the LLM token usage leaderboards etc; creating all the perverse incentives you'd expect.

So you can rely on income to precisely the extent you are willing to destroy your soul.