groby_b

What do you think happens to data center water? (It either evaporates or is recycled)

That page is a lot of ink for very little data. It's also somewhat wrong - no, you can't claim the axis marks and labels aren't data ink at all just because it doesn't directly reflect the measure. They are still data that's extremely relevant to reading the chart. Yes, even if Tufte says otherwise.

Meanwhile, even Tufte's advice was explicitly to grey them down, but not completely remove them.

(And we can debate Tufte's advice - lots of other comments on this page how he had great aesthetic sensibility, but his advice wasn't grounded in reason or actual readability research, and so is limited to the examples he chose)

I gave up throwing parties after the first two years in the States (1998-2000)

Americans throughout treat RSVPs as meaningless, show up if they feel like it, and think it's fine to cancel plans five minutes before the scheduled start.

Unless you're a glutton for punishment, you take the hint at some point.

Commonly held lore is that you do not install a macOS with a minor version of zero. That has been confirmed by shoddy first releases ever since there's a macOS X.

Wait for the next upgrade unless you're forced to go with the first release.

OK, but what does it do, without having to look at pages and pages of eye-searing graphics in the default AI style?

The one question you have to answer on your landing page, especially with AI making landing pages easy, is, "why should I care". And deliver that answer as quickly and clearly as possible.

It is Quanta, the home of sensationalist oversimplifcations. What did you expect, solid science?

> Also obligatory: Schmidhuber talked about this a long time ago.

That is basically axiomatic ;) (Seriously, Schmidhuber's body of research is mind-blowing)

A completely open uploader without any restrictions?

Will see CSAM in 3... 2... 1...

That's of course rather nonsensical.

In a "one LLM only" environment, your instructions are by default tuned for said LLM.

In a multi-LLM environment, roughly nobody will keep separate sets of instructions for each. It's not a realistic take.

On top of that: If your LLM is so bad at reading that it can't follow a set of instructions that wasn't specifically written just for that one single precious LLM, I sure wonder what that says about your employers repeated statements that ASI is definitely right around the corner.

The existence of attractors (e.g. "stable at iteration 31") for AI models is one of the weirdest things in AI land. They are a wonderful space to explore, fascinating and full of horrors.

Meanwhile, my hat's off to the (currently) last comment on the page: "I like how the shear-sheathed legs still earned their keep — probably load-bearing for Mrs. Regan"

Flip side? I have, in almost half a century of tech dabbling, never once needed a cable from that box that I couldn't have bought again.

I still keep _some_. A basic supply of standard cables "just in case", and clearly labeled things for "who the fuck needed to invent another connector" cases. (Looking at you, printer manufactures and WD hard drive enclosures)

But I strongly suggest tossing most of it. You need the space more than you need more cables.

The details on the lab are (understandably and simultaneously frustratingly) quite sparse.

But it's worth keeping in mind that theoretically, the mold on that cucumber in your fridge constitutes some sort of biolab. The main released criteria about those lab that raised alarm was the presence of specimens, which does not imply any capability of creating a weaponized strain.

It's still alarming, but (at least in my opinion), still doesn't prove an effective bioweapon is buildable in a small-scale operation.

Georgetown has a good article on the general problems with our current "AI and bioweapons" dialog: https://gjia.georgetown.edu/science-technology/rethinking-th...

Without a whole lot of tacit knowledge no LLM can provide you, bioweapons are still pretty much out of range for most of the population. Doesn't mean we should ignore the problem, but a more reasoned approach would stand to benefit everybody.

"indepdendent" does not mean "i see all parties as absolutely the same".

It means "I am not firmly affiliated". You still have the ability to make choices on who is a better likely outcome

Because it's completely dwarfed by florpedness. I can't define that either, but I just know it's more important.

Less facetiously: A debate of the importance of something needs a shared understanding of what is being debated. Without that, any discussion is merely people shouting that their belief is the right one, and the others are the heathens/idiots - because there isn't even agreement on what is debated.

That's the famous dropbox comment right there.

> 1. For a Linux user, you can already build such a system yourself quite trivially by getting an FTP account, mounting it locally with curlftpfs, and then using SVN or CVS on the mounted filesystem. > 2. It doesn't actually replace a USB drive... This does not solve the connectivity issue. > 3. It does not seem very 'viral' or income-generating."

Normal users don't care about running an extra mini-PC. If you do shoulder it, it becomes part of your attack surface as well, so it's non-trivial work.

And sure, you can find content anywhere - ordinary people won't.

Except... some of us like to stream content on TVs. And some of us really don't like the idea of maintaining yet another extra piece of hardware for that.

And so your TV does need Internet access.

The same holds for many other devices, in different shapes. Sure, you can opt out of civilization completely to avoid all the side effects of late stage capitalism, but maybe it's worth applying a bit more nuanced thinking to find a balance that doesn't throw you back to the 1800s. (Unless you like the 1800s, in which case, enjoy)

"I would guess I'm one of a handful of top individuals on Earth outside the frontier labs, in terms of my experience with top-end models."

I'd argue that number of tokens burned does not translate into "experience".

And if you needed to spend $300k or so to find out that process helps manage larger efforts (process is what "fences" are outside Yegge-land), you sure spent a lot.

I mean, I'm glad the guy's got a a hobby he enjoys, but there's less insight than you'd hope for.

This may be hard to understand, but that was an illustrative example on why sometimes a few degree difference have bigger absolute impact than you'd expect from the purely numerical value.

Nobody (well, almost nobody, I guess) considered it an explanation or model of climate change.

Mid 50s here - you can start at any point. Can confirm.

The past is in the past. It had good and bad outcomes for you. Those stand, and you're best served by making your peace with them. All that matters are the decisions you make now.

4A still doesn't permit you to destroy the evidence. Resist on 4A grounds, destroy the evidence, 4A reasons get overturned - you've got yourself a conviction. US v. Akram Musleh.

While I think it's an abuse of power from a moral point of view - yes, that would be the expected legal outcome. Under any administration.

You can refuse to hand over access. You can't go torch evidence. Caught Ollie North as well.

It's working reasonably well for me.

Given your description of the behavior, my guess is context window compression bit you. Less likely if it's a root level CLAUDE.md and you used a recent version of Claude Code.

Possibly. But telling it in a rules file eats significantly less context than if it needs to read a bunch of files every time just to find what "conformant" means.

AGENTS.md is a poor substitute for agents actually learning, but it somewhat works.

And sure, you can (and should) enforce constraints via githooks whenever you can (determinism is good), but if the agent only finds out at checkin time, it's created a whole bunch of code it needs to redo, and that doesn't run properly if you do the caveman thing and actually run your code before committing. Again, token & context waste.

If your writing gets called "slop", you might want to look at why. It's usually and indication it's considered low quality writing, and... you can fix that. (Unless you use AI, then you're doomed ;)

Uh. The .md files are also there so I don't need to say every single time "In this house, we believe in uv & PEP723. Also, no fucking localized imports, no just-in-case try/except."

(Or the equivalent incantations for other languages)

They aren't there for others, they are there to make sure the code produced measures up to my standards. Sure, it helps that it also applies to requests made by others, but that's a bonus.

It's... a weird take.

"Sure, the model makes repeatedly the same mistakes, and I could prevent that, but then it wouldn't be the same model" sounds like an excuse for masochism.

The one small bit of truth is that yes, instructions might become outdated, and they might affect negatively how the system performs. You fix that by... updating your instructions.

And if you deeply care about that potential negative impact (most models are pretty good at just ignoring long outdated issues), you run evals on model upgrades. You certainly don't refuse to use anything but the bare model.

(Of course, there are also folks who refuse to write code in any way except with sed, so there we go)

> you can always short the stock.

Which is going not too badly for the shorts, so far. There's a reason almost 50% of the float is loaned out.

We'll see once the earnings call rolls around.

Given doing better costs money, and people don't want to pay for better, leaving that place is hard. We've seen the same play out with physical goods - the cheapness of just-good-enough-but-right-now-cheaply outcompetes quality, which costs and takes time.

We as society are choosing to be in that place.

> Use Linux/KDE.

Yep. I, too, get more productive by spending a PhD worth of effort on choosing a window manager.

"No information about your text is ever sent to the server, and there’s no way to identify the author based on the minted certificate"

This means the certificate is independent of author and source text.

There's nothing stopping you from sending fake counts/duration to the semoi server. It's a certificate that only says "at this point in time, this is the information I was provided with".

You can then attach it to any piece of text you like.

At the very least, you'd need the ability to prove that there is an underlying event stream with these characteristics, and that this exact event stream creates the document in question. You still can fake that event stream, but it becomes enough work to distract at least casual abusers.

But really, it's the equivalent of saying "I wrote this without AI, honest" in-doc and signing that with your personal key. The value depends entirely on your willingess to be truthful. (IOW: I predict we'll see a resurgence of reputation systems, to some extent)