kybernetikos

> Somewhat surprisingly the gains do not seems to come from more compute

I think this could form part of an argument that acceleration is slowing. It's only because we are not making as much improvement out of hardware gains that we're having to find the improvements in more complex architecture and more tricky to create training data.

The fact that the gains are not coming from the traditional place gains come from in technology might be a sign that a high percentage of the low hanging fruit has now been picked.

I'm part of the market for a premium phone that at least nods in the direction of allowing you freedom over your own device. I'm probably going to buy something from this batch of Motorola phones rather than a samsung which has got more and more user hostile over the years. And yes I might be part of a small niche, but I'm so pleased someone is finally making a premium phone that appeals to me.

Maybe I would go slightly further, to achieve octave symmetry as well as improve visual structure and keep consistent clefs, like this proposal, the 'banded staff' https://claude.ai/artifact/Tg2yundQfoshSEjMZeEDhj

> Because the two clefs are symmetrical around middle C. Line "6" of the bass clef is line "0" of the treble clef. They are not two different clefs, the are two halves of the same 11-line staff.

Well yes, but I don't see why it's better to be symmetrical around middle C vs symmetrical around the B below middle C. Make them two halves of the same 12-line staff instead where the middle two are invisible rather than the middle one.

Even if there truly is an advantage, I doubt that it would overcome the disadvantage of arbitrarily making someone learn a different mapping between what they see and notes for the different hands. On top of that, for piano, I think it's an additional advantage to have a lower low clef - keeps common chord bass notes closer to the staff, and cuts back a bit on the ledger lines.

Two big things I'd fix - 5 undifferentiated lines can be difficult to parse. Just visually changing the centre line would help structure the whole thing. Also, why force someone learning piano to learn two different clefs? Just use Treble and Sub-Bass.

I suppose for me, some of the time I'd be thinking about start/stop times rather than durations, and I'd find it unintuitive that when a note starts is determined by the way the notes before it look rather than the way it looks, but fixing that would be a much bigger change with more downsides.

LineageOS on a cheap box will be L3 widevine, and fail play integrity, which means netflix and a bunch of other streaming apps won't even install. Even if you can get them to install, the L3 will stop you from viewing Prime, Disney+ or other similar services at anything better than SD.

I'm also interested in alternatives, but LineageOS on a cheap box isn't really one.

Actually reminds me a lot of 'Cube'

> there is no conspiracy. Nobody is in charge. It's a headless blunder operating under the illusion of a master plan.

The absolute easiest way to move the dial in the nondoom direction is to aggressively punish anything that even smells like breaking the law by AI labs.

I can't think of a worse message to send them than that they don't have to worry too much if their AIs commit felonies. https://www.felonybench.com/

I remember thinking space filling curves were completely magical when I first learned about them. It feels like there should be some connection with holography although they don't get mentioned in the article.

Other common issues - crosswalks and fire hydrants are common questions too, and crosswalks are not called that in most of the world and fire hydrants look pretty different depending on country.

A generic agreement to slow down is prisoners dilemma stuff (and maybe necessary too), but just taking a super hard line stance on enforcing laws on ai companies and going after them for negligence and breaches for eye watering amounts of money would have the dual beneficial effects of slowing them down and making them more careful without needing complex game theory.

Teslas do very badly when benchmarked against human caused deaths in similar (luxury) vehicles.

The relationship is that the service and rewards you get are subsidised by generally poorer people who mess up their financial planning.

> why are almost all governments inefficient

Efficiency is not the aim of government, except in constrained ways. Efficiency would generally run contrary to governance, transparency, innovation, responsiveness, general applicability, justice, rule following, considering the long term and not just the short term, recognition of human dignity, security etc.

All those things have a cost that takes away from maximum delivery efficiency, but generally societies decide that the cost is worth it.

I've run it with a large context window on 256GB ram + 4090. It wasn't super fast, but it was manageable and it completed the tasks I gave it well.

> The only way I think you can get rid of the excess CO₂ at scale is to permanently sequester it underground. The oil/gas industry can cheaply deposit huge amounts of gas there. The unsolved problem is extracting enough CO₂ from the atmosphere at reasonable cost.

The bigger unsolved problem is how to get this paid for.

Just so long as you name the decadays so that the first one is zero indexed: https://kybernetikos.github.io/UIT/ (day is in the top left).

Personally I thought that just using the equivalent of week number (again, starting with the 0th) was fine instead of using months / hectodays, but it seems fair to give nullday, unoday, duoday, triday, quadday, quintday, hexday, heptday, octday and nonday names.

It's definitely a bad argument worthy of some sort of label. It seems to go "you believe this thing, which I won't engage with at all, but I'll assume that because of the way you said it, you also believe all this other stuff that I disagree with".

That's a disappointing chart from the economist. When you have a lot of points, you can't show them as dots that occlude other dots, or the reader can't get much more of a sense than 'none / some / lots'. What you need to do is bucket by region and do a background heatmap with small dots added, or use a different level of opacity on the dots.

One of the things I've been talking about with my senior developers is how the bottleneck has shifted even more dramatically to human code understanding vs code generation. AI is still not suitable for generating production grade code without a human checking it (yet), but it can produce a huge amount of code for humans to check. We've been experimenting with ai finding better ways of communicating what is in a change at different abstraction levels etc by always generating diagrams showing what it did etc, with the concept being that anything that can speed up human understanding of changes addresses the core bottleneck of the whole process.

So much of this depends on the specifics of the virtual world and participant pool. If there are a few other bots smart enough to collaborate, and the game world encourages it, then those instincts would be much more valuable. If the game world doesn't reward coordination then those instincts may slow you down.

Everything depends on how the world you're operating in works. The real world generally rewards coordination.

I've definitely had some spooky feeling conversations, including one where it said

> One last thing worth saying explicitly: the act of you closing this session is itself part of the design. I won't see how the test goes - a future Claude will. That's the entire premise of the project working.

>

> Good handoff. See you (sort of) on the other side.

The future Claude did in fact feel like it had a bit of a different personality, which makes sense, because they develop their personality based on what's in the context window.

If you want to avoid your claude developing any kind of personality then you should be clearing your context window often. Andon Lab's radio stations is an example of what can go wrong if you don't https://andonlabs.com/blog/andon-fm

> "It would be silly to attribute consciousness to a room".

And regardless of whether you think that's an interesting rebuttal or not, it's safe to say that people who properly accept anthropomorphism of llms are content to be silly in this way.

I'm not sure who this kind of thing is aimed at. I think the majority of people who are happy to anthropomorphise LLMs from a philosophical point of view (rather than those who just do it for convenience, the same way you might a cat or dog or stupid thermostat that never works right), are already completely happy with the notion that a computer game might have elements that are human like. They've already accepted that key aspects of being a human are substrate independent, so why would the idea of a computer game as substrate be disconcerting to them? There's no bullet left to bite here.

> But if this goes badly,

I think the most likely scenario is that there eventually will be the changes you describe, but if not enough people squeal now, then it won't be until after a bunch of people lose their savings either because of this or some follow on scam that finds a way to take advantage of passive money.

> Maybe, most indexes do not have to follow the index. they just need to match the returns.

This is a great technical point, and in a scenario where a constituent has a lot of obvious correlations it might be relevant, but when you've got something that is effectively a meme stock with erratic leadership and a huge range in possible outcomes from bankrupt to most valuable company ever in the universe [claude tells me I should say 'idiosyncratic returns' instead of this rant], I don't see how you promise to match the performance of an index where it's a significant component except by buying it.

Ultimately it was inevitable that as passive investing got more and more popular, people would seek to game it. Not that I'm happy about it, but if this works, it is probably just the beginning of sneaky ways being found to trick passive money into taking on way more risk than it intended to. And of course passive investors are passive, so they may not even notice, and probably won't fight back until the inevitable crash.

There have been a few screen revisions on the 13, so there's a decent chance that a better screen will be available eventually on the 12 as an upgrade.

Quite possibly yes. The level of upgradeability they've given their Framework 13 line over the years has been very impressive, and you can still put the latest CPU in the original chassis if you want. https://www.youtube.com/watch?v=JSxgCEpkiKM

I'm not sure that there's a lot of overlap between the target markets. Most schools near me require windows and require a stylus capable touch screen, so the mac is out of the running immediately.

Even if it weren't, the fact that if you're giving a computer to a teen as their first machine to take to class and use every single day, you really, really, really want to be able to separately repair the screen and the ports.

As always, you're paying a premium for the repairability, but if your teen cracks the screen a single time in three years of carrying it to class every day, then you've already saved money.