ajb

I don't want to get into an argument about religion, so I will make a different point.

There are two ways to evaluate things. One is selection: look at the options available, decide which is best, and go all in on that one. The other is to understand things as deeply as possible, what the moving parts are,what makes them tick, what the unknown possibilities might be, etc.

We as individuals have to use selection a lot of the time. You can't do only the good parts of two jobs, or marry only the good parts of two spouses. But it's also profoundly negative to be obliged to only think in those terms: for example, when citizens are told that they have no choice other than to vote for the lesser evil, it lets politicians off the hook for only being slightly less corrupt, and reduces citizens to a passive role. Ironically, those who try to shut political discussion out of HN are doing the same thing: saying "just choose, do not think".

So, while I might be interested in instances where religious people have a longer term approach, it's not interesting to me as a "plus for religion". It's interesting if it leads to understanding and new ideas. Of course, as a non-believer I would like to separate any benefits from the need to be loyal to unprovable assertions. But that's actually secondary.

An uncorroborated claim of integrity is self-refuting.

You joke, but, leaving religion out of it, it's plausible that if the want long lived infrastructure that's maintained with integrity, it may be that tithing of some description is part of the solution. Currently the closest we have is patron, but most of those are still part of hustle culture rather than the supporters feeling a long term obligation.

'Some versions of the tale differ from Goethe's, and in some versions the sorcerer is angry at the apprentice and in some even expels the apprentice for causing the mess. In other versions, the sorcerer is a bit amused at the apprentice and he simply chides his apprentice about the need to be able to properly control such magic once summoned.[] The sorcerer's anger with the apprentice, which appears in both the Greek Philopseudes and the Dukas score (and its film adaptation Fantasia), does not appear in Goethe's "Der Zauberlehrling".'

Do you have any thoughts on Mozilla? It has a two-entity structure, but there are a lot of complaints that it is not sticking to its mission (or, slightly differently, that it's not succeeding at it, due to poor strategic choices).

(Apologies if you already addressed this somewhere. Thanks for doing this)

I don't think we necessarily disagree. I am pessimistic about laws being effective in this case, but that doesn't mean we should not try to find ones that are. I like your idea. Thinking and trials in that direction would be good.

Data using organisations often seem to prefer fig-leaf laws that aren't effective, and lobby against ones that might be effective. "My data use is a good use, therefore I should not be subject to restrictions and oversight". Instead, anyone with a use of data which is valuable to the public should not see themselves as on the same side as the advertisers and surveillance vendors. They should see themselves as on the opposite side.

That's all very well, but we just plain don't have a legal, economic, or technical system which will allow separation of the good uses from the bad uses. Once data is in someone else's possession, there's f-all way to prevent it being used to do whatever the possessor wants. Even if there is a legal agreement, it's easily abrogated, or overridden by insolvency law, or by a company having a "we can update our terms" clause. Some of this I can imagine how to address - insolvency law could be changed, for example - but in the absence of a fully robust system, promises of "we will only use your data for good" are not credible. Those who actually want to use data for good should be on the side of robust assurance of that, not just plead that they can be trusted and that no accountability is needed.

Yes, but this is preaching to the choir.

The counter must be as visceral is the claim. They make an emotional pitch:your children are in danger, surveillance is the solution. The counter must show the dangers in visceral, emotionally relevant way. This surveillance is actually a risk to parents and children as well - that by the accusation of an opaque, unaccountable system, you will be labelled a pedophile, and your kids taken away. That when sharing a picture of your own child with your own mother, you will have to worry about what the electronic bureaucracy will label your picture as.

Abstractions like privacy,and categorical claims, aren't going to reverse this. A properly pitched campaign could do. Sure, complain that politicians and the public are dumb. That may make you feel better but it won't change this an iota. Talking to people in the terms they care about might.

Depends how it works in the jurisdiction, but in common law usually this form of restriction is like a contract, but between two pieces of land, or between a piece of land and the public, rather than between two persons. In the former case only the current owner on the benefited property can complain. In the latter case, any member of the public can - but I'm not sure if a member of the public can create such a restriction.

Training and advice is about the expected disaster, not necessarily the one you are in.

In 9/11, lots of people escaped the south tower using elevators, which is the opposite of what you're told to do. There are people who died because they chose to follow the stair route. But others died because they were trapped in elevators.

I appreciate the sentiment, but this is exactly like the expectation that people can be responsible for intervening when self-driving or driver-assistance goes wrong. Human brains are strongly driven to conserve energy. If nothing seems to be happening - when errors become less and less frequent - the more difficult it becomes to guarantee intervention, and the less practiced the human will be at doing so.

I have written factory tests, in which I injected errors to make sure that the factory workers didn't develop "click next" syndrome and actually noticed errors. That's what you'd have to do. It's hard to get an organisation to stick to that, when they add up the time they're paying for in detecting fake errors.

Not necessarily, just that they don't have as many as they can make use of, and that xAI can't make more valuable use of them than renting them out.

This is valid for some use cases, but then it needs to be compared with other solar distillation methods, of which there are already a variety at different levels of energy efficiency, complexity, and land use.

It used to be that VCs wanted founders to pursue a singleton strategy, because they wanted the diversification to occur at the level of their portfolio - on the grounds that at the level of an individual startup, diversification would less likely to succeed. The catchphrase was that they wanted "pure play" investments. So it would be interesting to know why it's now a cause of friction. (The "pure play" thing was actually from before YC took off and gave founders a bit more agency; and before the "pivot" was popularised; I get the impression that VCs back then were not too happy for a founder to change direction and confuse their portfolio)

There is a fundamental minimum amount of energy needed to desalinate: you can't take less energy to do it,than you could gain back (from osmotic pressure) if you allowed the desalinated water to expand a cylinder containing the residual brine. This is large. This paper is a thermal method, so it doesn't have an electricity input, but to justify their efficiency claim, they should really compare against what you could do by using the same surface area for solar panels, driving a conventional setup. My (limited) understanding is that conventional reverse osmosis is not far from the theoretical optimum, energy-wise, the main difficulties being operational (the membranes need declogging). And of course RO is more expensive than rain.

This paper is interesting, however, in directly producing crystalline salt, which is lower volume than brine and easier to dispose of, maybe even valuable.

I think the GP's point is that accountability/responsibility isn't a substance, it doesn't have to be conserved like energy or momentum. I agree with them.

It would be perfectly valid for the law to be that individuals don't need to unpick the corporate web of relationships, but hold any of those who contributed (above some size threshold) culpable for the whole injury, and leave the corporates to arm wrestle about how culpability is assigned between them.

Odd - they've had that for years, but only on the command line. Wonder if it's different under the hood? They also have firefox containers which also never quite became a first-class feature (you have to install a plugin).

I don't know if it's specifically UK terminology, but here we don't call bathroom silicone "caulk" - we use that term for decorators caulk, which is much more sticky; so needs to be treated differently to silicone. Definitely true that the little tools are amazing for bathroom silicone, not sure what you do with the other stuff

SanDisk has designed a flash equivalent to HBM, which has 1.6TB/s of bandwidth. I expect that it will be available initially to server manufacturers only, but once supply ramps up will be built into individual machines. At that point it will be practical to run local inference on much larger models. Of course, maybe the SOTA providers will find some way to use even larger ones, but it seems like the returns to scale aren't as much as they were.

Not a direct answer, but there is a standard code for what actions emergency services should take, and the code[1] for this stuff is 3YE, which means "Use foam or dry agent, substance reacts violently/is explosive, BA [breathing apparatus] use is essential, evacuate vicinity, contain spill." So there must be some reason not to let the stuff out.

[1] This is the hazchem code. I think the US uses a different system. A list is here: https://www.ricardo.com/media/radn55jg/dangerous-goods-emerg...

That's a very interesting page, but the language ranking is wildly different for "average percentage" (python bottom) and "success rate" (python second). Sounds like there is some subtly about this.

The argument works at the margin as well, though. Suppose companies deliver value X and also extract rent Y, and defend Y on the basis that it would threaten the value of pensions to prevent it - this is rebutted in the same way.

It's true that the current pension provision depends on this. But if pensions are mainly funded by companies which extract monopoly rent, then it would actually be more efficient and be less distortionary to the competitive market to fund them directly out of taxation - one big, simple rent instead of 100,000 different ones clogging everything up.

In effect, robots.txt should have an "embargo" directive?

Interesting, but - it says it's a co-op, but this is a word with a legal definition. Every co-op I know publishes what it's legal entity is, how voting is done, etc. This just has "click to join". If your pitch is that you have more integrity than typical companies then you have to, you know, "walk the walk".

Also in some jurisdictions, calling yourself a co-op without actually being one will get you into legal difficulties. Companies that don't quite fulfil all the requirements are careful not to call themselves once. Igalia, for example, are very serious about being worker owned and run, but they made the choice to have a slightly different structure so they don't call themselves a co-op.

TBH something like this sounds useful even without LLMs (although I haven't fully grokked this yet). The problem with the operational level is that you can't express the invariants etc at the type level - not least because you're working across multiple languages - so the kind of dumb issues that we're beginning to rule out at the level of the language at the process level still require lots of diligence in operational code. Some kind of shared "operational type system" that could be integrated into relevant languages would potentially help a lot.

Amazing for 250 year old automaton. A bit odd that the show is it catching a fish! Swans are not predators. Given the immense diligence it must have taken to build, I hardly imagine they overlooked that fact - they must have cared less for ecological accuracy than for making a good spectacle.

If we are outdone in intelligence, diligence, and creativity, humanity will have to fall back on its weakest talents - integrity, fairness, and compassion. What a nightmare.

I think that may not be the potential used for electrolysis, but the chemical potential of the saltwater-metal boundary. But hopefully someone more knowledgeable will comment.

"We must now Single-click "Select" on that icon to actually bring the application to the forefront and activate it. I don't know what that's all about, but that's how it works."

What that was about was that all gui apps on riscos only ran one process, no matter how many files you had open. These machines had very little memory, so managing it was very important - there was actually a system panel you could open (I forget it's name) where you could drag sliders to change how much various things were allowed to allocate.

The downside, of course, was that if some app crashed, it would take out every file you had open with it. But then, it didn't really have very good isolation, so often a crashing app would take down the whole OS.