...that there's a LOT of minerals and other nutrients in food, only a fraction of which are produced from chemicals in fertilisers, O2, and CO2. If you produce too much with too little consideration of the impact on the soil, you can produce marvellous dust bowls but eventually that's ALL you will produce.

There's a lot of stuff that is on the Internet that doesn't end up in AIs, either because the guys designing the training sets don't consider it a particular priority or because it's paywalled to death.

So the imbalance isn't just in languages and broader cultures, it's also in knowledge domains.

However, AI developers are very unlikely to see any of this as a problem, for one very very important reason --- it means they can sell the extremely expensive licenses to those who actually need that information, who can then train their own custom AIs on it. Why fix a problem where the fix means your major customers pay you $20 a month rather than $200 or $2000? They're really not going to sell ten times, certainly not a hundred times, as many $20 doing so, so there's no way they can skim off the corps if they program their AIs properly.

Let's take a look at software sizes, for a moment.

UNIX started at around 8k, and the entire Linux kernel could happily sit in the lower 1 megabyte of RAM for a long time, even with capabilities that terrified Microsoft and Apple.

The original game of Elite occuped maybe three quarters of a 100k floppy disk and used swapping and extensive use of data files to create a massive universe that could be loaded into 8k of RAM.

On a 80386SX with 5 megabytes of RAM (Viglens were weird but fun) and a 20 megabyte hard drive, running Linux, I could simultaneously run 7 MMORGs, X11R4, a mail server, a list server, an FTP server, a software router, a web server, a web cache, a web search engine, a web browser, and stil have memory left over to play Netrek, without slowing anything down.

These days, that wouldn't be enough to load the FTP server, let alone anything else.

On the one hand, not everything can be coded to SEL4 standards (although SEL4, by using Haskell as an initial language to develop the core and the proofs, was able to cut the cost of formal programming to around 1% of the normal value). On the other hand, a LOT of space is gratuitously wasted.

Yes, multiple levels of abstraction are a part of the problem. Nothing wrong with abstraction, OpenLook is great, but modern abstraction is mostly there due to incompetent architecture on previous levels and truly dreadful APIs. And, yes, APIs are truly truly dreadful if OpenLook is the paragon of beauty by comparison.

The solution is easy. WiFi 6 is only just starting to come out in the marketplace. If TP-Link hijacks the standard development procedure, solidifies a workable WiFi 8 quickly, and manufacturers/users in Europe, Asia, and Oceana all start using WiFi 8, skipping WiFi 7 entirely, the US will be left with an inferior standard that only they have gear for, with no option to use WiFi 8 for many more years because the only manufacturers making it can't sell in the US.

We'll be fine, then. There's lots more gold than Sarah Conner clones.

Precisely.

There are 1500 genes involved. As effects are likely not merely down to specific genes, but gene interactions, you're going to need a model that can handle 2^1500 different permutations. That's simply not something that is classifiable.

As far as gene therapies are concerned, since autism seems to involve combining elements of Neanderthal neurology with homo sapiens neurology, the obvious fix would be to add further Neanderthal genes where combinations are known to produce adverse effects.

What did you expect from an algorithm named after the Roman Empire?

There are 1500 genes associated with autism and nobody has them all. This gives us 2^1500 different forms of the condition. So, yeah, more than one.

Autism is definitely a complex phenomenon. The number of people diagnosed today in the US is about the same as was being diagnosed in Europe 25 years ago, so no, I don't think anyone is jumping onto bandwagons, it's just Americans are being less stupid.

If human activity was on a fixed time, that point in time could never be visible to astronomers, at least not unless LIGO was rebuilt on the moon. This doesn't necessarily offer benefits, but it would be sensible if this was a possibility that was considered.

True, it means we can't use gravitational triangulation, but the detector is nothing like close enough to being sensitive enough to be useful there.

Either way, between radio astronomy, optical astronony, and gravitational astronomy now being largely defunct on Earth due to humans messing things up, we really need detectors on the moon or on Mars before we can do anything significantly beyond what we've already done. Space telescopes are just too small and although you could precisely measure 3D positions with sufficient precision to do interferometry, it would not be easy.

Basically, each space telescope would need to measure acceleration with incredible sensitivity and time with incredible precision, record over a very long time, then have a means of collecting the data on physical media and bring it back to Earth for combining with the other recordings. Real-time interferometry wouldn't be possible.

You're much much better off building your telescopes on the surface of a solid planetary mass like the moon or Mars.

We're learning that hallucinogens can indeed do wonders for depression and can even result in some degree of brain repair, but they are also capable of worsening depression, causing further brain damage, and even creating whole new conditions the patient hadn't previously suffered with.

This has been under discussion for well over two decades, in both the US and UK, and, frankly, I'm horrified that there hasn't been much, if any, meaningful research in many of the substances, with the result that the horror stories rival the success stories in magnitude. We could have avoided ALL of that simply by finding out who benefitted (is there a specific set of conditions? a genetic contribution to outcome?) and who worsened. It was gross incompetence by the governments of both countries and the corresponding health research grant bodies, who damn well AUGHT to have done the legwork, because it was perfectly obvious to everyone that if nobody got told anything practical, people would start experimenting on their own. A far worse outcome, because now we've no idea of why the difference in results, nor what can be used to repair damage where damage is repairable.

Ignorance is useless. The craving of it is, in all honesty, extremely irritating.

By now, after two decades of calls, we should know precisely what genetic and experiential outcomes make which substance applicable and what the correct and safe therapeutic dosage should be. We don't. We don't know anything. Oh, sure, we know it helps some people, but we can't predict who, why, when, or how to repeat those results without messing things up. We've had two flipping decades to learn that. We didn't.

I am, as you might have figured out, not happy, although that might not be obvious to all.

I am not going to say substance X is good/bad/ugly, because (a) no substance works that way, and (b) nobody did the proper research to find out.

(Yes, there has been a little bit, here and there, but nothing I'd consider systematic - it's very piecemeal and not much of it has been replicated. But it's not enough to have any confidence in reliable results.)

ChatGPT is struggling to produce valid SQL even for basic stuff. I use it to produce technical stuff, but everything is checked against at least 2 other AIs and is hand-verified by me, because otherwise there are too many errors.

This is not even close to usable for anything technical.

On one project, after 8 months, ChatGPT, Claude, and Gemini (combined) have managed to produce a technical specification that is as riddled with errors as a draft document produced by an engineer, so one could (I suppose) argue that it's not worse than a human, if given sufficient time. However, the specification has probably cost millions - if not billions - in electricity at this point. If the same amount of money had been spent on human engineers, over the same length of time, my guess is that the document would be of much higher quality and much closer to something that could be implemented.

True, I would never have been given that kind of cash, so it has democratised wasteful spending, but that's not necessarily useful.

The point is that people are investing vast sums of money to create elaborately-packaged boxed sets that are simply too vast to be actually enjoyed (apparently, the new boxed set Thunderbirds will include heavily restored footage that simply wasn't capable of being included in earlier releases), and upscaling a puppet show to 4K and still have it watchable is far from trivial -- those puppets were never made to be seen on such large screens at such high resolution. The scale of investment into making this publicity stunt and boxed set is incredible, the cost of the set isn't low, and the value of the material that's in the set - even to die-hard fans - isn't nearly as great.

Goblin/Guardian: The Lonely and Great God is an even more extreme example and includes 270 minutes of backstage footage, a large pack of publicity photos, scripts, and a tacky plastic sword. It's an extremely limited edition special collector's edition and the resale market is pricing it as though it includes a couple of solid gold ingots. People will certainly binge-watch the episodes once or twice, which will undoubtedly be in much higher resolution than the rare streamed versions, but not even the afficados will be watching all the making-of footage and the scripts will doubtless be on the Internet somewhere. Unlike high-end sci-fi, though, the storyline is simple so the difference between the scripts and fan-produced transcripts won't be vast. (It was a very good storyline, I was impressed, but it was hardly a case where the tiny nuances matter.) But K-Drama is milled in unimaginable quantities, so much so that many series just can't pick up any kind of audience and are abandoned. It's not produced for repeated watching and the odds of any show, however good, being repeatedly watched (the way fans repeatedly watch LoTR or SW) is essentially zero. But someone had to trawl through all the footage to put together the set, make the booklets, etc, and that wasn't cheap. The boxing is elaborate.

The importance of storytelling is high, but none of these are sophisticated stories. They're all pretty much on-par with Smith of Wooton Major - a great little read, but not one I'd pay £500 for, even if they did throw in a plastic sword. I'm not convinced anyone is buying these sets for the content, even though the content is enjoyable.

The degree of investment is phenomenal, the sophistication of presentation is exceptional, and the fans are buying in quantity. I'm just not sure what the benefit is, on either side.