Just as ironic as the search in Google Play Store. Also made by Google, also one of the most useless search features I've ever used. Unless you know the exact title of the app you're looking for, down to the letter, you have very slim chances of finding anything.

Yeah, I wonder why he wouldn't want to go into a cherry-picked study that just so happens to come to the opposite conclusion the tens of thousands of other studies and data series do...

Cooling period, are you for real? Where have you lived for the past 50 years? XD

I regularly have 2000+ open tabs. I probably mainly ended up with this method because Firefox specifically allows to do this without any slowdown or excessive memory use.

What usually happens is that I open a lot of "new" things of something in background tabs. New record release pages from a newsletter, articles from a news site, etc. I the go through them rather quickly, closing each as I'm either done with it or decide after a glance that I'm not interested (~90%).

This way I don't have to actively keep track of where I am in a list or what I've already seen, I don't lose any "unprocessed" items across sessions, and I get a sense of progress and how much I still want to check out. So I might open 2000 tabs when going wild with a few newsletters, and gradually work them off ovea a few days.

It's what's called voting with your wallet, and pretty effective when scaled up.

If I have several options that would satisfy a need, why wouldn't I pick the one where I *don't* support a person actively making society worse by sewing hate?

Same for me. In fact, I almost never get performance or memory leak issues on Firefox, but do so all the time on Chrome, Edge, and Vivaldi, despite using Firefox a lot more than the other three combined. Especially performance-wise, no other browser can hold a candle to Firefox, in my experience. Open more than 10 tabs, and all the others start panting. Firefox merrily skips along with 2000+ open tabs and doesn't even as much as slow down noticeably.

I do get the occasional problem on a website but, curiously, those tend to disappear immediately after enabling a User Agent spoofing extension to pretend that I'm using Chrome. Conferencing tools are the worst with this.

That's not at all clear. We're not talking about a knowledgeable personal tutor here, but a next token predictor optimised on stringing together sentences that the user is most likely to believe is an actual answer to their question. Their potential to be counterproductive is about as big as to be helpful.

Also, pointing out that something is conjecture isn't an ad hominem.

I keep hearing this online (though never ever from friends or coworkers). And I wonder how it's possible for experiences to be so different.

For me, Chrome is one of the most sluggish, laggy, leaky pieces of software I've ever used. It struggles to hande more than 10 open tabs, the UI keeps freezing, and it crashes all the time. This is true on all of my machines, across Windows and Linux, from a 20-year-old Thinkpad to a 2024 overpowered workstation.

Firefox, meanwhile? Handling my regularly four-digit open tab counts without as much as a noticeable delay in anything, ever. It's as snappy as I wish all software could be. Reasonable memory usage, maybe two crashes a year even though I sometimes don't close it for weeks on end (save OS sleep over night).

Chrome vs. Firefox has been snail vs. cheetah for me for more than a decade.

"Silicon Valley was founded on military development […]"

That's quite a blindered view. It was central for funding, sure. But that's ignoring who came up with all of the actually interesting, influential, revolutionary stuff in the Silicon Valley of the 60s and 70s. It was the original hackers, a generally very anti-war, anti-establishment, anti-MIC scene. The origins of the SC tech and founder scene are within the counterculture movement.

I find this story as funny as the next guy, if the next guy is a guy who finds the story very funny. But the way the story is reported in most places, including here, is exaggerated, bordering on disingenuous.

If you look at the company's marketing materials, they never claimed AI was going to build your app. Their sales pitch was "human-assisted AI". You were supposed to work out your solution with a chatbot, after which a human dev was going to take over and develop it for you. Supposedly more cheaply because of heavy leveraging of AI, but their hundreds of human developers in India weren't a secret. Apparently, they severely understated the extent of the human assist needed, but this wasn't really a moment of opening the door of the Mechanical Turk.

I think it's good that this weird thing went bust, though. If you want to sell outsourcing to Indian developers, just freaking say so. No need for all the nincompoopery.

Pre-2014 history about Russo-Ukrainian relations is irrelevant at this point, except for historians. One country started an unprovoked genocidal invasion on another. That's the whole story. Who gives a flying fuck what stupid justifications they try to come up with?

That analogy doesn't work. Automobiles get you the same places horses do, just faster and more comfortably.

A proper analogy to human writing vs. GenAI regurgitation slop would be an unlicensed, uninsured Uber driver who picks you up in a car with the backseat still covered in their last ride's vomit, then taking you to the wrong place and and insisting that this is where you actually want to be. I'll take the carriage, thanks.

Anecdotal, but one of our younger team members who's still at uni told us that the only lecture without rules against LLM use for programming exercises is the database one, because all the models fail to generate useful SQL anyway.

That being said, I sometimes try it for TypeScript and C#, and don't usually get results I'd consider useful either.

I know plenty of square people.
Also some spherical ones.

You must live in a very constricted media bubble. If you checked the actual numbers, you'd see that the major predictions, reaching back way into the 70s, have turned out to have been extremely accurate.

But for that, instead of looking for articles that go "look, I found a prediction that turned out to have been wrong", you'd have to look at what the actual "consensus models" (IPCC, etc.) said, under the preconditions that actually applied. "Dead on" would be a fair evaluation when you compare those to the actual measurements.

But we got the same schtick from the conspiracy heads during the pandemic. The actual Covid death toll in the UK was precisely in the range Neil Ferguson's models predicted. But because the conspirationists didn't like that result, they simply decided to compare the numbers to those Ferguson predicted had the UK not taken any measures against the pandemic at all. Well, whaddaya know, suddenly we can once more claim the predictions were grossly exaggerated, how neat!

The golden age of robots.txt might be approaching now, as a legal tool.

Sure, back when what it "protected" you against were search engine crawlers, its use was somewhat limited. You either benefited from search engines indexing your stuff, and if it was stuff you didn't want indexed, it probably shouldn't have been publicly accessible in the first place.

But that changed with LLMs, where being crawled is generally 100% to the disadvantage of the creator and publisher. LLM crawlers can and apparently do ignore it, sure, but it might be a good argument to have in a lawsuit.

If I engag my optimistic mode, I'd like to see a web where most worthwhile content isn't allowed to be used in GenAI training, and those companies who violate those terms are ruthlessly sued out of existence and into debt for damages.