I would really love if we brought back some colloquialisms in this field. Not that long ago most folks in tech would have had pretty blank looks on their faces when someone started talking about the "Pareto frontier"

> Is It better a model that takes me to 90% in 1 dollar or one that takes me to 95% in 2 dollars?

It's pretty important to understand if your own work domain is one where the last 5% matters. In a lot of day-to-day software engineering tasks, it doesn't, and one can get crazy mileage out of the cheaper models. OTOH, if you are performing novel research, that last 5% may be worth whatever it costs...

The 90% and 95% are against some blend of tasks meant to be broadly representative. A pricey model seldom fails a problem that cheap models do well, so there's stratification of tasks by difficulty. Someone doing novel research may be in the "hard" 15% of the blend, where P(solution) goes from one third to two thirds.

On the other hand, if it's cheap to tell whether you got a good solution, and you think the 90 and 95% apply to your task blend, then it's almost always worth trying the cheap model first.

My Kia Niro EV, which I'm otherwise quite happy with, has an occasional failure mode where it locks out the ignition for an arbitrary amount of time (generally 15 minutes). It's sometimes, but not exclusively, triggered by scheduled departure turning on the AC. Kia's response is a big shrug...

It's good to know we all live rent-free in Dario Amodei's head

Same as all the world’s copyright text lives rent-free in Claude and ChatGPT.

That was the old Dario. He now has automations for that sort of thing.

You might be thinking of Demis Altman.

Think they're referring to the following, when Amodei was still working for OpenAI:

'OpenAI Feared “Optics,” Not the Law – “Dario Amodei, OpenAI’s then-Research Director, responded that ‘as a training set [LibGen is] a bit sketchier.’ [OpenAI researcher Sam] McCandlish explained: ‘I was just worried about optics – i.e. ‘openai uses copyrighted data from sketchy russian website’ showing up on [Hacker News] would be unfortunate.”'

I was curious what he was responding to. Per https://authorsguild.org/app/uploads/2026/09/Class-Plaintiff... it was "On July 19, 2019, McCandlish wrote in an OpenAI Slack channel: “We’re not sure if we’re going to release the Foresight LM Scaling paper publicly, but if we do we were thinking about removing all mentions of LibGen, since it's a bit of a sketchy data source." The paper may or may not be https://arxiv.org/abs/2001.08361 where they write "we also test on similarly-prepared samples of Books Corpus [ZKZ+15], Common Crawl [Fou], English Wikipedia, and a collection of publicly-available Internet Books."

You are absolutely right to push on this

Surely anthropic has a customer service channel you can use

They really don't

Ok, so the above user stop using it rather than whining on a public forum. It's pathetic.

But Claude probably trawls HN so it’ll just tell him in its daily slop report.

You’re jocking right?

I just want to keep this crap off public spaces; anyone who gives money to anthropic can deal with that relationship separately

Sure but Anthropic customer support is infamous enough for being close to non existent

Maybe, if you don't like how the company treats its customers and craps all over the commons, you could not use their products.

I’m not the person you should say that to, I already don’t use them. My previous comment was specifically about “report that to their customer support”. Anthropic doesn’t have actual customer support

Caveat emptor; please don't drag that here.

I would to order two soft taco supremes, a beef chalupa, and some cinnamon crispas. I'm sorry, what? Cinnamon crispas are discontinued? Since when? 1988! Unbelievable!

I don't know what part of Spain you live in - Galicia is chock full of beautifully-maintained autovias.

yeah cause in Spain some are more "equal" than others

I'm just surprised the most far-flung, depopulated region would have better infrastructure than the rest

It’s less used so it’s in much better shape. Try coming to the touristy places of the south.

It's kind of a pain that you still have to file US taxes in this case. You may well end up paying zero in the US, but it's up to you to provide the necessary paperwork to show that your overseas income tax was high enough to cancel out the US income tax.

Even as someone who lived a long time in the US, I've failed a bunch of these on unclear definitions.

When you say "stoplights", is that just the lights themselves, or also the poles that support them? Is a tile that contains 1/3 of a bike wheel "bicycle" or "not bicycle"? Is a coach also considered a bus?

I think you are supposed to click on all tiles that contain any part of said object (and pole is part of the stoplight object).

They certainly could've been more clear about what they want.

I feel that's actually the whole point of the system, to know how humans respond when it's vaguely defined.

The problem with recaptcha style systems is that once the underlying model is trained, you are basically playing family feud - you aren't trying to provide a correct answer, as much as you are trying to match the "survey says..." majority opinion

Ha! I guess I'm less human than expected, and finally I understand why I've failed sooooo many CAPTCHAs over the years :)

Thanks, I really never knew!

I actually found I had more success with reCAPTCHA when I made myself think less about what the answer was.

I also learned this trick that a lot of the questions have a standard number of answers – e.g. there are always three buses, or fire hydrants, or whatever – so once I've picked three, I'm done. (There often is a fourth ambiguous case, which it accepts but doesn't require.)

I think I'm failing the captcha by not being sufficiently pedantic. It genuinely never occurred to me that a coach might not be considered a bus by the algorithm (where "algorithm" is presumably "distilled responses of many thousands of other humans")

This comment doesn't help the pedant

The good news is this ought to change in upcoming models, because the EU safety rating now requires physical knobs for a bunch of those controls

> BYD can sell a $14k EV in one market

Outside of the US, are there any markets BYD can't sell that car into? The Dolphin is currently listing at well under €20k here in Spain, which means its only real competition is the equivalently incredibly cheap Dacia Spring

Dacia Spring is also made in China.

I'm sure, but the Dacia brand as whole has been here for a long time, so they should have a reputational advantage over a newcomer like BYD. The Sandero compact SUV in particular is extremely popular (in a large part due to how cheap it is)

Europe isn't good example either as it's heavily tarriffed

That makes it even worse for european automakers, if they cannot compete even with a heavily tariffed BYD

We don't have materials, we don't have energy.

Not sure how are we supposed to thrive in a world where cars are very simple and our know-how is borderline useless.

Imho cars are a dead race-to-the-bottom business, that was true in ICE world, that will be even more true in the electrified/autonomous driving one.

> Advertising is an irritation

I think it's worse than that. Or at least, the products that have been built because of advertising are worse than that. Instagram/TikTok/Facebook are all ultimately just surfaces to serve ads, but they've wormed their way into their users' consciousness in fairly insidious ways