fph

> my money is on food being a major contributing factor

That's not how science works either.

American big tech companies pay very few taxes, compared to European standards. This is part of what makes it difficult to compete with them.

Everyone in Europe is saying that the CCP is illegally subsidizing EVs and solar panels to conquer the European market, and that we need to protect from them by imposing tariffs: but if you think about it, this isn't too different from what America is doing with tech.

Ah ok fair, if your GrapheneOS isn't running the sandboxed google play services, not even in a separate profile, then I don't know if it works.

A dozen is the number of banks that whitelist Grapheneos keys via hardware attestation.

Curve in Europe (which can be linked to any card) and Paypal.

Same as what happened to "crypto", unfortunately. It's now hard to explain people that Bruce Schneier isn't in the same field as Sam Bankman-Fried, which is a shame.

Because that doesn't void your home insurance, as GP wrote.

> IoT and anything that cannot receive time updates have trouble with leap seconds.

Could you make an example application where it could be an issue? I imagine it would have to be an appliance that needs high precision time, is not connected to the internet, and hence it has an internal clock that is accurate over long times.

It might actually be the other way round: Fortran adopted the convention that i,j,k are integer variables because that was the standard in maths.

Warning: sneaky paywall that appears only after scrolling.

Much better security. Sandboxing and app isolation actually works on Android.

You might be mixing them up with Mullvad.

So a fair headline would be "OpenAI tried formalizing more of their proofs, and found 33% of them had a mistake".

No. They have a new upper bound for matrix multiplication, but it's O(n^2.25). You can't do less than n^2 for obvious reasons (size of the input and output).

Many mathematicians I know are shaken or depressed by these news of machiines that might put them out of business. Especially those who don't have tenure yet.

...but we're not talking about diseases. Publishing an AI-generated Navier-Stokes solution does not save lives. (And, in fact, it harms some.)

Most mathematical results are shared on Arxiv. Journals add peer review.

Is it better than regular uBlock origin? I thought (possibly wrongly) that the "lite" version was for inferior browsers with a crippled Adblocking API.

But, most importantly, is Bundesunterhaltungsministerium a single token in Kolibri? This could make a huge difference for the model's performance in German.

> akin to gatekeeping how someone should breathe air.

The law is also a sequence of letters available to anyone. Yet you cannot practice without passing the bar exam. And that is not the only example; many professions can only be practiced by licensed individuals: medical doctors, journalists, electricians...

To be fair, you picked a well-known tricky benchmark for LLMs: When working on an embedding spelling disappears after the embedding level. I imagine modern frontier models have tools that let them read back their input to work around this issue.

The classical solution to the secretary problem works under two very unrealistic assumptions: (1) you win only if you pick the best candidate. Picking a close second is worth nothing. (2) you have no prior information about the expected distribution of the skills of possible applicants.

So I am not sure we should really follow that heuristic in real life.

Probably the price for Moto signature is that low only because they're right about to launch the next version.

Not a very effective way to support the devs though: only about 60% of what you pay gets to the developers, because of Steam's cut and VAT.

It's in the alpha channel for now, you have to join the alpha channel to get it immediately (at your risk).

Maybe more like the baseband modem in a phone?

The huge UX drawback of password managers vs. passkeys is sites that have arbitrary requirements. "Your password must be between 9 and 12 characters long, and include at least two capital letters and a special symbol". Well, my password is randomly generated, and enforcing that there are two capital letters would just reduce entropy.

If sites fixed this issue instead of pestering me for a passkey I'd be happy.

Is it a land border? Because if they are separated by a stretch of sea, well, the same is true for Canada and mainland France.

Do I understand correctly that the main workflow is "pick the first Github repo returned by the search, check that it is legit by clicking the link, then install the app from it"? It seems very easy to make a mistake and install a malicious fork.

Why would they support SMS then?