timjver

The whole point of Lean is that you don't need to understand the entire proof to be sure that it's correct. You only need to understand the definition of the theorem being proven, and you need to trust that the relatively small core of Lean is correct.

That doesn't really answer the question of how they're going to detect it at scale. Ban repositories with long comments?

> A 3rd party hosting platform like fireworks.ai that hosts open weights models won't provide any of that at all.

Why couldn't they?

So then bots will just intentionally fail the test?

> Over treatment can be a real problem.

Indeed, but having more data might be able to solve that? The whole problem seems to be that benign conditions sometimes look scary because we're currently not able to predict well enough whether it's something that will eventually cause problems.

It's equally signaling that other US-based labs can't provide reliable access to their closed-weight models.

Verifying that every step in a (potentially long) proof is sound can of course be much, much harder than verifying that a definition is correct. That's kind of the whole point.

By "online wallet" they were likely referring to the Bybit website being the wallet of those customers that held their coins there rather than keeping them in their own private wallets, and not whether the hack involved a hot wallet or a cold wallet. Calling it a custodial wallet would have been more accurate.

> It would be similar to if I claimed that an LLM is an expert doctor, but in my data I've filtered out all of the times it gave incorrect medical advice.

Computationally it's trivial to detect illegal moves, so it's nothing like filtering out incorrect medical advice.

> in the absence of other celestial bodies the satellite would be in a stable orbit

Presumably entering such an orbit is only possible due to forces from other celestial bodies in the first place, since otherwise if you reversed time it would spontaneously leave its orbit. In other words, the act of the earth "capturing" the object is ultimately performed by external forces?

They said nothing about not printing any non-primes.

VanMoof bikes are known for breaking quickly. Calling them "so much better" than Cowboy based on a single bad experience doesn't seem totally reasonable.

Simply twisting a single corner piece or flipping a single edge piece achieves that already, without having to mess with the stickers.

Whether or not it is a coincidence depends on whether Obsidian planned this. Apple isn't going to base their release schedule on something like this.

Give Copilot a try, it has been way more reliable for me in terms of giving good code suggestions than ChatGPT so far.

That's probably an impossible task. The best they can do is ask contestants nicely not to do this, but that opens the can of worms of whether tools like GitHub Copilot should not be allowed, either.

Just in case performance matters, there is a more efficient way: have the tortoise stay in place and advance the hare only one node at a time, and assign the hare to the tortoise every time a power of 2 number of steps have been made. This is known as Brent's algorithm, and it requires fewer advancements than the original tortoise and hare algorithm by Floyd.

Another notable advantage of Brent's algorithm is that it automatically finds the cycle length, rather than (in Floyd's case) any multiple of the cycle length.

https://en.wikipedia.org/wiki/Cycle_detection#Brent's_algori...

Right, but will they be eligible the year after? If not, then this still means that not surpassing the threshold could increase their revenue in the long run.

Not to mention that that was a comment on GitHub, not on Reddit, even though the "far, far, far over the line" statement was about "the now-usual Reddit uproar". I guess Klabnik just really dislikes Reddit.

Yeah, that comment on the GitHub issue was definitely over the line, but it wasn't on Reddit. The comments in the subreddit were higher in volume, but way more reasonable.

This could very well be thanks to the moderators, but I hardly saw any nasty comments on r/rust. People were critical of the actix maintainer, sure, but I didn't see anything that crossed the line. Some comments in the GitHub issues were indeed nasty, but those were actually called out on the subreddit.

There's no way for me to know for sure, but it seems as though Klabnik was exaggerating here.

You're not wrong, but the author's code does in no way rely on the fact that you could theoretically iterate over all the valid Int values in Swift, so it would have made no difference if the Int type was somehow truly infinite.

And the author clearly acknowledges this:

>it’s best to think of Int as modeling the infinite set of all integers

I'm not a huge fan of Optional.matching (also commonly called Optional.filter) because you may want to "filter" non-optionals as well, i.e. T -> Optional<T> given a predicate. Unfortunately Swift doesn't (yet?) allow writing extensions on any generic T and the alternatives are a free function and an Optional initialiser, neither of which allows you to easily chain them. (Another alternative is a custom operator, but let's not.)

So perhaps the best we can do is Optional.matching as defined in the post, combined with turning a non-optional into an optional using Optional(...) in case we want to "filter" non-optional values using a predicate.

I think the lack of left-right symmetry was the surprising bit.

>but still we may introduce the problem to places that did not have it before.

And perhaps in some places this will cause the problem to disappear. I think it will be fine.

I doubt the difference will only be 0.2%.

Or a problem of mixing class inheritance with protocol conformance, perhaps.

Isn't operator overloading something else than the ability to define custom operators?

What makes it immoral for Buterin to have a large amount of Ether? Would it be immoral for a very early investor to buy a lot of Ether for a very low price?

>Considering his share of ethereum is worth billions

Citation needed.