It's probably not impossible to build a proof language that makes that kind of thing doable, but I suspect that (a) it would be genuinely difficult to operate it skilfully, much as being a really good developer is difficult, and (b) it would take a huge collective effort on behalf of each research community to prove the foundational results everyone relies on.

Whereas the system we have right now, despite sounding kind of weird to outsiders, mostly works okay? I'm just not sure a switch to formal proofs would be worth the time investment -- or that you could convince the many researchers less interested in tech than myself.

A typical research paper is written at a very high level; often steps in the argument will assume the reader is also a skilled mathematician, and invite them to fill in the lower-level details themselves. This reasoning is as much intuitive as it is formal; I suspect it will be a while before proof assistants are as intelligent as the typical reader of a mathematics paper.

I think there's an outside perspective that if math isn't 100% logically verified, it's worthless, which doesn't really match up with my own experience. Most results rely more on the intuition of the authors than on the precise logic they write down; thus the surprising result that papers with logical gaps are, very often, still correct.

Comments URL: https://news.ycombinator.com/item?id=26875298

Points: 9

# Comments: 3

I think side projects can be valuable when you want to demonstrate competence at something, and you don't otherwise have experience to draw on. If you're a barista who's learning to code in their spare time, or you want to understand a hot new technology, sure. Otherwise, I'd rather ask you about your actual work.

It’s such an obvious deal-breaker, I don’t understand how “let’s delete your stuff all the time, even in offline mode” made it through testing.

For the Americans reading, they have a US site too, might be worth checking out?

Definitely if they start trying to haggle, or pressure you into something, that's a hard pass, I'll say whatever bullshit it takes to get them out the door. I'm happy to help out the former at the expense of fielding the latter, though I can understand why not everyone is!

For what it's worth, as an interviewer I'm happy to give feedback in person, at the end of the interview, if the candidate asks for it. It's much easier to do when you're both in the same room, and asking e.g. "is there anything you think I could improve on?" makes a positive impression either way.

Comments URL: https://news.ycombinator.com/item?id=21248168

Points: 2

# Comments: 0

You can just set individual projects, it's not that hard.