[electrograv]

> I'm earnestly trying to avoid downplaying his work, but it's only a package manager.

I'm earnestly trying to avoid downplaying your comment, but you sound like you have no clue how much engineering effort and talent is actually required to build and maintain a reliable, functional, production-grade package manager that a huge number of people use.

Being able to completely implement and support a sophisticated software product end-to-end is a FAR better indicator of engineering talent than isolated algorithm puzzles on a whiteboard. Whiteboard problems are used in interviews simply because individually executed complete projects are so rare (at all, let alone publicly visible).

In fact, whiteboard interviews have all but been proven to be among the least useful at distinguishing good vs bad candidates. The only reason it works is because everyone knows it's a game, studies the game, and gets tested at the same game. It's essentially a disguised IQ test that you have to study for.

The problem is, if you don't study for it specifically, you're at a huge disadvantage. This leads to famous engineers (clearly talented) getting rejected because they don't practice jumping through the particular hoops the company makes everyone jump through.

It seems there are more than several cases where Google needs a high profile engineer more than that engineer needs Google, and as a result the engineer doesn't study the hoop-jumping Google wants. Google rejects them out of bureaucratic process, and loses out on a good hire.

On the opposite side: I'm not a famous engineer, so I suck it up and practice whiteboard algorithm problems like most everyone else -- and as a result, I've never had a problem passing coding interviews. But just because most everyone jumps through hoops to play the game, doesn't mean everyone should have to.

[godot]

To add to your points, the problem is not just missing good engineers because they don't study and play the game, but also hiring bad engineers because they study and play the game.

I personally know of people who habitually study FAANG interview questions, and make it their career to jump ship every few years for salary boosts. I've worked with some of these people; while they are nice and friendly folks and I consider some of them good friends of mine, they are much less-good engineers compared to many others I know, who stay in smaller startups and want to get things done and don't want to play the interview game.

[diracpam]

I'm one such 'engineer' who is good at whiteboarding algorithms. I can do well on 'system design' questions but dont feel confident at all about really designing such a system from ground up. I, for one, am glad that I can find gainful employment by memorizing 200 pages.

[vanderZwan]

I'm happy for you, but I hope you're also that honest with yourself and your employers about your limits when they want you to built something that you know is over your head

[saagarjha]

> I'm earnestly trying to avoid downplaying your comment, but you sound like you have no clue how much engineering effort and talent is actually required to build and maintain a reliable, functional, production-grade package manager that a huge number of people use.

It's also entirely possible that Google saw Homebrew as neither reliable, functional, or production-grade enough for their taste. While Homebrew is somewhat decent today (though, it's still broken in some ways that I will not go into here and I'm under the impression that it's being worked on), but it wasn't always particularly reliable or stable. The only real benefit it had (and arguably still has) is that it has massive traction.

[electrograv]

That's a very good point -- I've never had any issues with it, but I've also not used it very extensively, so my anecdotal experience doesn't count. If it's true that it's more broken and unreliable than alternates, then I'd agree with you.