Hacker News Reader

I haven’t seen any good data that shows that the current system is actually good at doing that.

Have seen metrics that show a good programmer from a bad one?

That show that the current interview system filters for “good programmers” as defined by the company’s performance system or whatnot.