> If 80% are good submissions it’s a useless filter and should be dropped.

How did you arrive at this conclusion? I can easily imagine that this process is so laborious that plenty of people don't even submit anything. Like the person who wrote about this on reddit. -- Now what reason would we have to think that the people who don't want to go through the process would also be 80% good? This may be a very valuable filter for Canonical!

I'm of the same mind. Interviews are not like school assignments, there's not some reason why the average person should "pass" most questions. Every question or step should provide clear information that lets you decide who to hire.

In practice it's not like this though, it's probably true that < 10% of what goes on in most interviews actually differentiates candidates.

I think part of it is it's just hard to figure out which 10% is the useful part. Though certainly if 80% are passing an early step in the process, it should just be scrapped altogether.

That wouldn't necessarily be true even if 100% of submissions were good, as you've completely discounted the cases where no submission is made. You can argue about whether the whole thing has any point or use, but a low % bad on submissions isn't on its own a sign that such a filter isn't useful.