If you’ve only ran into 2 of them in your entire career it seems like there we’re probably better off trusting people’s verified experience than wasting millions of hours per year trying to a filter out the tiny fraction of people capable of pulling that off.

And if you did happen to hire one of those people, it should be pretty easy to figure out that they aren’t producing any actual output fairly quickly. Then you fire them.

Now instead of wasting 1 hour on every one who applies, you've wasted 40 hours testing them each for a week?

I’m suggesting you conduct interviews for experienced candidates like virtually every other industry and trust that if they have 10 years of verifiable employment experience as a programmer that they aren’t complete charlatans.

If 1/100 people are (and in my experience and the person I was replying to it is significantly less than that) you’ve wasted less than a half hour per candidate.

You’re not wasting 40 hours on the 99% who aren’t frauds.

Oh, if we're talking about 10+ years, then I definitely want to test that they can code, and make sure I get a good read on their attitude. You get a lot of "architect" types who think they are too good for coding now and can hand down their glorious wisdom from above, when in reality they were kind of mediocre engineers to begin with, are now even more divorced from the reality of their decisions, and come up with worse designs than if you had just left it to some juniors to figure out themselves.

But yes, for senior candidates, I'm doing a lot less coding and a lot more design and architecture, because that should be what they can offer differentiated from their less experienced peers. And I'm also going to be looking for their interviewing skills and philosophy, because they will probably end up doing more of the interviewing and thus team-building, not to mention the culture they set by all the juniors looking up to them.