I've had both I've had ones that only tell you what the next interview looks like. Unless you specifically prior it out of the recruiter's hands.

The only way to test actual work is to do actual work, should be obvious but here we are.

It is obvious, but it’s also very time-consuming. You can’t solve ambiguous open ended problems in a set amount of time while being watched closely.

I'm not talking about being watched closely, I'm talking about regular work and then making a decision. You have to leave people space to do their thing if you want to see what they're capable of.

Sure, that would be even better. But how would that even look?

In the best case applicants needs to apply multiple companies. Companies need to interview multiple applicants and have a way to compare those applicants.

Those are the most basic constraints I can think of. How do you make that cost tens of hours for each round?

You would have to stop optimizing people to hell and back and start committing to a few at a time. Sounds like a really good idea to me.

That doesn’t answer the question.

For me as a job applicant even in the best case I would need to do 3 to 5 interview interviews. The same is true for companies in the best case it will take at least 3 to 5 interviews to find somebody. Are they supposed to have 3 to 5 temporary staff for weeks at a time?

How much time should that take per interview? How would somebody that currently has a job manage that kind of time commitment?

Yeah, you would need to change your expectations, I figured that much was obvious.

I felt like I was doing that with what I described.

What changes to expectations are you talking about?

You seem stuck on the idea that you need to verify an employee to hell and back before investing anything, that's not what I mean by committing. If they look like they might be a reasonable fit, give them a month to prove it, then you'll know for sure. One interview per person should be enough to make that decision.