I believe the big problem people have with take-home work is trust. Nobody wants to do a take-home assignment that doesn't really impact their prospects of getting a job, or, worse, that has only downside for them. Hiring teams give assignments and then run a standard interview gauntlet. On both sides, most people believe it's the interviewers making the real decisions, without much consultation to the tests.
The candidate thinks, what's the point? It's a seller's market. I'll work somewhere that doesn't haze candidates.
We run work-sample challenges now, and (loudly) did so for about a decade at Matasano. We had essentially no complaints once we locked our process in. Part of the reason why, I think, is that we tried hard to be very clear about our process. But more importantly, the work-sample challenges made most of the decision. When we told candidates the time they spent on the challenge offset time they'd have to spend in person performing for an interviewer, we weren't lying. A 4-5 out of 5 average on our challenges more or less totally technically qualified a candidate; the odds were overwhelming that such a candidate would get an offer from us, and we told them so before they came in for interviews.
What I think I see now is hiring teams who want to have things both ways. They don't want to take a leap of faith with a new process, so they're still abdicating their responsibility to random subjective interviewers. But they want to look on-trend and competent, so they assign homework.
When you get on the phone with one of these teams and want to check them out, ask about the rubric for the challenge, and ask if they're going to do any on-site technical qualification after you do them (if they're giving challenges after the interviews instead of before, hang up the phone on them.) I'd be surprised if even half the companies that assign homework even have a rubric.