The current issue is that most of the effort is done by humans right at the beginning of the process: CV screening & first calls.
If you're a recruiter, by the time you actually get to talk to a candidate, you're tired of screening and most of the time you do a half-assed call.
Plus it boggles my mind how very very few companies train the recruiters in understanding the technologies they have in their stack.
There are a lot of factors that add an incredible amount of friction and make it a very asymmetric process (as other have said above):
• plain human error • job specifications with boilerplate text not being clear on what the role actually does • often times the people doing the first screening don’t have a technical background • resumes come in all shapes and formats, so it’s difficult to efficiently read them
And because of that asymmetry and supply-demand ratio, it’s actually quite hard to hire software developers that are suited to a company, tech environment & business needs.
(disclaimer, I work at WorksHub)
The way we're trying to solve the issue at WorksHub is by trying to get companies to open-source same part of their code-base so that a developer could submit a few PRs during the interviewing process.
This solves the issue of irrelevant technical interviews, gives an idea of how your coding environment will be right off the bat.
Basically the thing is to do as much of the process as possible using only tech and then leave the last part for the actual human interaction.