The IBM anecdote in the comments captures something important. The old system had a lot of problems, but it did one thing well: it forced a much faster path from "can this person do the work" to actually finding out. You walked in, they gave you an aptitude test, and the test result was the decision.

We've somehow ended up with a hiring process that's simultaneously more information-rich and worse at answering the basic question. You can see someone's entire LinkedIn history, their GitHub, their blog posts, their conference talks, and still have no idea whether they'll be effective in the role. The signal-to-noise ratio went the wrong direction.