That would seem premised on the idea that it is better to have someone barely capable in the position than to go without. In my own personal experience I'd rather be short handed than work with someone incompetent.
Your outlook is only correct if jerking applicants around with meaningless trivia questions and irrelevant tests actually, somehow, probably accidentally, results in the selection of world class developers.
If there were 20 plumbers and only 10 plumber jobs, I could get away with jerking them around forcing them to do crossword puzzles before hiring them to fix my toilet. That doesn't mean the selection pressure would magically result in the guy most likely to fix my leaky toilet; much more likely it would result in the guy best at doing crossword puzzles. Unfortunately I now have a concrete deliverable of a leaky toilet to fix, and the best crossword puzzle solver in the market trying to fix it. That isn't going to end well. Especially when the best leaky toilet fixer in the world knows he can refuse the crossword puzzle test and get a job via his rep or connections so he never even applies. So its not just a random selection of applicants, but a random selection of an inferior population.
That would indeed be a terribly way to hire plumbers! But happily, that is not anything like what the article is suggesting.
The article suggests (1) using a simple but relevant exercise and (2) using it as a screen, to select people for a real interview.
The analogy with plumbers would be something like asking them to list the tools they would need to replace a hot water cylinder. For someone who knows their stuff, it's trivial. But it will weed out the guys who are bullshitting you. The guys who can give a good answer, you ask to come in for a proper interview.
I understand where you are coming from but personally I believe that asking a programmer to produce a small amount of code is certainly relevant.
Don't get me wrong, it would be possible to pick nearly worthless questions which prove nothing about a candidate but a well designed problem requiring a few dozen lines of code and no 'trick' or specific knowledge can tell you more about an applicant than many hours of trying to code on a white-board or doing silly logic puzzles.
If there were 20 plumbers and only 10 plumber jobs, I could get away with jerking them around forcing them to do crossword puzzles before hiring them to fix my toilet. That doesn't mean the selection pressure would magically result in the guy most likely to fix my leaky toilet; much more likely it would result in the guy best at doing crossword puzzles. Unfortunately I now have a concrete deliverable of a leaky toilet to fix, and the best crossword puzzle solver in the market trying to fix it. That isn't going to end well. Especially when the best leaky toilet fixer in the world knows he can refuse the crossword puzzle test and get a job via his rep or connections so he never even applies. So its not just a random selection of applicants, but a random selection of an inferior population.