Initially you'd want to hire randomly to seed your 'gene' pool and then fire those who are unfit. This will provide feedback as to which measures best predict the outcomes the business wants to optimize for. And would take into account the fact that different businesses require different hiring practices and need to optimize for different outcomes.
This is just another idea for hiring that provides no information as to whether it's a better approach and completely ignores the rest of the employee life cycle.
The article talks about confirmation bias but provides no indication as to how hiring by interest changes anything about confirmation bias in the hiring process.
Seriously the product 'you all want to make' is a piece of shit that is essentially design by committee. There will be vocal people who DON'T want to make that product because by the nature of the process it has become bland and unimaginative.
My personal and completely unscientific approach to hiring is towards interesting people rather than people interested in the particular problem we're currently solving. I have no idea whether this is better / worse that credentials or hiring based on interest.