Companies like Microsoft want the best people. It's not brick laying, there's no "qualified for the job" tick box. The effects in software are non-linear, and one brilliant hire can create more value than 100 mediocre ones. These companies are actively hurt by having to hire second-best diversity hires, even if they're technically qualified for the job (whatever that means).
You're trying to argue axiomatically and I'm relating an empirical fact: Microsoft hires from elite engineering schools, and many of those hires wash out. It's not improbable that there are coding camp people who would perform well at Microsoft. I have seem people with similar backgrounds perform well in other elite engineering environments (cryptography engineering, kernel software security, to name two).
"Technically qualified for the job" isn't some ineffable abstraction. Most programming jobs at Microsoft are quite well defined, and qualifying people for them mostly means extracting solved problems from the work and presenting them uniformly to a pool of candidates. You don't need science to figure out how to do this, although if you want it, it was all worked out and written down in the 1950s.
What does best mean? That is very hard to quantify especially when the job is more than just LeetCode. It requires communicating with other teams, writing skills, general statistical thinking and data analysis skills (something not tested on leetcode), ability to receive feedback, ability to understand customer requirements, etc.
I do not mean to imply that "diversity" hires have those properties and non-diversity hires do not. However, I'd argue that if you don't make an effort to at least talk to everybody you can (phone screens), then you are going to miss a lot of people who are great.
