The filters that companies and recruiters use mostly trend in a meaningful direction. There are a higher percentage of good programmers among MIT grads than SUNY Potsdam grads.Imagine how mistaken that would sound if you substitute "startup founders" for "programmers."
The reason you feel company filters are anything except random noise is personal bias. It's disheartening that your article notices that company preferences are essentially random noise, and then you don't really internalize that fact or take it to its logical conclusion:
Interviews where programmers do anything except work-hire tests tuned to the signal that the companies want are doomed to fail.
And it's almost impossible to do a work-hire test in an interview. A real work-hire test. The type of test where you actually do some work and then show that yours is effective.
We're so backwards that people probably read this and think I'm saying whiteboard work or something stupid like that. No. Real, actual work. The problem you're working on can be fake, but the problem needs to be literally the same type of thing that the company does on a day-to-day basis.
Want a 100% success rate? Do that, and then don't pay attention to anything but the results. Including whether the candidate has an MIT degree, or any degree at all. If you actually set up a work-hire test, a real one, and then stop asking about all the other stuff that doesn't matter, then you will get a 100% rate. Again, you have to tune the test to what the company is looking for, but that's not hard. "Here is an iOS app. Find and fix these bugs, then add a new feature."