>if Elon Musk comes out and says SpaceX considers a cert from $NetflixOfEducation as equal to an engineering degree from MIT on a resume... then it's all over.

If 50 developers apply for a job and 3 of them are from MIT and 47 are from lesser known schools, using the MIT signal as a filter is efficient. If $NetflixOfEducation is too accessible and half of the applicants have it, then it loses its value as a hiring signal. Of course there are other filters too. Previous employers is an important one. You can create comprehensive professional examinations like they do in medicine, where a passing grade is a major achievement. But these don't scale well because the more they scale the easier it is to cheat, especially if the risks of getting caught cheating are minor (i.e. there aren't previous barriers that must be overcome via investment of time in other institutions first).

The other dynamic of top schools is the networking effect. There is real value in concentrating thousands of the "best and brightest" together in one place and bouncing them off each other. It doesn't scale, which is both a problem (for qualified people who are excluded for whatever reason) and a benefit (for those who participate, there is value in the scarcity).

Only for companies who hire for signals.

For companies that hire for throughput this does not hold.