Ultimately, it boils down to filtering by reputation. The data says that most jobs are acquired through existing connections before it is ever advertised to the public, but in those outlier cases, a degree gives a nod to you from a known credible source.
In the world of social networks and artificial intelligence, it is interesting that nothing has come up to provide better recommendations. Perhaps because it's not really needed in the software industry at this point (at least if you believe there is a shortage of talent), so it is not noticed as a problem by the programmers who are capable of making such a system a reality.
I've seen a poster around here who is involved with a startup (can't think of the name, sorry) that is working on a problem like that. Amalgating prospective hires and ranking them based on various assets of problem solving, contributions to Github, technical experience, etc.
I wouldn't be surprised to find this sector increasimgly revisited in years to come. I've recently come to experience that programmers without much experience, no matter how well-intentioned, sometimes just stomp around doing far more damage than good on projects.
In the world of social networks and artificial intelligence, it is interesting that nothing has come up to provide better recommendations. Perhaps because it's not really needed in the software industry at this point (at least if you believe there is a shortage of talent), so it is not noticed as a problem by the programmers who are capable of making such a system a reality.