No, this is 100% battle tested. I can't drop numbers but we just did a study at work comparing referrals to non-referrals... its night and day. Referrals are out performing across the board. the only problem is that eventually you run out of referrals

I am aware of several startups that started this way, and have been involved in some. The quality of the results depends on who you seed the graph with, of course; but I’ve seen it work well.

As requested by the original poster, it doesn’t scale.

+1, referred candidates at my company perform way better in interviews and on the job compared to cold applicants