To me, Jon seems to have demonstrated exactly like the type of resourcefulness YC asks for in its applications [1].

Unfortunately there was no way for YC to predict the Jon had these chops at the time. The fact that YC can make such predictions with high-accuracy, from simply an application and a 10-minute conversation, is very impressive. There are bound to be borderline calls now and then.

If you have de-identified application data, it would be interesting to know what early patterns in applications predict future success.

[1] Jon hustled his way into YC, found a way to make the move when his co-founders weren't willing to, kept the company running and profitable meantime, and closed a large seed round with top-notch investors, in perhaps roughly the same timeframe that he would have with YC.