That's a silly way to think about the issue. Of course, the error in every measurement is compounded by errors in any related measurement, but these aren't really measurements we're talking about; they're opinions. There is no error in opinions.

It would be crazy to judge someone based solely on one factor of their entire existence. Sure, the more metrics you have, the more complex the matching is going to be; but then again, these people have made their lives out of looking at success rates and making connections (I would imagine not solely based on similarity). Theoretically, you might be right; but theory is oftentimes hard to match to practical reality.