They didn't search every @gmail user, they searched what would be roughly equal to (in gmail's commercial case) a payment provider for gmail, and people loosely related to such a provider. It's a flawed comparison, obviously.