Absolutely. So now that we've established the possibility, the question is, why exactly is it infeasible to use our existing means to identify spammers. And what do we have to do make it feasible?