O1 is also starting to get "gamed".
When you actually want to identify talent, it works the same way everywhere. You find other experts in the field you can trust and ask them to use their subjective judgment. Governments don't want to do that unless it's absolutely critical, as it goes against the prevailing theory of how government bureaucracy is supposed to work. Even if the government is willing to give outside experts the power to make the decisions, the experts probably don't want to accept the liability that comes with it. Which leaves the final decisions to government officials who rarely have the expertise to evaluate the claims on their own.