On attack vectors in Pseudonym Pairs

To be verified in the pair, mutual verification needed. If you, as a government official, do not think the other lives up to your idea of a person (maybe it is a bot, or a lol cat?), then you just break your pair, and you are assigned instead to be verified by a full other pair, 2-on-1. Since you are a person, you'll be verified (just think of Pseudonym Pairs as any government, except you are also an employee and a government official, and usually if you have done nothing wrong, and to the extent the government is not biased, it should verify you), and, if the other was a bot, they will not be. In that situation, you cannot loose, the other cannot win, so, there is no possibility to attack that way.

This here is the only attack vector, and it is very minor, not significant. It is mentioned on https://panarchy.app/Proof-of-power.pdf

The colluders get colluders/population more than they get otherwise. If 10% of the entire human population collude and together attack the network, they get 1/10 more than base line, if 5% of the entire population attack the network, they get 1/20, 5% (0.05x personhood tokens per attacker, exactly how that is divided between the attackers, they get to decide for themselves. )