On the game theory why collusion attacks are not a security risk

If 2% of all people collude, they get bots 0.03% of all personhood tokens, that's 0.015x whatever a person-token provides, per person colluding. One of the benefits of personhood tokens could be basic income, for a 1000 dollar UBI that would be 15 dollar per person colluding. The bots who collude with humans need a much larger population of humans than they get bots, because of the randomization of pseudonym pairs, and for humans to collude, there are game theoretical rules, you need to pay them. I'm not sure that it is feasible economically to get people to attack their network for 15 dollars. If 10% collude, the total share per personhood token per person is 0.05x so slightly more, 50 dollars from a 1000 dollar basic income. If 25% collude, 3% of all personhood tokens, 0.12x a personhood token per person, 120 dollars per person, getting a bit more significant but 1/4th of a population is quite large. If 50% collude, 15% of all personhood tokens, 0.3x, 300 dollars, and at that point half the population is attacking.

The bots need something for themselves as well, if 50% of the profit from the attack goes to the colluding humans, that is even less, a 1000 dollar UBI gives 7.5 dollars per person in a 2% attack, and so on. I may be wrong, but getting 0.0075x less from attacking the network than you get from the network itself, does not really seem that big of an incentive.