The existence of reputation doesn't give rise to these bots. The bot operators could calculate their own "rep score" whether we provide one or not. More likely, they will have a far more sophisticated rep score that is a better predictor of post value.
Our rep system errors on the side of tolerance (or tries to). Theirs would aim for predictive power.
I mean it's a very easy implementation. Not everybody knows about predictive modeling while voting bots are freely available.
How do you see this holding up to bot owners simply voting their accounts in the positive again? Are there algorithms in place to prevent recurring votes from being as impactful as varied ones?
Yes, they already do develop own "rep score" , mostly it is just a list of users to vote right after if someone of the list made an upvote. There's much more ways to be creative on that field though.