I think I see now that approval voting has a different dynamic even if all shareholders voted for many delegates of the same person that is operating many delegates (delegate-xeroc-1, delegate-xeroc-2, delegate-xeroc-3 are all controlled by xeroc).
Example:
One shareholder (called "X") has 10% of the stake and it is a bad shareholder. The rest is good stake.
In approval voting: X puts up 52 delegates and votes for them with his 10 %. Now if the other 90 % of the stake vote for 5 other people / delegate operators by voting for the 20 delegates of each of the 5 trusted individuals. Result: The bad actor is definitely out.
But if everyone in the 90% group only voted for one delegate of the 5 trusted people then X has all his 52 delegates in the top 101.
In delegation voting: X puts up 1 delegate and votes for him with his 10%. X's delegate is in with 10%. If pool sizes would be limited to say 1% then X would just put up 10 delegates he votes in and nothing changed. I left out the possibility for negatives votes but I guess that wouldn't have changed anything anyway: Everyone negative votes against X. But X would also negative vote against everyone else which would make it equal again. Suggestion: Negative votes could be limited to 5 negative votes which would not allow the bad actor to negative vote against everyone.
Conclusion: Approval voting makes the result more extreme, all depending on whether shareholders get it that they should vote with their 101 votes or not. A bad actor can be kicked out best with approval voting but might also get full control over the network if shareholders only vote for a few delegates (as opposed to delegates 1-x of the individuals they trust).
Solutions:
- RDPOS is a suggestion that would max out the voting capacity of everyone to 101 votes.
- Quasi-Delegation voting. I though of the following but that would equal delegation voting, as I see it: Not limit one delegate's "seats" to 1 in 101 seats and base his seats on the amount of approvals he got.
- Real delegation voting.