That's a very interesting discussion going on here, thanks for bringing it up @scipio!
I personally don't feel that we need to be able to explicitly disagree on a witness, since not voting is already a clear statement to not support that specific witness.
Reducing the witness votes to 7 (as it has been suggested here) seems to be the better idea to me.
With regards to the UA Score, IMO it's totally fine to consider witness rankings in the way it is done now. However, as the situation is right now, someone who's not the ability to set up a witness node, would never be able to achieve a top 100 ranking.
That is why I'd appreciate to see the algorithm evolving in a sense that other parameters are included as well. Currently, engagement from a UA standpoint is exclusively defined by number of followers. Apparently, one needs to engage with the community in order to grow one's followership. Yet, new users would always be dwarfed by those who joined the platform early and could start building their brand at a point in time when it was much easier to gain attention.
Here's an idea on how to improve the algo and make it more competitive/attractive:
Only consider active followers when calculating one's influence / UA score, being active those followers who've been using the BC at least once during the past 6 (4?) months e.g..
Then nobody can just sit on their followers but needs to stay engaged with the community in order to keep their score high. Like that, even new users were able to climb up the ladder.