To the extent that the reward pool is just used to funnel coins back to people for holding SP it just becomes interest on SP. If people were proposing eliminating the reward pool and increasing the SP interest rate to some big number like 200% people would see that it's economically nonsensical, bigger numbers of coins but the value evaporates from the coin because there's no usage model besides "the number gets bigger". I haven't looked into how UPVU works, but it seems like it's basically a self-vote aggregation service -- instead of jumping through the hoops of creating ten self-votable posts per day and self-voting them you delegate to a big user who votes to give you the equivalent reward on a single post (except more, because the "convergent linear" rewards model punishes smaller votes). I am sympathetic to the problem of there not being enough easy-to-find good content to use legitimate upvotes for, but if the system is just largely self-votes then it doesn't work.

If we're pondering troubling possibilities related to overly-centralized witness vote control it might also be worth considering that Witness proxies also impact DAO proposal voting. Someone with enough witness votes can just start directly transferring lots of SBDs to themselves if they want.