In and by itself that observation is both correct and "just how things are" currently; it's a direct consequence of dPOS, which also has its advantages and disadvantages of course.
The "witness rank" component, via which an initializing "UA trust vector" is assigned to all Steem accounts, surely has its pitfalls -- I'm not denying that of course. On the one hand, I agree with you that indeed a few very large stakeholders (e.g. @freedom) have an immense influence on -a- the witness rankings and therewith also on -b- all UA scores. But on the other hand, the question begs: "is that therefore 'wrong', or a good indicator indeed?"
@freedom is massively "influential" on Steem, yet due to not openly blogging via the @freedom account (arguably one of the best example accounts on why UA functions better as an influence metric than "reputation"), it has no "votable objects" and conversely no "reputation earnings" either, but it does have a relatively high UA score (and rightfully so!). And therewith, I argue that it is justified that the witnesses @freedom / @pumpkin votes on should be "powered with" lots of "UA Trust".
So is the UA witness ranking component "perfect"? Nope. Is it "justified"? Yes, I think it is.
Right, it is indeed kind of an 'influence score' which need not to be bad. Nevertheless, people with much influence still can have a very bad reputation (think at Donald Trump for example ...). :-)