I was more talking about "before EIP" where this was a common practice. Right now we don't have that problem because of the curve, since its not lucrative. And if the votes are big enough you can detect and downvote them easily. (Not so much when the votes are 1$ each).
Even vote circles, should be easily brought down with downvotes nowadays. Shouldn't it?
Not that easy unfortunately, as the risk of retaliation makes a lot of people stay away from the bigger ones.
Detecting bigger self-voters/circlejerks only works at amounts which are way over the threshold. Raising it can't be a solution either, as that would affect even more honest accounts. Imo it would be possible to set up analytical tools which can detect the kind of abuse we're trying to prevent. Yes, that would require more effort, but it wouldn't hurt all the small accounts in the same run.
In war terms, I think of it like the difference between a targeted strike on an enemy key figure, and carpet bombing the whole town. The latter is easier to set up, but at what cost?