That is a bad idea. In fact the ideal number of witness votes should be unlimited and it only isn't because of an annoying exploit with backups (and it is indeed frustrating and perhaps worth reconsidering that the little 1/21 slot forces us to weaken the voting system and indeed the whole platform).
This may seem counterintuitive, but the idea of the approval voting method is to vote for all witnesses you believe are competent and not malicious. Those evaluated as such by the most stake are chosen. Smaller limits of votes leads to more gaming attempts to manipulate the list, and in doing so very likely makes the system less secure.
Right now our Top 20 witnesses are basically all being decided by just a couple of the largest stake holders
At the same time this is false (for example several of us don't have a vote from freedom yet we are still there), and also ultimately the way it is supposed to work. The people with the most at risk should have the most say, it should require a lot of stake to get any witness into the list, and also you don't want relatively small(er) amounts of stake to ever be able to elect any witness, and certainly not multiple witnesses (see above re. "less secure").
In terms of the biggest stakeholders having huge influence, I think the bigger concern might be people not voting at all, and a lot of stake sitting on exchanges in liquid non-voting form. Because when it comes down to it the biggest stakeholders aren't even that big (freedom owns about 3%!), so their relatively large influence over witnesses is an indication of not that much stake actually voting. I really don't know how to do anything about that.
RE: Our Plan for Onboarding the Masses