This assumes NO support from any other stake, which is unlikely and unrealistic, especially after any degree of marketing, campaigning, offering incentives, etc. In reality, a dominant stake would still be to override, say an 80-20 split of the rest of voters against what they want to do, just not 100%. That's a little better from that perspective and in that particular scenario, but nothing close to a panacea, and this also significantly weakens the chain because any malicious minority attacker can still push in a minority of witnesses who can cause harm in various ways, instead of being kept out by broad stake consensus as the current rules are intended to do.

I think people are being a little selective in thinking here. At the moment, "we" are in the minority therefore giving a minority more influence is "good" . But there are also plenty of situations when "we" won't be in the minority and giving a minority more influence would be bad. Once you put rules into place you can't pick and choose how they will be used.

Also, I think it might make sense to break out the policy voting and block producer voting roles (as I believe was the case in Bitshares, and was dropped from Steem without much explanation other than "simplfiication"). In the case of voting on forks and policy, I agree there benefit in not "amplifying" a majority and giving smaller voting blocs a seat at the table (though this still allows a malicious minority to block policy decisions i.e. hard forks which may in fact be beneficial), but in the case of block production, this really isn't true. Minority influence on block production is pretty much only bad. It can't create a longest chain and can only cause trouble (including outright failure).