If we had vote ranking at the national level, like Minneapolis does, this wouldn't be an issue.
> Following the primary election in September, a poll on October 20 showed Humphrey leading 35% to Coleman (34%) and Ventura (21%)
That's pretty far behind. But I'm not going to research this because you're the one trying to make the point, and you need to defend your position that it's a useful one, not switch topics to alternative voting systems.
If people followed your suggested voting tactic then Humphrey or Coleman would have been governor. They didn't, and ended up with Ventura. Not only was he a plurality winner, but from what I can tell he was a Condorcet winner too.
The voters got who they wanted. Not what the polls said was the lesser of two evils as you would have had them choose.
I don't like making these kinds of calls, though, which is why I support vote ranking or some other way to make third parties more viable.
In the 2000 election, "Gore was leading Bush and third parties with 44-42-4 among registered voters" in late October according to one poll.
In Minn. it was Humphrey 35%, Coleman 34%, and Ventura 21%. You posited "Humphrey was way worse than Coleman". That's inverted compared to 2000. Assume that a Ventura supporter considered Coleman way worse than Humphrey.
In that case, what should the Ventura supporter do? I believe you would say to vote for Humphrey. By ignoring that advice, the Ventura supporters ended up with Ventura.
Yes, vote for Humphrey. If Ventura then loses by one vote, you still made the best choice with the info you had. 9 times out of 10, or whatever high percentage, the 3rd place poller would've lost even with your vote, in which case you did the best by voting for the lesser of the top 2 evils.
Perhaps Ventura won because Ventura supporters felt that Coleman and Humphrey were equally worse, in which case voting for Ventura wasn't risky.