In a general sense: you’re right that Duverger’s is not an iron law, and right that even under FPTP, party systems can undergo revolutions. I think that evolution is healthier, but there’s room for debate there.
In the specific case: Katy Levinson was suggesting that voting for a third party, from a safe state, was a good idea in the upcoming election. I have several responses.
As one of her arguments in support of this, she cited the Presidential Campaign Fund, available to parties which got more than 5% in the most recent election, with certain conditions attached (such as not fundraising separately). I pointed out that the effort to reward ratio of getting said funds was singularly poor.
I also think that, if you want to try to upset the party system without changing the voting system, you should have more of a plan than “let’s get a bit more money so we can make a bigger effort next time”. Every time you make an attempt like that, the possible direct electoral consequences are: if your efforts are weak, you make no difference at all; if they’re moderate, you end up hurting your own interests and helping the “greater evil” win; or, if they’re successful, you win the election, and a revolution in the party system. Of those three possibilities, history shows the third is the least likely.
What about the indirect consequences? Ideally, even if you lose, one of the parties would try to coopt your ideas, and actually implement them. In the current election, the Democratic platform has already gone an unprecedented distance to coopt Green/Sandernista ideas. If the consequence of that is that the Greens do particularly well, do you think that will strengthen the hand of the progressive wing of the party?
As for the Libertarians: yes, I agree, from that perspective the two current candidates really are uniquely bad, and the prospects of supplanting the Republicans as one of the major parties are uniquely good. But still, it seems pretty clear to me that in order to reach 50%, the Libertarians will have to form a coalition with somebody; there are not enough voters who buy the pure version of their ideals. I also think that voting reform would help foster the kind of open debate and negotiation that it would take to build such a coalition. So I guess I can see the argument for voting Libertarian from a safe state, but only in combination with aggressively pushing for voting reform.
