I think you missed the part where I said that it can happen, but that it’s rare and hard to predict.
Yea, sorry, my wording wasn't the clearest. I meant to say that it is actually not that rare, and hoped that the linked source would help support that claim. From the same website:
We can [assume that] "all votes [are] equally likely except that the probabilities that A,B,C will be middle-ranked of the three in that vote are 30%, 30%, and 40% respectively" where C is the 3rd-party candidate. Then in IRV as #voters→∞, C's probability of winning is probably exponentially tiny so that Joe Voter is justified in assuming C only a very tiny [...] chance of winning. Indeed C only has a tiny chance of merely surviving the first round.
However, Joe reasons, if Joe and friends by honestly-ranking C top do manage to make C survive the first round, then that will almost certainly happen only at the cost of eliminating Joe's second-favorite candidate A. If the A votes then transfer equally to C and B (which in "1-dimensional politics" with C A B arranged along a "line" in that order, seems likely) then C will almost certainly still lose, and will have deprived A of victory in the process.
The idea then would be that the behavior of mid-ranking the 3rd party candidate would be self-reinforcing in IRV: an assumption of a slight bias that way like we just made (40% versus 30% [...]), then leads to it being strategically wise for Joe Voter to do it, leading to a larger bias that way, etc. – positive feedback, self-reinforcing 2-party domination.
Approval Voting is bad because of the simple fact that it doesn’t let you express any preference.
I agree and that's why I support Score Voting over it! The mechanism to express that one candidate is better than another one is to just give them honest scores! And there's studies proving that's the reality is, the vast majority of people are at least somewhat honest when filling out a Score ballot
Would this work like an open-source strava alternative?