If you read my comment, you see I don't believe these are mutually exclusive at all. There's no reason a voting machine could be made reasonably tamper proof. Or at the very least, air gapped and uncaring of the names of the candidate. It literally only needs to be a tabulating machine for punched cards, or something similar.

I really dislike instant runoff voting and similar systems like STV. They are better than FPTP, but not by much. One big issue is discussed here: https://www.youtube.com/watch?v=7Q7rzqJ0YS8 But my main concern is they might tend to elect more extreme candidates, because they quickly drop candidates that don't have enough first votes, even if they are everyone's second vote.

Ideal election system > paper count with representatives of the party > actually existing electronic voting.

The problems of actual existing electronic voting are a microcosm of bigger problems. We get insecure, ineffective voting machines because procurement is difficult and tends to be captured by lobbyists.

Condorcet superiority seems not all that important in practice: http://www.fairvote.org/why-the-condorcet-criterion-is-less-... , and if you have suitable multi-member constituencies for STV then you tend to end up with a couple of "extremists" and a large group of moderates in the middle.

Again, let's not let the best be the enemy of the good.

The article you linked to goes on and on about the exact issue I mentioned. IRV tends to select for extremist candidates. Condorcet methods tend to punish them. The article tries to defend this, but I see this as a very bad outcome. Extremist candidates can do much more harm than good. Centrist candidates are more rational and tend to be less likely to do crazy things. Condorcet methods ensure the winner will be the best possible compromise, IRV gives very inconsistent behavior but can favor extremists, but can also punish third parties (IRV still suffers from vote splitting in many cases, see video I posted.)