imagine a theoretical society comprised of 1/3rd left, center, and right factions, electing 300 representatives to the legislature. in a proportional representation form of government, 100 seats are awarded to each of these three factions.
now instead imagine we elect all the seats using single-winner districts, with a centroid-finding voting method such as score voting, star voting, or approval voting. there will surely be some diversity in the legislature, arising from the fact that various districts will lean further left or right, but on the whole, it’ll be a centrist-dominated legislature.
now, which of these arrangements will produce better government? and by “better”, i mean that the policies put into law — and ultimately the state of the world — will produce greater overall voter satisfaction.
advocates of proportional representation naively assume that proportional representation will lead to better results. but this generally flows from the fallacy of treating the legislature as and end unto itself, rather than merely a means to an end, that end being a particular state of the world. in engineering parlance, we would say that that pr advocates are focusing on an implementation detail rather than a goal.
while there are some reasons to suspect pr to be better, such as more powerfully mitigating gerrymandering, there are also reasons to believe the opposite. the all-centrist legislature would have the same ideological center as the proportional legislature, so there’s no particular reason to expect dramatically different legislation to ultimately pass. however, the more centrist body would tend to be comprised of more amicable and less divisive officials, working more harmoniously and efficiently.
there is also a theory that pr leads to more diverse perspectives being considered, but i’m skeptical of that. candidates running for office, aiming to be the most broadly appealing, would already have an incentive to consider multiple perspectives and find the most effective policies.
either way, we can only guess as to how it all shakes out. any claim that proportional representation is obviously superior is inherently fallacious, as it confuses ends with means.