That may be "obvious from the beginning" but it's far from clear to me that it's correct. Here are some reasons why. * The appeal of a political party to a given voter is not simply a matter of computing some measure of similarity between its principles and the voter's. Some of the other things it depends on -- e.g., how trustworthy the party's people seem, whether the party succeeds in arousing enthusiasm rather than mere consent, whether the party's statements are clear and vivid enough to get through to the voter -- mat well favour more extreme positions. * To win elections, a party must not only get voters on side but also get them out of their houses and into the polling stations on election day. Again, this is a matter of enthusiasm as well as consent, and may favour more extreme positions. * In many countries, political success is not a matter of a simple nation-wide majority vote. There are constituencies and electoral colleges and the like. This means that political success may depend on identifying particular spatially-correlated groups of people and appealing to them, and there is no guarantee that this looks anything like appealing to the nationwide median voter. * When there are more than two candidates, or more than two parties, you can win by appealing to a reasonably-sized minority, and their preferences may be some way away from the "centre". * When multiple issues are at play, you can't just arrange parties on a linear scale and ask where the centre is. Aiming for the centre on every issue may result in every voter finding your party mediocre and preferring another party that's extreme according to their highest-priority issue, and success may depend on finding a bunch of specific issues and adopting specific (perhaps "extreme") positions on them. * It may be worth noting that one of Cummings's claims is exactly that looking at everything on a left/right axis and aiming for the centre is a big mistake and misunderstands what issues people