Lots of good responses to this question. The most plausible solution being pushed right now is the one advocated by Lee Drutman and Fix Our House, proportional

Lots of good responses to this question. The most plausible solution being pushed right now is the one advocated by Lee Drutman and Fix Our House, proportional representation via large multi-member districts: https://www.fixourhouse.org/ https://www.fixourhouse.org/ It's more plausible than others because it doesn't require amending the U.S. Constitution. I'm not convinced you couldn't still gerrymander a bit to tilt this system one way or the other, but it's way more robust than what we have now. My ideal system has the same basic design, a combination of local and proportional representation, and would look like this: (1) Redistrict algorithmically using a simple algorithm based on population density alone, minimizing a relatively easy-to-understand objective function (e.g., average distance of people to other people in their district). The goal of this algorithm is not to be "best" or "fairest" in any sense, but to be robust to attacks like John Roberts's frightening line about "sociological gobbledygook." Once you have subjective measures of fairness, you open things up to political maneuvering (and accusations thereof). (2) Assign multiple members of Congress to each district. An election for a district involves picking n candidates via instant runoff voting. (3) Members don't get 1 vote. They get a vote proportional to the fraction of people that voted for them in the election. That way, if you have a 4-member district, with the top 4 vote-getters being a Democrat with 45% of the last-round vote, a Republican with 43%, a Libertarian with 7%, and a Green with 5%, they each get 0.45, 0.43, 0.07, and 0.05 of a vote, respectively. (4) Do this for the Senate too, ha ha ha. Senate still serves a purpose in this fantasy: it's a longer-time average of public opinion with less time spent running for re-election.