US House elections are designed to have approximately equal populations per district. This is accomplished by a census every 10 years. Redistricting is tricky if it is controlled by them majority party because they can manipulate the boundaries in ways that maximize their number of majority seats. One technique is called packing. If you can pack an 80% district for your rival party voters, that makes the outcome guaranteed, while thinning rival voters in "your" district.
https://www.thebulwark.com/what-role-did-gerrymandering-play-in-giving-the-gop-its-house-majority
Redistricting to your favor is called gerrymandering.
This simplified example shows an extreme case. A 60% blue majority can win 5 of 5 seasons simply by distributing voter equally (middle). But a 40% minority party can win 60% of the seats by packing 2 100% rival voter districts.
https://fairvote.org/new_poll_everybody_hates_gerrymandering/
Gerrymandering is considered unethical but necessary in a race-to-the-bottom. If Republicans gerrymander a disproportionate majority in Texas, then Democrats will feel the need to gerrymander a disproportionate majority in California. It is better for voters if an independent court or commission is set up to redistrict, but there's still no "right" answer. Or the primary criteria would be "compactness", closer to a circle than a snake, and using fixed city or county boundaries.
One confusing thing is that courts have ruled that racial minorities "deserve" representation, so then you might intentionally gerrymander "snaking" districts that maximize blacks into a supermajority district, which helps them always pick their candidate, but dilutes their voters elsewhere, if they lead towards a given party (namely Democrat), hurting that party's competitiveness in other districts.
The natural solution to this is to expand representation beyond single winner districts. A Proportional Representation voting method allows a 4 seat election where any united group of 20% of voters can pick a winner, and then at least 80% of voters can support someone they like.
A good method is called Single-Transferable-Vote (Variant of Ranked Choice Voting). Voters rank choices. Lowest candidate are eliminated, and fraction of winning votes can transfer to a next choice. Flowchart!
STV sounds a bit complicated, and it is is a bit complicated to vote, but it is easy to vote. You just rank candidates. You can rank as deeply as you'd like, but you don't have to rank all. The cost of not ranking is if all your choices are eliminated, you don't get a next choice. or if one candidate wins with a surplus, you may lose your fractional surplus transfer vote.
https://en.wikipedia.org/wiki/Single_transferable_vote
@tomruen STV is not hard when votes are hand counted, but when you get to hundreds of thousands of votes being cast the reality is you need machine counting, but having a human readable paper ballot is also highly desirable. I have not seen an example of a paper ballot that a machine can read, allows candidate ranking and obvious and user friendly.
Oakland, for example, uses The Grid, which has the down side of not allowing a voter to rank more than 5 candidates (limited by the number of columns) and has shown itself to over voting ( accidentally ranking the same candidate as 1st and 5th choice or ranking 2 candidates 3rd. for example).
@antares Hard is always relative. Count is "hard" if you're error-prone and don't have good error-correcting systems in place.
And given Kari Lake just lost close election in Arizona and hinting at fraud against her, anything that makes elections more complicated has some risk.
@antares And STV does have a "randomness" with it, partly connected to a property called "nonmonotoicity" which basically means sometimes voting against your preferences can help you.
If you vote true preference A,B,C, you may find both A,B, losing and getting C, but if you (and others) voted B,A,C you might have A winning!
None of this is LIKELY, and you have no control, but it comes down to the idea that adding losing candidates can change winners due to sequential elimination order.
@tomruen There is a mathematical proof (Arrow's Impossibility Theorem) that for an election with more than two choices any Voting System will produce perverse incentives -- that is a voter with complete knowledge of all other votes might be better off casting a vote not reflecting their true preferences.
The good news is that some systems have much smaller regions where strategic voting is more optimal than true preference. STV is one of the systems with the smallest regions; so much so that in elections where the number of voters is much much larger than the number of candidates it becomes statistically unlikely.
@antares Agreed. People bring it up in RCV, single winner mostly, and worst there. But also exists in any runoff process.