It's a good joke, but obviously there are benefits to having "your" representative. Who do you go to in a multi member state with say 10 reps. Which one is yours? Do you just contact all 10?
If so now all 10 are getting everyone in the state. That's kind of hard to handle, and really not efficient.
Excellent point. Proportional representation solves (maybe) *one* problem, that of fair partisan balance, while throwing away the benefits of having a one to one relationship between officeholder and constituent.
Look at Germany and New Zealand for examples of mixed-member proportional systems that have both geographic representation and party lists to achieve proportionality.
All "likes" should be aggregated and then distributed proportionately across all comments based on an algorithm designed by me to make sure the allocation is entirely fair.
You seem unclear on what the underlying logic of my algorithm would be, or how the word "proportionately" is meant in my definition, and what results it would produce.
Well, yes, because you *left* it rather unclear. But you said it would be "entirely fair," and since I know that you are an entirely fair person I assume that you would make sure that the distribution would begin by allocating at least 1 like to every commenter, and then distribute the balance in proportion to...some other metric (e.g. whether the commenter's initials were "M.R."?)
Of course, Congress could repeal that. But it never will, because the status quo is advantageous for one side (the Republicans) while you'd need bipartisan support to change the law.
And it wouldn't be enough just to legalize multimember districts - even if you did that it wouldn't make sense for blue states to implement proportional representation unless red states also did it, which they wouldn't. You'd have to get Congress to not only legalize multimember districts but mandate them.
I think this is a feature not a bug: proportional representation (unlike electoral college or senate reform, but like prohibiting partisan gerrymandering) wouldn’t require a constitutional amendment, just a simple act of congress.
There is a certain amount of rotisserie politics on Slow Boring. This is a fun distraction from the death of democracy, but won't get implemented. My personal preference for this kind of post is high speed rail maps and (alt) history (lessons). For me, this post is a little to close to the depressing political reality, while only offering a solution in a fantasy world where we have brave politicians who focus on the greater good at the expense of their party's and own self-interests.
Yeah. This is like watching people in a burning house fantasize about the great sprinkler system they were planning to install. I’m all in favor of policy posts that have a long-term flair, but this blog increasingly seems increasingly like a missive from some parallel universe.
You could expand the house and then have a rule that the number of candidates per district is the number of representatives allotted to the smallest state.
Why is the status quo advantageous for the Republicans? The distribution of House seats by party tends generally to reflect the overall national House votes by party. I don't think either side has an advantage in the present system.
The issue isn't a gap between national vote share and representation; it's between state vote share and representation. Illinois Republicans are barely represented in the House and Arkansas Democrats have none.
In terms of the national vote, over the last 10 elections Republicans won, on average, 1.2% more seats than expected by the national house vote and they've won more seats than the national house vote in 7 of 10 elections.* There are likely two reasons:
1) The republican structural advantage lets them control more state legislatures and draw more maps.
2) Small states with few districts are more likely to lean right. Out of the 21 states with 4 or fewer districts, 13 are right-leaning 7 are left-leaning, and one is purple. Of the 46 seats in these states, republicans won 70% in both 2020 and 2024.
Whether a 1.2% bias is an problem to solve or not is debatable, but there is a general republican advantage that sometimes is enough to swing control of the House.
* I calculated the gap based on a two-party election. The two parties received 97% of the house vote in the last 10 elections, but captured all house seats in all but two and 434 of 435 in the other two.
>over the last 10 elections Republicans won, on average, 1.2% more seats than expected by the national house vote
Your 'expectations' are wrong. Single member districts are not proportional representation. If you get say 47% of the national vote in 435 separate contests, there is almost no way to translate that into 47% of the legislature. If anything, it's a miracle that the US system gets anywhere near so close.
PR systems don't necessarily translate exact vote % into exact seat %, for boring logistical reasons including thresholds and which exact algorithm is used.
Your two reasons give me, a former political scientist, a headache :) Just think about how hundreds of separate contests play out. If you run up your score in a district that you win, the extra votes are all 'wasted'. It's inevitable that the parties will run up the score in any number of seats. Sit down and make a 10 district model in Excel, and play around with the vote totals, and I think it'll start to make sense
I was just replying to the original comment that suggested there's no problem due to the similarity in national vote share and house allocation.
I agree with all your points and don't think it's a reasonable expectation that there is a match.
I do think that a crude degree of proportionality should be expected, such as Arkansas having a reasonable chance of electing a democrat to the house from time to time.
But, seeing how close the national vote share and the house composition is made me less favorable to Matt's point. Even if proportional representation would improve the proportionality of state results, the degradation of accountability of reps to their cities is a big downside.
The balance between seats won and distribution of the vote at the state level is a secondary issue. Which party controls the House of Representatives is the primary issue. It can make a world of difference, as we can see right now.
Now it is possible that the Republicans have a few more seats in their majority than they might otherwise, but to win *any* majority in the House -- which is what the parties strive for -- almost always requires one party to win a majority of the national two-party vote. (2012 is the only exception I'm aware of.) Having extra seats in your majority is nice, but seeing how the parties can pass their highest priority legislation even with paper-thin majorities, it's not that crucial.
If the only thing you care about is who wins the house and the breakdown of seats matches the national vote share, than you're right. But, obviously, people might weigh factors differently. I personally think that the state issue and direct representation is more important than making every single election perfectly match the national house results. And this perspective explains Matt's article, which was the intent of my comment.
If the only thing you care about is who wins the house, then the national bias also matters. Proportional representation would also improve this. It would make it less likely that a party would lose the national house vote and still win the house, which happened in 2012 and 2016, 20% of the last 10 elections (and also happened in 2004).
I think the advantage comes from having more opportunities to gerrymander (fewer GOP states with independent redistricting) not from anything inherent in district based elections
I'm the captive of a one-person bubble, but the fact that proportional representation is never mentioned by anyone in American politics is truly mind-boggling.
What I continue to be frustrated by is why some municipality doesn't try it. The federal and state hurdles are well known, but those usually don't exist on the municipal level. Especially in municipalities whose constituency are more populous than some states.
Greetings from Cambridge, MA, which elects its city council and school board by PR, and has since 1941. A bunch of cities did around that era, and since then have mostly switched back.
In theory, any of the city councilors; they're equally responsive (or not...) across the city. There are some ideological groupings at times, where councilor X is known as the one about a particular neighborhood or cause, so you might pick them, or you might pick the one you remember meeting at a campaign event, or so on.
Not much different from talking to an at-large member of a more conventional city council instead of the one from your district or ward.
Yeah, I have to reiterate Nathan's point, I've actually had more luck getting a response on certain issues I care about from a city councillor that doesn't geographically represent me than I have from my local councillor. This is in Seattle, where council seats are largely defined by geography, with a couple at large seats that represent the entire city, which is why I copied the entire council on my complaint and how the councilman in the district over came to respond to me.
Unfortunately, some time in the middle 20th century, multimember districts became synonymous with racial exclusion, because many cities had separate seats that all shared a district, in order that the white majority would win every election. (Some cities like Austin had a gentleman’s agreement to let black people run in one of the seats and Hispanic people run in another seat, so that it could be unified partisan control without racial uniformity drawing attention from the law.) As a result, a lot of remedies for racial exclusion took the form of “best practices” advocating single member districts, taking advantage of racial segregation to ensure racial representation.
It would be interesting for some smart PhD candidate to do a thesis on why there is so little experimentation among the fifty state laboratories of democracy.
While PR is vastly more important, I think it should be paired with the cube root method for House expansion. Small states (I live in Vermont) should have a minimum of two representatives.
House size would be determined by the cube root of the population of the United States. IIRC, that would put the country around 697-698 members as of now.
If you draw a graph of all the (democratic) national assemblies in the world by population size and number of representatives, then a cube root turns out to be close to the line-of-best-fit (and is certainly the simplest formula anywhere near that line).
But what is the principled reason for this? I can see 150 as a reasonable upper bound for a legislative body on the basis of Dunbar’s Number, but other than the observed empirical trend, what is the *reason* to favor a number of representatives that grows as the cube root of population? Heck, at least logarithmic growth would seem to match population dynamics (albeit I believe strict choice of appropriate base would be birthrate-dependent, so let’s just use e.)
I haven't done the math on this, but if countries that don't use the cube root method and are below that line of best fit (like the US) then adopt that method and increase the size of their legislature, will the cube root method still be the best and simplest formula?
Doesn't this imply that at population of 999,999,999, you should have 999.999999667 representatives, and at a population of 1,000,000,000 you should have 177.83?
"During the drafting of the Constitution, George Washington played a key role in determining the size of the House of Representatives. Initially, the delegates at the Constitutional Convention proposed a ratio of one representative for every 40,000 people. However, at Washington's suggestion, this was changed to one representative for every 30,000 people. This was the only time Washington offered an opinion on the specifics of the debates during the convention."
I don't see how this doesn't relegate power even more to leadership in the house. Backbenchers now are able to accomplish little. With 11k representatives, they could just come to Washington, vote for the leadership, then go home. Or just skip coming to Washington and vote for leadership and then go home.
With that many representative you introduce hierarchy, with the 11k house electing its own representatives (possibly iteratively) until you get to a manageable size.
Ha, I suppose that's true, but yes, what I mean is executive branch notice and comment rule making ought to be be done in house committees and subject to congressional vote.
I'm more probably more sympathetic to the "representatives should be primarily accountable to the voters of a particular, coherent, geographic locality" argument than most, which is why I support vastly increasing the size of the house. Small, geographically compact districts are good for making reps directly accessible to the precise people that vote for them and less beholden to broader, nationalized party politics. The downside of PR systems is the homogenizing effect which probably also increases polarization and maybe even the influence of "the groups".
The actual thing that I get more aggravated by with gerrymandering than the proportionality is precisely the incumbent protection aspect Matt describes. That NY map that's arguably more gerrymandered but has far fewer safe incumbent seats would be a way better situation than the current status quo.
The problem is we are polarized. In a less polarized system you can focus on creating contiguous districts and local service for constituents. But in 2025, the Congressman is just going to go to Congress and vote party line down the line anyway, so local representation is dead.
Besides voting though, there’s a whole portfolio of constituent services that would very likely improve with smaller districts and more responsive reps. That seems like a real benefit.
Yea I think this gets it perfectly correct. Before maximally nationalized politics, it made a ton of sense. But there's probably no putting that toothpaste fully back in the tube, and devoid of that, PR would be much better, and in tow with that, a cube root scaled house.
That being said, it does in theory mess with my preferred non amendment requiring Electoral College reform, which is that all states should adopt Maine Nebraska style district based Electoral Vote allocation. However maybe it could be even more efficiently done. They could all be allocated proportionally. State has 10 reps and you won 60%? You get 8 Votes, (6 +2 Senators). Suddenly becomes much harder to have PV EV splits.
I hear what you're saying, but considering the patchwork manipulation of the system, I imagine it would be a net good for the minority party in each state to have greater representation. Sure you lose specific local reps, but you have less tyranny of the minority and the need to account for more than just your base.
If you don’t do literal party list proportional representation, but instead do something like single transferable vote, you can get the benefits of both. People wouldn’t have to list all democrats in a tie above all republicans, or vice versa, but they could rank regional candidates above others if they care strongly about regional representation (or could rank YIMBY candidates above others, or whatever other issue they care more about). Regions that care strongly about regional representation would have it.
Yeah, I've heard a different arrangement where the House is as specified (though I'd be open to expansion) and the Senate goes from 2 Senators per state (which makes no sense at all) to PR (and this could be at either the state or national level). So you can yell at your local rep in the House and in the Senate you angrily vote for the Green or Tea Party and now the Dems and GOP both agree that they don't want PR...
What is a state? It's the people living there, right? Why would 40 million Californians have the same amount of representation in the Senate as 500,000 Wyoming residents (Wyomins?)? And no California Republicans (about as many voted for Trump as Texas Republicans) represented at all.
Ah that's because I don't believe that all states require equal representation. "All states are created equal" isn't something I've ever heard anyone say. If Congress is supposed to legislate on behalf of the American people, then the people should have roughly equal say. Giving a Wyomin 80x the voting power in the Senate is not roughly equal.
States are not mere administrative regions—they are separate sovereigns. Equal representation in the Senate was, 1) a necessary compromise to get all of the original states to agree to form a union, and, 2) a way to ensure that the large states can’t steamroll the small states.
I suppose you must disagree with one or both of these facts if you don’t believe states are equal.
I agree that proportional representation is likely a better way for the makeup of Congress to reflect public opinion.
What I don’t really understand, though, is the flavor of procedure discourse that goes like: “X disadvantages my party right now. You know what would be great? Getting rid of X!” (Without any plausible argument about why the other side would unilaterally agree to disarm.)
1. It reminds me a bit of when conservatives say academia should hire more conservatives. Ok? Why? Maybe it would be better, but there’s no serious account of how colleges would be incentivized to do this. Or how it could be done without hiring lower-quality researchers. It’s just begging.
2. This article kind of implies that Democrats don’t quite act like they’re at war. They’re sitting around trying to make safe seats while Republicans try to win. This is a broader issue where Democrats kind of bend rules, so no one thinks they’re really acting in a principled way, but when they do bend rules, they don’t do so with the intent of beating Republicans.
3. For the various procedural questions that are often debated (e.g., court packing) I think it would be much more interesting to know how pundits stand on changing the rules several years out (so that we don’t know which party will be at an advantage when the rule changes).
“This is a broader issue where Democrats kind of bend rules, so no one thinks they’re really acting in a principled way, but when they do bend rules, they don’t do so with the intent of beating Republicans”
Yes, agree. In a system in which grinding partisan polarization is the one constant, I have to have at least a vague concept of how to get there politically before I can even entertain the thought experiment. It's like imagining the future of energy by starting with "first, assume we've invented cheap fusion reactors," except cheap fusion feels more realistic than getting Rs and Ds to cooperate on systemic procedural reform. Not that I begrudge Yglesias for indulging in some fantasy; he spends more time dealing with the limitations of the real world than most, so I suppose he's earned it. But still, I want to hear the theoretical road map here.
It's easy to incentivize universities to hire more conservatives. Just threaten to take away the billions of federal dollars they use to find cures for cancer.
1. I think not discriminating against conservatives would be sufficient. Also, for universities where they are still actually trying to find truth, you tend not to find it if everybody is already in agreement and there are no dissenting opinions.
This is mostly impishness on my part, but what if conservatives have a population-level deficit of academic ability? Is it fine if the discrimination-free equilibrium ends up being 80/20 or 90/10 toward the left anyway?
You certainly don't want to hire anyone that can't do the job. And this is probably very much position specific. But if you have two candidates that are close for say a sociology position, maybe there's a benefit to having someone that approaches the subject from a different perspective.
But again, I don't think you need any type of affirmative action program for conservatives, just stop the discrimination.
Would proportional representation reduce polarization, or increase it? For the dems specifically, I could imagine a world where it increases the power of the Groups: in a big state where people aren’t voting for a specific person I could imagine voters just lazily going with whoever gets the right endorsements, which I think plays right to the Groups. Additionally, if you can get—in effect—one Dem safe seat in a state like Arkansas, then would that person have any incentive to be a moderate, or would they be a standard-issue progressive of the sort that could never win a senate seat?
With proportional representation the parties would most likely split into a Progressive party and a Liberal one. They could form alliances in Congress as needed depending on the issues.
There are proportional representation methods (most notably STV) where voters specifically choose candidates.
STV in turn probably requires something like 3 or 4 representatives per district (since the messaging from groups would be to rank particular ways with a constraint that the groups have to basically guess how much support they actually have when deciding how many candidates to include in their recommended rank).
Proportional representation would not, in fact, do this. Republicans won the house popular vote in 2024, 2022, 2016, 2014, and 2010. It probably would've helped Democrats (we'd have won in 2012, which isn't nothing!) but it wouldn't be a permanent D majority.
Of course it would increase polarization. To the extent that providing constituent services is non-partisan (i.e., close to 100%) and to the extent that doing so makes voters more inclined to support you (almost certainly pr>0), then removing that tie to voters and basically having them vote by party label would make the system more polarized.
I don't think it's quite so clear. There's probably a u-shaped curve of polarization based on district size. The smaller the district, the less likely that voters are aware of candidates and will vote based on heuristics like name ID, party slates, endorsements, etc. When districts get bigger and/or more competitive, there is a greater incentive to reach voters through campaigning and advertising, increasing awareness. On the other hand, the bigger the district, the more likely that national issues will trump local concerns, leading to party-line voting, as happens with most senate seats today. These have countervailing effects, so there is a sweet spot somewhere in the middle.
The case against strict PR in the American system of government as it is practiced is that representatives often are intentionally the single point of contact for escalated government services. For individuals, this is everything from Capitol tours to flag requests to help when Federal agencies ignore you or requests for “private” bills for immigration purposes.
Under PR, when 2 or 3 or 9 representatives are accountable to you nobody is accountable to you.
I like the Germany/New Zealand approach of voting for a member with a clearly defined constituency as well as voting for a party to top off any national imbalances caused by local level concerns.
The idea I had during the 2020 reapportionment to achieve MMP that would also sate the "expand the House!" people would be to expand it to, say, 601 members, but the additional 166 members (27.6% of the new total) would all be chosen by PR. This protects current incumbents from having to deal with PR, and can also protect any incumbents from being gerrymandered out in states that lose population. Then, in 2030 you change the geographic/PR ratio to 35%, and in 2040, change it to 50/50.
This what was I was going to bring up. As someone that once worked in constitutent services for my local Congressman, having a point person the federal government was kind of a big deal to some people! Is it not anymore? I feel like any time I hear my Congressperson now, she talks a lot about the local projects she is doing. Our district (urban/suburban) is very unique to the rest of the state (rural), and has unique needs that she addresses, that I'm not sure a state-wide elected Rep would also address.
Also, I think town halls are a really good thing, even if they're not just intended to scare the bejeezus out of Republican House members. Would PR render those extinct?
The idea that Congresspeople provide individualized constituent services seems poorly advertised and not something that I would expect to work practically in view of how many constituents each Congressperson has. Potentially this understanding is wrong! But it's also the obvious default assumption.
Proportional representation might have a good outcome when it comes to partisan control of congress, but that's not the only political outcome that would be affected. Going back maybe 20 years, Persson and Tabellini found that, relative to majoritarian elections, PR increases taxation and welfare spending but also decreases accountability, which results in greater corruption. Non-local officials are also less likely to care about things like closing local hospitals or factories. That may be an ok tradeoff, but it's still a trade off. There's probably other research on this I'm not aware of.
The benefit of the current system is supposedly the increase in localism but that seems like it's long gone, making the system reflect current political realities is i think much more important than some ideal where everyone will know every politician individually and vote on local issues.
And when you draw districts the way we do now, there's not much 'local' anyway.
It wouldn't prevent town hall type situations, you can always as a rep go somewhere in your state and let the constituents yak.
I know there's a romanticism behind the idea that house reps will think about local concerns but honestly my experience is opposite. Most current reps are basically soldiers for national parties and that's that and that's what voters have largely preferred in action, if not words.
The representatives felt pressured enough not to go, so they got the sense that voters _cared_ about these things, which surely affects their future decision making at least a little.
Neither would I, the point is that even the decision not to go affects them. I'm positing that if it's PR, that decision no longer exists so the effect doesn't exist.
The existing system does have 'localist' benefits. Of course these would be enhanced if the House was substantially expanded, you're not going to get all that much in the way of 'local' benefits from 700,000-800,000-pop. districts.
I mean that national matters are prioritised is true in countries that have much smaller populations per member as well. But the 'benefits' do not only come at election time; in the UK, MPs end up adopting a role that is essentially 'social worker of last resort', dealing with individual constituents' problems dealing with the bureaucracy. The extent to which this is *actually* a 'benefit' or not is (very) debateable but it's something that's genuinely possible when your constituency (district) has a population of about 70-80k, ie 10 times smaller than a House district.
Yes, that 'from White House to shitehouse' line - in addition to being a good joke - is very much how a lot of British legislators (especially the more ambitious ones) feel about these 'localist benefits', but they mostly lack the imagination to see how they could change things.
Also worth noting that Republicans find Democrats' anti-gerrymandering stance absurd because Democrats are very pro-gerrymandering when it comes to majority-minority districts.
That’s half the story. In the 1980s when the VRA was amended to require minority-opportunity districts the RNC chair was a big fan. Drawing new black majority seats in the South helped Republicans beat a lot of white Democrats and they wouldn’t have flipped the House under Gingrich without it. If VRA districts were no longer required Democrats could draw much more efficient maps in southern states they control.
Yeah the racial gerrymandering thing has a different valence when it was more about White Democrats vs Black Democrats in the South. Now that a state like Mississippi is almost perfectly racially polarized, it's just pure partisanship.
It’s not gerrymandering to ensure that communities of interest that have long thin regions of habitation get represented. A district that follows the banks of the Mississippi, or that gets the core of several segregated cities, is much more natural than a square district based on nothing meaningful.
I think that again comes down to how you define gerrymandering. For example, in Louisiana, the districts are not compact and the borders look funny. However, the representative is proportional to voting: Republicans got 65% of the votes and 66% of the seats.
The gerrymandering by both sides tends to cancel each other out and on the whole the House reflects the national will of the people.
What I think is bad is tossing the norm to only do redistricting once, after the census. Otherwise, it's just a chaotic free for all. So I'm happy to see Democrats pushing back against Texas and the Republicans in this instance.
1. It's a radical solution that would make titanic changes to the political order the US has had for 250ish years. It's a little a strange for an essayist whose whole schtick is 'radical change is probably bad, let's do incremental change instead' to advocate for. In terms of radicalism I'd rank it slightly ahead of 'outlawing private health insurance and creating an American single payer system overnight'.
If you wanted to try this out first in like a small state or county government, I mean, sure.
2. Mixing PR & a separately elected President famously doesn't work very well- proven not just throughout Latin America but also when Israel briefly tried a similar system to this for a few years. That's why literally not one single rich country structures their political system this way. This is doubly true with bicameralism and 2 year terms. The House is made up of Parties ABC, the Senate of parties CDE, the President from party B. To get any legislation passed, you have to navigate cross-cutting alliances of 6+ separately elected parties, all responsible to a different small voting/more ideologically intense voting base. Then, these coalitions turn over every 2 years! The House and the Senate would be in a constant churn of parties engaging with each other, and then with the opposite house- overturned every other year.
What's happened in dozens of Latin American countries over 150+ years of trying this is that it's too difficult to get *anything* passed this way. Too much squabbling, too many separate constituencies. Consequently the President gains more & more power, just in an attempt to get something- anything- passed. This way lies Caesar!
(If you don't know the difference between a PR parliament & a PR presidency, please do not respond so as to not raise my blood pressure. Thanks).
3. The US is already veering towards becoming more & more like a Latin American state. "Let's adopt the political system of Brazil, it's going to totally work out great this time, I promise" is not very convincing! Again, not one single rich country mixes PR & a President!
"If you wanted to try this out first in like a small state or county government, I mean, sure."
Yes, this is one of the things that I find tiresome about electoral reformers -- there are a bunch of states where the state constitution can be amended by voter-initiated amendments (and in some of those states even just a bare majority of the popular vote is enough to win), but it's very rare to have any sort of electoral reform proposed via ballot measure. Make California a parliamentary system or get Illinois to adopt proportional representation for the state legislature and you'll likely be able to get more of a hearing about making these changes at a national level. (Cf. giving women the right to vote -- about half of states adopted that on their own before the 19th Amendment was passed and having those examples almost certainly made passage and ratification of the amendment easier.)
The US has had the same 2 parties since the 1850s, which is older than most democracies today. I think it's weird that the writer who's against radical change thinks we should switch our entire electoral system, at the national level, overnight! It's not something that can be done incrementally or in stages
The problem with this is that PR is unilateral disarmament for any state with a partisan skew. And in practice, since Republicans aren't interested in voting system reform, this would mean asking a state like Massachusetts to adopt PR and hand over more seats to Republicans while knowing that a Republican-controlled state like Wisconsin is never going to return the favor. The game theory of piecemeal PR doesn't make sense.
For state level office voters care much less about partisan affiliation. Not zero (good luck electing a Republican mayor in NYC these days), but also you would never get Charlie Baker to represent MA at the national level.
That's why I explicitly suggested "Illinois to adopt proportional representation ***for the state legislature***" (emphasis added). Additionally, states almost certainly could not adopt proportional representation for U.S. House seats without amendments to existing federal laws.
"2. Mixing PR & a separately elected President famously doesn't work very well"
So make the presidency a PR system as well! I look forward to President Kamala Harris in 2026 and 2028* and hope we survive the 2025 and 2027 reign of terror by President Donald Trump.
Republicans might be over interpreting the 2024 race result and underestimating the plasticity of voters. They have failed to deliver on promises of prosperity, they are actively terrorizing Hispanic communities, and are just being an utter clown show. How will they preform in 2026 when all their stagflationary policies come to roost?
2. On a per-state basis, why not? It doesn't completely solve electoral college disproportionality, but it massively reduces it.
3. Depends: if you use STV or open lists, then there are no safe seats; if you use closed lists, the people at the top are safer than any safe seat under FPTP.
2. You'd have to put in a floor (10%? 15%?) for winning an electoral vote. Otherwise weirdo third parties would pick up enough EVs to throw a lot of elections into the House of Reps. There's a similar rule in the presidential primaries - you need to hit a minimum percentage to get any delegates.
Yes, you likely would - though note that there are only four states with 20 EVs or more, so even just a state rule in those four states would reduce that problem enormously. A party could get represented on <2% in California and <3% in Texas, which is much more of a problem than 9% in Wisconsin.
Ideally, you'd be doing this through a constitutional amendment that would let you replace the House of Reps rule - either with a runoff, or with a system that lets parties combine their votes to elect a coalition president (simple version: have the electoral college meet in Washington instead of in the states and let them have multiple ballots)
Given the massive failure it was in the last electoral cycle to get RCV implemented anywhere, I don't see STV catching on anywhere. Granted, we're also talking about an idea that's not even on the radar in PR.
Ranked choice has been doing quite well! It had one bad cycle, but it’s actually been picked up in several states and cities over the past decade or so, which is more than any other significant change in elections.
in some countries you vote for a specific person on the list when you cast your actual ballot and then the order within each party list is chosen by # of votes
however i would just let the party decide the party list
It should definitely apply to the electoral college so long as we're going to have the electoral college (it would solve most of the problems), but I think it would be as hard as or harder than just switching to a popular vote. House elections can be changed by ordinary federal legislation, but the only way to force every state to switch to proportional election of presidential electors would be a constitutional amendment. And doing it without an amendment would require separate legislation in every state (or at least a lot of states with the stragglers being evenly divided), which would require as much political consensus as an amendment anyways.
The state-by-state legislative approach to a popular vote (the NPVIC) is, assuming it's constitutional, actually easier because you only have to get to 270 electoral votes to bring it into force. There's no easy trigger point for a proportionality compact.
I'd use a jungle primary (like CA and WA) combined with ranked choice voting based on the number of state seats to determine slates of reps. Then voters in the general select a party. Seats would get allocated to the parties based on their proportion of the vote with the ranking from the primary used to determine who wins the seats.
So if a state has five seats, voters rank 1-5 in the primary across all reps. The top 5 Republican and Democrat vote getters fill their party's slate for the general. If Democrats won 60% of the vote their top three reps from the primary would go to Congress and Republicans would get their top two if they got 40%.
This could also lead to third party reps if the state is large enough and you only need to meet a small threshold.
The people at the very top of lists in PR would be pretty safe unless their party is wiped out completely. This can be solved if you include the ability to rank candidates on your party’s list in some way, which would also function as a kind of mid-election primary.
Another solution is to allow multiple slates to be ran under the same party headline. They’re aggregated for the party’s total vote tally and seat allocation, but then those lists themselves act basically as sub-parties for allocation of the party’s delegation.
1. I would think that if you have, say, three members in your multi-member district, the three top vote-getters in the primary win the nominations. The question is, should every primary voter get one vote or three votes? Probably one vote, but I could see that going either way.
And thinking about this from the article: "...it’s also bad to let weird aspects of geography rather than actual public sentiment determine election outcomes."
I really wish Matt had spent a couple paragraphs level-setting on what, exactly “proportional representation” means here. I’m not a moron, and I think I can intuit it, but I would’ve been nice to have the concept defined for clarity.
e.g., I’d have 7 Congressmen, all charged with constituent service? We get one or seven votes?
There’s a few different systems that all get called “proportional representation”. There’s pure party-list PR, which is what the term usually suggests - everyone just checks off a party, each party has a list of candidates, and based on the number of votes they get, the candidates down to a certain number on that list get elected. There’s also mixed-member proportional, which Germany and New Zealand do, where you vote for a local candidate and a party, and the local candidate with the most votes gets elected, but an additional party list is used to top up the legislature to make the total representation for each party proportional. The one I like best is Single Transferable Vote, which works like Instant Runoff/Ranked Choice does in single-member districts - everyone ranks all the candidates running; each vote first gets allocated to the first-place candidate on that ballot; if any candidates have more than 1/n of the vote, all the ballots that went for them get part of their weight moved down to their second choice candidate; this process is repeated until there’s no one above threshold; if you haven’t yet assigned all the seats, then candidates with the fewest votes are eliminated and their votes passed down; in the end you have n candidates selected. STV is great when there are multiple dimensions of representation different voters care about - it would be perfect for Belgium, where some people might rank the French socialists above the Dutch socialists above the French conservatives above the Dutch conservatives and other people might put the French socialists above the French conservatives above the Dutch socialists above the Dutch conservatives. But it’s also good for places where there are several issues that cross party lines, like housing policy or local concerns. It’s hard for electing a lot of seats though, since voters need to rank a lot of candidates. (It’s helpful if there can be slates of rankings drawn up by various interest groups that people can just copy with a single button.)
Some may like reading “Shape” by the mathematician Jordan Ellenberg, which gives an entertaining and approachable treatment to gerrymandering for a few chapters.
One of the points there was that the best test of whether single member districts are fairly drawn is NOT compactness or proportionality (both have logical problems). The best test is whether the probability is reasonably high that the allocation of seats resulting from the the district map in use would have occurred if you were picking uniformly at random from the set of all possible maps. This test is easy to apply using computers. The Wisconsin map fails this test badly.
That’s not a test of whether the districts were *fairly* drawn - that’s a test of whether the districts were drawn in a way that isn’t sensitive to allocation of seats. There’s an important philosophical question of whether this is what “fair” means. Is it “fair” that in states like Massachusetts and Oklahoma, where 30% of the public votes for the minority party, there is no political representation for that party? Is it “fair” that in states like Texas with large black and Asian populations, there are several seats controlled by black voters but none controlled by Asian voters? If you’re going by random sampling of geographic districts, you have to say yes. But there’s no philosophical reason why this is what it means to be fair. Fairness in representation doesn’t obviously seem like it should have anything to do with where you’re physically located, such that segregation makes it “fair” to represent you and desegregation makes it “unfair” to give you representation.
Wouldn't it be fair to say that with Ellenberg's random walk approach you can identify gerrymanders, and maybe pick a better one that is not an outlier. Its main purpose is to detect unfair maps in a single member per district winner take all system.
A multi member PR system is an alternative, a different domain altogether.
Both good points. I share the intuition that a group of X people who are geographically dispersed within a state is entitled to as much representation as a group of X people who live near one another. But I also think we have to give up on single member districts altogether to accommodate a principle like that. If we take it for granted that we are going to have single member districts, then I like Ellenberg’s fairness criterion.
I think Single Transferable Vote is much better, since it allows people to pick what dimensions they particularly care about representation on. Some people would care more about party, some would care more about race, some would care more about geographic representation, some would care specifically about housing policy, etc. It’s very difficult for voters to do this if you’re ranking several dozen candidates, as you would have to be for districts with more than about 6-8 seats, but if there are several preset rankings you could easily pick from, it would work fine.
I don’t see much of a problem with the current system, i.e., a community selecting their representative. The argument against gerrymandering, in this view, is that it does not respect communities.
No proof, but I think most people could name the party they voted for last election, but they'd probably have a harder time naming their congressional representative. People already are voting based on party ID at least at the national level.
I will henceforth accept “likes” only from commenters in my district.
It's a good joke, but obviously there are benefits to having "your" representative. Who do you go to in a multi member state with say 10 reps. Which one is yours? Do you just contact all 10?
If so now all 10 are getting everyone in the state. That's kind of hard to handle, and really not efficient.
Excellent point. Proportional representation solves (maybe) *one* problem, that of fair partisan balance, while throwing away the benefits of having a one to one relationship between officeholder and constituent.
Look at Germany and New Zealand for examples of mixed-member proportional systems that have both geographic representation and party lists to achieve proportionality.
Also the larger the district the less attention paid to local issues like a toxic waste cleanup site in a rural part of a big state
"...Who do you go to in a multi member ....?"
Why not a central number with a cab-rank rule?
We can get credit for this? No one's been liking my comments for years!
Have I found a kindred spirit in my dear blorpington?
"..We can get credit for this?...."
If you save up enough "likes," you can exchange them for S&H Green stamps.
Are you in SoCal? If not, I can never like any of your comments again, sorry.
All "likes" should be aggregated and then distributed proportionately across all comments based on an algorithm designed by me to make sure the allocation is entirely fair.
"...aggregated and then distributed proportionately...."
I like this idea! It would amount to a Universal Minimum Basic Like.
Any more than that would lead to UMBL bragging.
You seem unclear on what the underlying logic of my algorithm would be, or how the word "proportionately" is meant in my definition, and what results it would produce.
"...You seem unclear on...."
Well, yes, because you *left* it rather unclear. But you said it would be "entirely fair," and since I know that you are an entirely fair person I assume that you would make sure that the distribution would begin by allocating at least 1 like to every commenter, and then distribute the balance in proportion to...some other metric (e.g. whether the commenter's initials were "M.R."?)
That's *exactly* what I would do.
[Whistles. Looks around innocently.]
It's not your initials that earn you my likes, it's the quality of your comments.
You have 56 members in your district 😄
Worth noting that multimember districts are illegal under the Uniform Congressional District Act. https://en.wikipedia.org/wiki/Uniform_Congressional_District_Act
Of course, Congress could repeal that. But it never will, because the status quo is advantageous for one side (the Republicans) while you'd need bipartisan support to change the law.
And it wouldn't be enough just to legalize multimember districts - even if you did that it wouldn't make sense for blue states to implement proportional representation unless red states also did it, which they wouldn't. You'd have to get Congress to not only legalize multimember districts but mandate them.
Well obviously a national anti-gerrymandering law would need to repeal that
What do you think of the Fair Representation Act? If has ranked choice voting, multi member districts and districting reform: https://fairvote.org/our-reforms/fair-representation-act/ Would ranked choice be better?
I think this is a feature not a bug: proportional representation (unlike electoral college or senate reform, but like prohibiting partisan gerrymandering) wouldn’t require a constitutional amendment, just a simple act of congress.
There is a certain amount of rotisserie politics on Slow Boring. This is a fun distraction from the death of democracy, but won't get implemented. My personal preference for this kind of post is high speed rail maps and (alt) history (lessons). For me, this post is a little to close to the depressing political reality, while only offering a solution in a fantasy world where we have brave politicians who focus on the greater good at the expense of their party's and own self-interests.
Yeah. This is like watching people in a burning house fantasize about the great sprinkler system they were planning to install. I’m all in favor of policy posts that have a long-term flair, but this blog increasingly seems increasingly like a missive from some parallel universe.
You could expand the house and then have a rule that the number of candidates per district is the number of representatives allotted to the smallest state.
Why is the status quo advantageous for the Republicans? The distribution of House seats by party tends generally to reflect the overall national House votes by party. I don't think either side has an advantage in the present system.
The issue isn't a gap between national vote share and representation; it's between state vote share and representation. Illinois Republicans are barely represented in the House and Arkansas Democrats have none.
In terms of the national vote, over the last 10 elections Republicans won, on average, 1.2% more seats than expected by the national house vote and they've won more seats than the national house vote in 7 of 10 elections.* There are likely two reasons:
1) The republican structural advantage lets them control more state legislatures and draw more maps.
2) Small states with few districts are more likely to lean right. Out of the 21 states with 4 or fewer districts, 13 are right-leaning 7 are left-leaning, and one is purple. Of the 46 seats in these states, republicans won 70% in both 2020 and 2024.
Whether a 1.2% bias is an problem to solve or not is debatable, but there is a general republican advantage that sometimes is enough to swing control of the House.
* I calculated the gap based on a two-party election. The two parties received 97% of the house vote in the last 10 elections, but captured all house seats in all but two and 434 of 435 in the other two.
>over the last 10 elections Republicans won, on average, 1.2% more seats than expected by the national house vote
Your 'expectations' are wrong. Single member districts are not proportional representation. If you get say 47% of the national vote in 435 separate contests, there is almost no way to translate that into 47% of the legislature. If anything, it's a miracle that the US system gets anywhere near so close.
PR systems don't necessarily translate exact vote % into exact seat %, for boring logistical reasons including thresholds and which exact algorithm is used.
Your two reasons give me, a former political scientist, a headache :) Just think about how hundreds of separate contests play out. If you run up your score in a district that you win, the extra votes are all 'wasted'. It's inevitable that the parties will run up the score in any number of seats. Sit down and make a 10 district model in Excel, and play around with the vote totals, and I think it'll start to make sense
I was just replying to the original comment that suggested there's no problem due to the similarity in national vote share and house allocation.
I agree with all your points and don't think it's a reasonable expectation that there is a match.
I do think that a crude degree of proportionality should be expected, such as Arkansas having a reasonable chance of electing a democrat to the house from time to time.
But, seeing how close the national vote share and the house composition is made me less favorable to Matt's point. Even if proportional representation would improve the proportionality of state results, the degradation of accountability of reps to their cities is a big downside.
The balance between seats won and distribution of the vote at the state level is a secondary issue. Which party controls the House of Representatives is the primary issue. It can make a world of difference, as we can see right now.
Now it is possible that the Republicans have a few more seats in their majority than they might otherwise, but to win *any* majority in the House -- which is what the parties strive for -- almost always requires one party to win a majority of the national two-party vote. (2012 is the only exception I'm aware of.) Having extra seats in your majority is nice, but seeing how the parties can pass their highest priority legislation even with paper-thin majorities, it's not that crucial.
If the only thing you care about is who wins the house and the breakdown of seats matches the national vote share, than you're right. But, obviously, people might weigh factors differently. I personally think that the state issue and direct representation is more important than making every single election perfectly match the national house results. And this perspective explains Matt's article, which was the intent of my comment.
If the only thing you care about is who wins the house, then the national bias also matters. Proportional representation would also improve this. It would make it less likely that a party would lose the national house vote and still win the house, which happened in 2012 and 2016, 20% of the last 10 elections (and also happened in 2004).
>a party would lose the national house vote and still win the house, which happened in 2012 and 2016
?? Republicans won 1.3 million more votes in 2016 in the House https://en.wikipedia.org/wiki/2016_United_States_House_of_Representatives_elections
Nor did it happen in 2004, when Republicans won about 3 million more votes...... https://en.wikipedia.org/wiki/2004_United_States_House_of_Representatives_elections
I think the advantage comes from having more opportunities to gerrymander (fewer GOP states with independent redistricting) not from anything inherent in district based elections
And yet in almost every cycle, the party that wins the greatest number of votes across the nation achieves a majority of House seats.
I'm the captive of a one-person bubble, but the fact that proportional representation is never mentioned by anyone in American politics is truly mind-boggling.
What I continue to be frustrated by is why some municipality doesn't try it. The federal and state hurdles are well known, but those usually don't exist on the municipal level. Especially in municipalities whose constituency are more populous than some states.
Greetings from Cambridge, MA, which elects its city council and school board by PR, and has since 1941. A bunch of cities did around that era, and since then have mostly switched back.
Do you know why they’ve switched back over the years?
Much has been written. Here's one article: https://fairvote.org/archives/a-brief-history-of-proportional-representation-in-the-united-states
They cite, among other reasons:|
- political machines trying to get back into power
- dissatisfaction with other reform elements (like city managers) spilling over onto the entire project that came in with that reform, like PR
- racist and anti-Communist horror at the way that PR made it possible for racial or political minorities to ever be elected.
Thanks!
So when you have an issue with the city, whose door do you knock on?
In theory, any of the city councilors; they're equally responsive (or not...) across the city. There are some ideological groupings at times, where councilor X is known as the one about a particular neighborhood or cause, so you might pick them, or you might pick the one you remember meeting at a campaign event, or so on.
Not much different from talking to an at-large member of a more conventional city council instead of the one from your district or ward.
So in other words, you the constituent don't have a city councilor you can go to. If they're all your representative, none is your representative.
I think that's excessively cynical.
Or, alternative view, your direct rep can already decide to blow you off even when you are a constituent, and they often do.
Yeah, I have to reiterate Nathan's point, I've actually had more luck getting a response on certain issues I care about from a city councillor that doesn't geographically represent me than I have from my local councillor. This is in Seattle, where council seats are largely defined by geography, with a couple at large seats that represent the entire city, which is why I copied the entire council on my complaint and how the councilman in the district over came to respond to me.
We do each have two Senators. So while I think the question is fair, there is probably a solution.
Unfortunately, some time in the middle 20th century, multimember districts became synonymous with racial exclusion, because many cities had separate seats that all shared a district, in order that the white majority would win every election. (Some cities like Austin had a gentleman’s agreement to let black people run in one of the seats and Hispanic people run in another seat, so that it could be unified partisan control without racial uniformity drawing attention from the law.) As a result, a lot of remedies for racial exclusion took the form of “best practices” advocating single member districts, taking advantage of racial segregation to ensure racial representation.
It would be interesting for some smart PhD candidate to do a thesis on why there is so little experimentation among the fifty state laboratories of democracy.
While PR is vastly more important, I think it should be paired with the cube root method for House expansion. Small states (I live in Vermont) should have a minimum of two representatives.
“… the cube root method….”
Say more?
House size would be determined by the cube root of the population of the United States. IIRC, that would put the country around 697-698 members as of now.
So when we finally get to One Billion Americans, we'll have a 1000-member House of Representatives.
ONE BILLION CONGRESSMEN.
What would this do to Congress's surface area?
“… cube root of the population of….”
Thx!
If you draw a graph of all the (democratic) national assemblies in the world by population size and number of representatives, then a cube root turns out to be close to the line-of-best-fit (and is certainly the simplest formula anywhere near that line).
But what is the principled reason for this? I can see 150 as a reasonable upper bound for a legislative body on the basis of Dunbar’s Number, but other than the observed empirical trend, what is the *reason* to favor a number of representatives that grows as the cube root of population? Heck, at least logarithmic growth would seem to match population dynamics (albeit I believe strict choice of appropriate base would be birthrate-dependent, so let’s just use e.)
I haven't done the math on this, but if countries that don't use the cube root method and are below that line of best fit (like the US) then adopt that method and increase the size of their legislature, will the cube root method still be the best and simplest formula?
This should be generalized to:
#reps = pop^1/N where N = 1+ number of commas in pop.
(or if you are in France, then the number of periods, which is virgually the same.)
Doesn't this imply that at population of 999,999,999, you should have 999.999999667 representatives, and at a population of 1,000,000,000 you should have 177.83?
"... at population of 999,999,999, you should have...."
I believe it was Leibniz who said:
"Nature does not make jumps: that's the job of bureaucracies."
How is that better than “the least populous state gets one rep and everyone else gets floored multiples of that”
"During the drafting of the Constitution, George Washington played a key role in determining the size of the House of Representatives. Initially, the delegates at the Constitutional Convention proposed a ratio of one representative for every 40,000 people. However, at Washington's suggestion, this was changed to one representative for every 30,000 people. This was the only time Washington offered an opinion on the specifics of the debates during the convention."
George Washington was right.
11,300 would be a big House
Sounds good. The house should be doing a lot of the work of the administrative state. They need the capacity.
I don't see how this doesn't relegate power even more to leadership in the house. Backbenchers now are able to accomplish little. With 11k representatives, they could just come to Washington, vote for the leadership, then go home. Or just skip coming to Washington and vote for leadership and then go home.
With that many representative you introduce hierarchy, with the 11k house electing its own representatives (possibly iteratively) until you get to a manageable size.
I know what you really mean, but for those who don't, this would sound like a call to parliamentarianism.
Ha, I suppose that's true, but yes, what I mean is executive branch notice and comment rule making ought to be be done in house committees and subject to congressional vote.
I think we'll have to agree to disagree on what role we think the House should have in politics
Have them work in shifts. Blocks of representatives casting votes together on issues with one person present for them.
Because people like having "their" representative.
I'm more probably more sympathetic to the "representatives should be primarily accountable to the voters of a particular, coherent, geographic locality" argument than most, which is why I support vastly increasing the size of the house. Small, geographically compact districts are good for making reps directly accessible to the precise people that vote for them and less beholden to broader, nationalized party politics. The downside of PR systems is the homogenizing effect which probably also increases polarization and maybe even the influence of "the groups".
The actual thing that I get more aggravated by with gerrymandering than the proportionality is precisely the incumbent protection aspect Matt describes. That NY map that's arguably more gerrymandered but has far fewer safe incumbent seats would be a way better situation than the current status quo.
This loss of locally bound representation seems to me to be a big con that Matt didn't mention at all in this article.
The problem is we are polarized. In a less polarized system you can focus on creating contiguous districts and local service for constituents. But in 2025, the Congressman is just going to go to Congress and vote party line down the line anyway, so local representation is dead.
For the vote, sure, but a lot of stuff happens before the vote where the party hashes out the legislation they are going to vote on.
Besides voting though, there’s a whole portfolio of constituent services that would very likely improve with smaller districts and more responsive reps. That seems like a real benefit.
That's a fair point.
On the other hand, it's already pretty hard to manage a congress of 435 people. That gets even harder with say 1,000 people.
You just can't know that many people
Yea I think this gets it perfectly correct. Before maximally nationalized politics, it made a ton of sense. But there's probably no putting that toothpaste fully back in the tube, and devoid of that, PR would be much better, and in tow with that, a cube root scaled house.
That being said, it does in theory mess with my preferred non amendment requiring Electoral College reform, which is that all states should adopt Maine Nebraska style district based Electoral Vote allocation. However maybe it could be even more efficiently done. They could all be allocated proportionally. State has 10 reps and you won 60%? You get 8 Votes, (6 +2 Senators). Suddenly becomes much harder to have PV EV splits.
He did mention it, people have cars and phones and internet connections these days
I hear what you're saying, but considering the patchwork manipulation of the system, I imagine it would be a net good for the minority party in each state to have greater representation. Sure you lose specific local reps, but you have less tyranny of the minority and the need to account for more than just your base.
If you don’t do literal party list proportional representation, but instead do something like single transferable vote, you can get the benefits of both. People wouldn’t have to list all democrats in a tie above all republicans, or vice versa, but they could rank regional candidates above others if they care strongly about regional representation (or could rank YIMBY candidates above others, or whatever other issue they care more about). Regions that care strongly about regional representation would have it.
Yeah, I've heard a different arrangement where the House is as specified (though I'd be open to expansion) and the Senate goes from 2 Senators per state (which makes no sense at all) to PR (and this could be at either the state or national level). So you can yell at your local rep in the House and in the Senate you angrily vote for the Green or Tea Party and now the Dems and GOP both agree that they don't want PR...
“2 Senators per state (which makes no sense at all)”
How else would the states have equal representation?
What is a state? It's the people living there, right? Why would 40 million Californians have the same amount of representation in the Senate as 500,000 Wyoming residents (Wyomins?)? And no California Republicans (about as many voted for Trump as Texas Republicans) represented at all.
California is a state. Wyoming is a state. Each has two senators because each state gets equal representation.
What I asked you is how could the two states have equal representation if they have different numbers of senators?
Ah that's because I don't believe that all states require equal representation. "All states are created equal" isn't something I've ever heard anyone say. If Congress is supposed to legislate on behalf of the American people, then the people should have roughly equal say. Giving a Wyomin 80x the voting power in the Senate is not roughly equal.
States are not mere administrative regions—they are separate sovereigns. Equal representation in the Senate was, 1) a necessary compromise to get all of the original states to agree to form a union, and, 2) a way to ensure that the large states can’t steamroll the small states.
I suppose you must disagree with one or both of these facts if you don’t believe states are equal.
Isn't the Constitution very clear on what is required here?
"That NY map that's arguably more gerrymandered but has far fewer safe incumbent seats would be a way better situation than the current status quo."
agreed, yes to swing districts.
I agree that proportional representation is likely a better way for the makeup of Congress to reflect public opinion.
What I don’t really understand, though, is the flavor of procedure discourse that goes like: “X disadvantages my party right now. You know what would be great? Getting rid of X!” (Without any plausible argument about why the other side would unilaterally agree to disarm.)
1. It reminds me a bit of when conservatives say academia should hire more conservatives. Ok? Why? Maybe it would be better, but there’s no serious account of how colleges would be incentivized to do this. Or how it could be done without hiring lower-quality researchers. It’s just begging.
2. This article kind of implies that Democrats don’t quite act like they’re at war. They’re sitting around trying to make safe seats while Republicans try to win. This is a broader issue where Democrats kind of bend rules, so no one thinks they’re really acting in a principled way, but when they do bend rules, they don’t do so with the intent of beating Republicans.
3. For the various procedural questions that are often debated (e.g., court packing) I think it would be much more interesting to know how pundits stand on changing the rules several years out (so that we don’t know which party will be at an advantage when the rule changes).
“This is a broader issue where Democrats kind of bend rules, so no one thinks they’re really acting in a principled way, but when they do bend rules, they don’t do so with the intent of beating Republicans”
This is a good way to put it!
Yes, agree. In a system in which grinding partisan polarization is the one constant, I have to have at least a vague concept of how to get there politically before I can even entertain the thought experiment. It's like imagining the future of energy by starting with "first, assume we've invented cheap fusion reactors," except cheap fusion feels more realistic than getting Rs and Ds to cooperate on systemic procedural reform. Not that I begrudge Yglesias for indulging in some fantasy; he spends more time dealing with the limitations of the real world than most, so I suppose he's earned it. But still, I want to hear the theoretical road map here.
It's easy to incentivize universities to hire more conservatives. Just threaten to take away the billions of federal dollars they use to find cures for cancer.
"... they use to find cures for cancer...."
Yeah, that'll teach those pointy-headed nerds to respect conservatives!
1. I think not discriminating against conservatives would be sufficient. Also, for universities where they are still actually trying to find truth, you tend not to find it if everybody is already in agreement and there are no dissenting opinions.
This is mostly impishness on my part, but what if conservatives have a population-level deficit of academic ability? Is it fine if the discrimination-free equilibrium ends up being 80/20 or 90/10 toward the left anyway?
Probably yes.
You certainly don't want to hire anyone that can't do the job. And this is probably very much position specific. But if you have two candidates that are close for say a sociology position, maybe there's a benefit to having someone that approaches the subject from a different perspective.
But again, I don't think you need any type of affirmative action program for conservatives, just stop the discrimination.
Conservatives are inherently close-minded which is not a good fit for academia.
Would proportional representation reduce polarization, or increase it? For the dems specifically, I could imagine a world where it increases the power of the Groups: in a big state where people aren’t voting for a specific person I could imagine voters just lazily going with whoever gets the right endorsements, which I think plays right to the Groups. Additionally, if you can get—in effect—one Dem safe seat in a state like Arkansas, then would that person have any incentive to be a moderate, or would they be a standard-issue progressive of the sort that could never win a senate seat?
With PR, a green party would become viable, which would take pressure off the mainstream Dem party to be progressive.
With proportional representation the parties would most likely split into a Progressive party and a Liberal one. They could form alliances in Congress as needed depending on the issues.
There are proportional representation methods (most notably STV) where voters specifically choose candidates.
STV in turn probably requires something like 3 or 4 representatives per district (since the messaging from groups would be to rank particular ways with a constraint that the groups have to basically guess how much support they actually have when deciding how many candidates to include in their recommended rank).
So long as there’s a permanent Democratic majority, who cares about the power of the Groups?
Proportional representation would not, in fact, do this. Republicans won the house popular vote in 2024, 2022, 2016, 2014, and 2010. It probably would've helped Democrats (we'd have won in 2012, which isn't nothing!) but it wouldn't be a permanent D majority.
Yeah we need partisan disenfranchisement too
May as well ask for world peace while you are at it
It is the only way to save America, though
It's like imagining a world without lawyers: https://www.youtube.com/watch?v=--vLL3Kd664
Finally a solution we can all get behind!
"So long as there’s a permanent Democratic majority, who cares about the power of the Groups?"
I assume that's sarcasm, but if it's not, people that want good legislation passed would care.
Of course it would increase polarization. To the extent that providing constituent services is non-partisan (i.e., close to 100%) and to the extent that doing so makes voters more inclined to support you (almost certainly pr>0), then removing that tie to voters and basically having them vote by party label would make the system more polarized.
I don't think it's quite so clear. There's probably a u-shaped curve of polarization based on district size. The smaller the district, the less likely that voters are aware of candidates and will vote based on heuristics like name ID, party slates, endorsements, etc. When districts get bigger and/or more competitive, there is a greater incentive to reach voters through campaigning and advertising, increasing awareness. On the other hand, the bigger the district, the more likely that national issues will trump local concerns, leading to party-line voting, as happens with most senate seats today. These have countervailing effects, so there is a sweet spot somewhere in the middle.
The case against strict PR in the American system of government as it is practiced is that representatives often are intentionally the single point of contact for escalated government services. For individuals, this is everything from Capitol tours to flag requests to help when Federal agencies ignore you or requests for “private” bills for immigration purposes.
Under PR, when 2 or 3 or 9 representatives are accountable to you nobody is accountable to you.
I like the Germany/New Zealand approach of voting for a member with a clearly defined constituency as well as voting for a party to top off any national imbalances caused by local level concerns.
The idea I had during the 2020 reapportionment to achieve MMP that would also sate the "expand the House!" people would be to expand it to, say, 601 members, but the additional 166 members (27.6% of the new total) would all be chosen by PR. This protects current incumbents from having to deal with PR, and can also protect any incumbents from being gerrymandered out in states that lose population. Then, in 2030 you change the geographic/PR ratio to 35%, and in 2040, change it to 50/50.
That’s something I’ve thought of too and agree that it would, at least in theory, be the ideal way forward in implementing PR.
This what was I was going to bring up. As someone that once worked in constitutent services for my local Congressman, having a point person the federal government was kind of a big deal to some people! Is it not anymore? I feel like any time I hear my Congressperson now, she talks a lot about the local projects she is doing. Our district (urban/suburban) is very unique to the rest of the state (rural), and has unique needs that she addresses, that I'm not sure a state-wide elected Rep would also address.
Also, I think town halls are a really good thing, even if they're not just intended to scare the bejeezus out of Republican House members. Would PR render those extinct?
The idea that Congresspeople provide individualized constituent services seems poorly advertised and not something that I would expect to work practically in view of how many constituents each Congressperson has. Potentially this understanding is wrong! But it's also the obvious default assumption.
You could still give reps a nominal "district" like SCOTUS justices have a "circuit" for that purpose.
I agree. Mixed member proportional systems are better than pure proportional.
Proportional representation might have a good outcome when it comes to partisan control of congress, but that's not the only political outcome that would be affected. Going back maybe 20 years, Persson and Tabellini found that, relative to majoritarian elections, PR increases taxation and welfare spending but also decreases accountability, which results in greater corruption. Non-local officials are also less likely to care about things like closing local hospitals or factories. That may be an ok tradeoff, but it's still a trade off. There's probably other research on this I'm not aware of.
The benefit of the current system is supposedly the increase in localism but that seems like it's long gone, making the system reflect current political realities is i think much more important than some ideal where everyone will know every politician individually and vote on local issues.
And when you draw districts the way we do now, there's not much 'local' anyway.
Right, if you like localism, convince voters to be less partisan.
It's not really all voters though is it? Isn't it really primary voters
No, it's pretty much all voters, the casual non-primary voters also don't pay attention to local policy.
How would town halls work? Lots of constituents were angrily at recent ones.
It wouldn't prevent town hall type situations, you can always as a rep go somewhere in your state and let the constituents yak.
I know there's a romanticism behind the idea that house reps will think about local concerns but honestly my experience is opposite. Most current reps are basically soldiers for national parties and that's that and that's what voters have largely preferred in action, if not words.
“Lots of constituents were angrily at recent ones”
Who cares about angry people?
The representatives felt pressured enough not to go, so they got the sense that voters _cared_ about these things, which surely affects their future decision making at least a little.
Who wants to deal with angry people? I wouldn’t go either.
Neither would I, the point is that even the decision not to go affects them. I'm positing that if it's PR, that decision no longer exists so the effect doesn't exist.
The existing system does have 'localist' benefits. Of course these would be enhanced if the House was substantially expanded, you're not going to get all that much in the way of 'local' benefits from 700,000-800,000-pop. districts.
In theory, but in practice, no matter what the district is, house reps are running on national politics, not local issues.
I mean that national matters are prioritised is true in countries that have much smaller populations per member as well. But the 'benefits' do not only come at election time; in the UK, MPs end up adopting a role that is essentially 'social worker of last resort', dealing with individual constituents' problems dealing with the bureaucracy. The extent to which this is *actually* a 'benefit' or not is (very) debateable but it's something that's genuinely possible when your constituency (district) has a population of about 70-80k, ie 10 times smaller than a House district.
This situation is wonderfully parodied in this scene in "In The Loop"
https://www.youtube.com/watch?v=b5kdOvsyv98
Yes, that 'from White House to shitehouse' line - in addition to being a good joke - is very much how a lot of British legislators (especially the more ambitious ones) feel about these 'localist benefits', but they mostly lack the imagination to see how they could change things.
Maybe. But the greater the number of representatives, the less ability each one will have to intervene with the bureaucracy to help constituents.
Can you provide a link?
Also worth noting that Republicans find Democrats' anti-gerrymandering stance absurd because Democrats are very pro-gerrymandering when it comes to majority-minority districts.
That’s half the story. In the 1980s when the VRA was amended to require minority-opportunity districts the RNC chair was a big fan. Drawing new black majority seats in the South helped Republicans beat a lot of white Democrats and they wouldn’t have flipped the House under Gingrich without it. If VRA districts were no longer required Democrats could draw much more efficient maps in southern states they control.
Yeah the racial gerrymandering thing has a different valence when it was more about White Democrats vs Black Democrats in the South. Now that a state like Mississippi is almost perfectly racially polarized, it's just pure partisanship.
Mercifully it appears SCOTUS will resolve this problem by next June
Oh I'm fairly certain that GOP states in the south will keep majority-black districts. Going all-geographically-compact means fewer 60% GOP districts.
Ehh... I think Mississippi could easily make a solid 4-0 map.
Yes, the VRA serves as a protection racket for a small number of Black incumbent officeholders.
It’s not gerrymandering to ensure that communities of interest that have long thin regions of habitation get represented. A district that follows the banks of the Mississippi, or that gets the core of several segregated cities, is much more natural than a square district based on nothing meaningful.
I think that again comes down to how you define gerrymandering. For example, in Louisiana, the districts are not compact and the borders look funny. However, the representative is proportional to voting: Republicans got 65% of the votes and 66% of the seats.
The gerrymandering by both sides tends to cancel each other out and on the whole the House reflects the national will of the people.
What I think is bad is tossing the norm to only do redistricting once, after the census. Otherwise, it's just a chaotic free for all. So I'm happy to see Democrats pushing back against Texas and the Republicans in this instance.
1. It's a radical solution that would make titanic changes to the political order the US has had for 250ish years. It's a little a strange for an essayist whose whole schtick is 'radical change is probably bad, let's do incremental change instead' to advocate for. In terms of radicalism I'd rank it slightly ahead of 'outlawing private health insurance and creating an American single payer system overnight'.
If you wanted to try this out first in like a small state or county government, I mean, sure.
2. Mixing PR & a separately elected President famously doesn't work very well- proven not just throughout Latin America but also when Israel briefly tried a similar system to this for a few years. That's why literally not one single rich country structures their political system this way. This is doubly true with bicameralism and 2 year terms. The House is made up of Parties ABC, the Senate of parties CDE, the President from party B. To get any legislation passed, you have to navigate cross-cutting alliances of 6+ separately elected parties, all responsible to a different small voting/more ideologically intense voting base. Then, these coalitions turn over every 2 years! The House and the Senate would be in a constant churn of parties engaging with each other, and then with the opposite house- overturned every other year.
What's happened in dozens of Latin American countries over 150+ years of trying this is that it's too difficult to get *anything* passed this way. Too much squabbling, too many separate constituencies. Consequently the President gains more & more power, just in an attempt to get something- anything- passed. This way lies Caesar!
(If you don't know the difference between a PR parliament & a PR presidency, please do not respond so as to not raise my blood pressure. Thanks).
3. The US is already veering towards becoming more & more like a Latin American state. "Let's adopt the political system of Brazil, it's going to totally work out great this time, I promise" is not very convincing! Again, not one single rich country mixes PR & a President!
"If you wanted to try this out first in like a small state or county government, I mean, sure."
Yes, this is one of the things that I find tiresome about electoral reformers -- there are a bunch of states where the state constitution can be amended by voter-initiated amendments (and in some of those states even just a bare majority of the popular vote is enough to win), but it's very rare to have any sort of electoral reform proposed via ballot measure. Make California a parliamentary system or get Illinois to adopt proportional representation for the state legislature and you'll likely be able to get more of a hearing about making these changes at a national level. (Cf. giving women the right to vote -- about half of states adopted that on their own before the 19th Amendment was passed and having those examples almost certainly made passage and ratification of the amendment easier.)
The US has had the same 2 parties since the 1850s, which is older than most democracies today. I think it's weird that the writer who's against radical change thinks we should switch our entire electoral system, at the national level, overnight! It's not something that can be done incrementally or in stages
The problem with this is that PR is unilateral disarmament for any state with a partisan skew. And in practice, since Republicans aren't interested in voting system reform, this would mean asking a state like Massachusetts to adopt PR and hand over more seats to Republicans while knowing that a Republican-controlled state like Wisconsin is never going to return the favor. The game theory of piecemeal PR doesn't make sense.
You could switch for state office without changing federal office.
Prove that CA or IL could be successfully governed that way.
For state level office voters care much less about partisan affiliation. Not zero (good luck electing a Republican mayor in NYC these days), but also you would never get Charlie Baker to represent MA at the national level.
For State Legislature they care a whole ton about it. And that's who would be elected through PR, not Governor
That's why I explicitly suggested "Illinois to adopt proportional representation ***for the state legislature***" (emphasis added). Additionally, states almost certainly could not adopt proportional representation for U.S. House seats without amendments to existing federal laws.
My bad, misinterpreted.
"2. Mixing PR & a separately elected President famously doesn't work very well"
So make the presidency a PR system as well! I look forward to President Kamala Harris in 2026 and 2028* and hope we survive the 2025 and 2027 reign of terror by President Donald Trump.
* Minus a couple months.
Republicans might be over interpreting the 2024 race result and underestimating the plasticity of voters. They have failed to deliver on promises of prosperity, they are actively terrorizing Hispanic communities, and are just being an utter clown show. How will they preform in 2026 when all their stagflationary policies come to roost?
How do you primary in multi member districts?
Should proportionality apply to the Electoral College?
Does it make elections more contestable, fewer "safe seats?"
1. STV
2. On a per-state basis, why not? It doesn't completely solve electoral college disproportionality, but it massively reduces it.
3. Depends: if you use STV or open lists, then there are no safe seats; if you use closed lists, the people at the top are safer than any safe seat under FPTP.
2. You'd have to put in a floor (10%? 15%?) for winning an electoral vote. Otherwise weirdo third parties would pick up enough EVs to throw a lot of elections into the House of Reps. There's a similar rule in the presidential primaries - you need to hit a minimum percentage to get any delegates.
1. incentivize enough weird third parties to pick up EVs that the house always picks the president
2. lower threshold for impeachment & removal to house majority alone
3. parliamentary government achieved!
Yes, you likely would - though note that there are only four states with 20 EVs or more, so even just a state rule in those four states would reduce that problem enormously. A party could get represented on <2% in California and <3% in Texas, which is much more of a problem than 9% in Wisconsin.
Ideally, you'd be doing this through a constitutional amendment that would let you replace the House of Reps rule - either with a runoff, or with a system that lets parties combine their votes to elect a coalition president (simple version: have the electoral college meet in Washington instead of in the states and let them have multiple ballots)
Given the massive failure it was in the last electoral cycle to get RCV implemented anywhere, I don't see STV catching on anywhere. Granted, we're also talking about an idea that's not even on the radar in PR.
Ranked choice has been doing quite well! It had one bad cycle, but it’s actually been picked up in several states and cities over the past decade or so, which is more than any other significant change in elections.
And TIL! https://wonkhe.com/blogs/the-history-of-university-representation/
Can we also get University representation along with STV ;)
its simple, abolish primaries
Okay who has the power to decide what actual people get to go on the party list and in what order then?
The parties themselves (though I am not a fan of party-list proportional: STV is my thing).
in some countries you vote for a specific person on the list when you cast your actual ballot and then the order within each party list is chosen by # of votes
however i would just let the party decide the party list
Letting parties fully control the lists is terrible. This just leads to party splintering if you don't get along with the boss (see: Israel)
The DNC/RNC decides who is on the party list? :)
who else?
It should definitely apply to the electoral college so long as we're going to have the electoral college (it would solve most of the problems), but I think it would be as hard as or harder than just switching to a popular vote. House elections can be changed by ordinary federal legislation, but the only way to force every state to switch to proportional election of presidential electors would be a constitutional amendment. And doing it without an amendment would require separate legislation in every state (or at least a lot of states with the stragglers being evenly divided), which would require as much political consensus as an amendment anyways.
The state-by-state legislative approach to a popular vote (the NPVIC) is, assuming it's constitutional, actually easier because you only have to get to 270 electoral votes to bring it into force. There's no easy trigger point for a proportionality compact.
I'd use a jungle primary (like CA and WA) combined with ranked choice voting based on the number of state seats to determine slates of reps. Then voters in the general select a party. Seats would get allocated to the parties based on their proportion of the vote with the ranking from the primary used to determine who wins the seats.
So if a state has five seats, voters rank 1-5 in the primary across all reps. The top 5 Republican and Democrat vote getters fill their party's slate for the general. If Democrats won 60% of the vote their top three reps from the primary would go to Congress and Republicans would get their top two if they got 40%.
This could also lead to third party reps if the state is large enough and you only need to meet a small threshold.
The people at the very top of lists in PR would be pretty safe unless their party is wiped out completely. This can be solved if you include the ability to rank candidates on your party’s list in some way, which would also function as a kind of mid-election primary.
Another solution is to allow multiple slates to be ran under the same party headline. They’re aggregated for the party’s total vote tally and seat allocation, but then those lists themselves act basically as sub-parties for allocation of the party’s delegation.
1. I would think that if you have, say, three members in your multi-member district, the three top vote-getters in the primary win the nominations. The question is, should every primary voter get one vote or three votes? Probably one vote, but I could see that going either way.
2. Yes.
3. Seems like it.
I like the solution.
And thinking about this from the article: "...it’s also bad to let weird aspects of geography rather than actual public sentiment determine election outcomes."
That's a good reason to abolish the Senate.
It would also be a good reason to annex Canada.
Speaking of things that will never happen
I really wish Matt had spent a couple paragraphs level-setting on what, exactly “proportional representation” means here. I’m not a moron, and I think I can intuit it, but I would’ve been nice to have the concept defined for clarity.
e.g., I’d have 7 Congressmen, all charged with constituent service? We get one or seven votes?
There’s a few different systems that all get called “proportional representation”. There’s pure party-list PR, which is what the term usually suggests - everyone just checks off a party, each party has a list of candidates, and based on the number of votes they get, the candidates down to a certain number on that list get elected. There’s also mixed-member proportional, which Germany and New Zealand do, where you vote for a local candidate and a party, and the local candidate with the most votes gets elected, but an additional party list is used to top up the legislature to make the total representation for each party proportional. The one I like best is Single Transferable Vote, which works like Instant Runoff/Ranked Choice does in single-member districts - everyone ranks all the candidates running; each vote first gets allocated to the first-place candidate on that ballot; if any candidates have more than 1/n of the vote, all the ballots that went for them get part of their weight moved down to their second choice candidate; this process is repeated until there’s no one above threshold; if you haven’t yet assigned all the seats, then candidates with the fewest votes are eliminated and their votes passed down; in the end you have n candidates selected. STV is great when there are multiple dimensions of representation different voters care about - it would be perfect for Belgium, where some people might rank the French socialists above the Dutch socialists above the French conservatives above the Dutch conservatives and other people might put the French socialists above the French conservatives above the Dutch socialists above the Dutch conservatives. But it’s also good for places where there are several issues that cross party lines, like housing policy or local concerns. It’s hard for electing a lot of seats though, since voters need to rank a lot of candidates. (It’s helpful if there can be slates of rankings drawn up by various interest groups that people can just copy with a single button.)
Some may like reading “Shape” by the mathematician Jordan Ellenberg, which gives an entertaining and approachable treatment to gerrymandering for a few chapters.
One of the points there was that the best test of whether single member districts are fairly drawn is NOT compactness or proportionality (both have logical problems). The best test is whether the probability is reasonably high that the allocation of seats resulting from the the district map in use would have occurred if you were picking uniformly at random from the set of all possible maps. This test is easy to apply using computers. The Wisconsin map fails this test badly.
That’s not a test of whether the districts were *fairly* drawn - that’s a test of whether the districts were drawn in a way that isn’t sensitive to allocation of seats. There’s an important philosophical question of whether this is what “fair” means. Is it “fair” that in states like Massachusetts and Oklahoma, where 30% of the public votes for the minority party, there is no political representation for that party? Is it “fair” that in states like Texas with large black and Asian populations, there are several seats controlled by black voters but none controlled by Asian voters? If you’re going by random sampling of geographic districts, you have to say yes. But there’s no philosophical reason why this is what it means to be fair. Fairness in representation doesn’t obviously seem like it should have anything to do with where you’re physically located, such that segregation makes it “fair” to represent you and desegregation makes it “unfair” to give you representation.
Wouldn't it be fair to say that with Ellenberg's random walk approach you can identify gerrymanders, and maybe pick a better one that is not an outlier. Its main purpose is to detect unfair maps in a single member per district winner take all system.
A multi member PR system is an alternative, a different domain altogether.
Both good points. I share the intuition that a group of X people who are geographically dispersed within a state is entitled to as much representation as a group of X people who live near one another. But I also think we have to give up on single member districts altogether to accommodate a principle like that. If we take it for granted that we are going to have single member districts, then I like Ellenberg’s fairness criterion.
Serious question: Why should there be proportional representation based on political parties?
I think Single Transferable Vote is much better, since it allows people to pick what dimensions they particularly care about representation on. Some people would care more about party, some would care more about race, some would care more about geographic representation, some would care specifically about housing policy, etc. It’s very difficult for voters to do this if you’re ranking several dozen candidates, as you would have to be for districts with more than about 6-8 seats, but if there are several preset rankings you could easily pick from, it would work fine.
I don’t see much of a problem with the current system, i.e., a community selecting their representative. The argument against gerrymandering, in this view, is that it does not respect communities.
We are all one big happy community Ken!
No proof, but I think most people could name the party they voted for last election, but they'd probably have a harder time naming their congressional representative. People already are voting based on party ID at least at the national level.