A modest and timely proposal

Democrats are fundamentally disadvantaged when it comes to winning U.S. House majorities. This is because their votes concentrate in population-dense areas. Independent redistricting cannot fix this. On all this, read Rodden (2019) and McGann, Smith, Latner, and Keena (2016).

Depending on the outcome of Georgia’s two Senate runoffs, Democrats may be in a unique position to fix their “geography problem.” Unified government would make it possible to adopt a modest form of proportional representation.

If Democrats do not do this, they are likely to lose the House again in 2022. That will mean a return to gridlock — and all that it entails for Democrats’ brand.

Here is one plan going forward:

1) Eliminate the Senate filibuster with respect to election legislation.

2) Adopt the Fair Representation Act, which calls for PR in districts of 3-5 seats.

3) Replace its ranked voting provision with open-list PR (OLPR). In a “one-vote” version of OLPR, the voter would pick their favorite candidate. That vote would count for a person and their party. Within a district, parties then receive seats in proportion to their vote shares. Within a party, votes for people determine which ones get the seats. One-vote OLPR is minimally disruptive for several reasons:

3a) From the voter’s perspective, there would be almost no change — pick your most preferred candidate.

Sample OLPR ballot

3b) From the incumbent’s perspective, there would be almost no change. Have each incumbent campaign as they do now. Carve multi-seat districts into geographic areas that correspond to the old districts. In comparative politics, these regions are known as “bailiwicks.”

3c) For the purpose of winning seats not currently held, run new candidates in those regions.

3e) From the election official’s perspective, there would be almost no change — a larger ballot, but without any need to implement ranked-choice voting. Just count up the votes, and report the totals.

3f) Have each state’s chief election official allocate seats, based on party-level vote totals, via D’Hondt or Hare or whatever. There are many options.

4) Get over the allergy to third-party politics. Take a serious look at what states and districts are likely to produce third-party winners. Others may disagree with me, but I suspect gains would be isolated and modest.

No, this doesn’t “work” for states with one or two seats. Any “spoiler” problems would need to be solved without ranked voting (which requires coordination anyway).

What about nominations? I suggest letting (state) parties decide how to structure these. One option is to use “nominating districts,” i.e., the current single-member congressional districts (which might collapse into general-election multi-seat districts). Then there might need to be some provision for candidates not strong in any one “nominating district” — maybe a multi-seat-district-wide primary at the same time. Another option is just to get rid of primaries, then have a party committee (local, state, and/or district) decide whether some candidate can appear on its list. There are many options. (Full credit goes to Matthew Shugart for the “nomination districts” idea. This is fleshed out at his blog. – JS, 2021-01-23)

Let me stress: if Democrats win these runoffs, they will have a rare and brief opportunity to “fix their geography problem.”

Photo credit: James Godowic.

24 thoughts on “A modest and timely proposal”

  1. It appears that Jack thinks his OLPR proposal would be easier to get adopted, and/or easier to implement, than the current version of the Fair Representation Act. But I think it has three practical political disadvantages which make STV, as in the current version of the act, easier to win rather than harder.

    First, STV is the only form of PR that can answer the complaint that PR gives “too much power to political parties”. (I don’t think that’s a problem myself, but the practical reality of electoral reform in the U.S. is that it has to be addressed.)

    Second, STV seems less “foreign” and less radical than any form of list PR. Maybe this is just a restatement of my first point in different words, but I don’t think so.

    Third, Jack’s specific proposal includes enough complexity in the way it deals with geographic boundaries and nomination procedures to match, or nearly match, the (alleged) complexity of STV. If the intent is to placate defenders of single-member districts by maintaining links between representatives and plots of ground, then I think MMP is probably a more straightforward choice.

    But in order to get adopted in the U.S., MMP needs some way to be fair to “non-partisan” candidates and voters. That can be done, but it means it would appear more complicated here than it does in New Zealand and Germany. STV really isn’t so intricate after all.

    1. The goal was to find something that can be adopted quickly and easily. There is no question that, from an administrative and voter-education perspective, one-vote OLPR is quite easy to implement. As for geographic representation, the difference between lists and districts is overstated (see, e.g., Fiva & Halse 2016). I agree that something is needed to make “independents” feel that they have been treated “fairly,” and maybe this can be done at the state-party level (essentially what happens now).

  2. If one believes in an electoral system with strong political parties, the list system is the way to go. If one believes in non-partisan elections then either STV or an open list PR will work. As far as ease for the voter then either of the list systems work best. Each party can use its primary to arrange their closed list, or taxpaying voters could save millions of dollars by abolishing the primary and have all qualified candidates (who gather nomination signatures or have a party submit a list of names) placed on the general election ballot. In any case, most Democratic and Republican politicians and their supporters will oppose anything that weakens their hold on power.

  3. I am skeptical about replacing RCV with OLPR. Under the FRA district magnitude rules, the average district magnitude is only 3.7. A proportional system, and particularly a list system, would undoubtedly draw more votes to third party candidates hoping to win seats under the lower threshold. However, the threshold would still be too high for most (maybe all) of them. Under OLPR, those votes would all be wasted. This is both unfair to the third party voters and could cause an overall distortion in favor of the less fragmented party (probably the Republicans, unless the GOP splits into two parties or something).

    The ranked ballot would ensure that votes that would otherwise be wasted could go to back-up choices. There would still only be modest (if any) gains for third party candidates (at least at first), but third party voters would have more power, especially because major party candidates would want their back-up choices. The gains you’d get from using OLPR instead of RCV would be offset by some pretty serious disadvantages of that system.

    All that said, any proportional system would be miles better than the current system, and so I would argue that the focus should not be on policy benefits of one or the other, but should be focused almost exclusively on which is more likely to pass (though the relative policy benefits are relevant to that inquiry). On that front, RCV has a lot more momentum. Honestly, it is hard for me to see OLPR catching up.

    1. There are two kinds of wasted votes. One kind is cast for a viable party but does not “count” due to geography. The other is cast for a hopeless party. My proposal is meant to deal with the former.

  4. I think that “one-vote OLPR” is equivalent to Balinski’s Fair Majority Voting. If so, I agree with Santucci that it’s an easier, quicker fix than multi-seat¹ STV. But it still raises the thorny issue of party thresholds; set them too low, and you get an Israel problem of excessive party fragmentation, but set them too high, and you get vote-splitting problems and wasted votes.

    I’ve been saying for a long time that we desperately need a healthy, realist debate about proportional representation methods for the US. Santucci is showing the way here: we need to make it past idealist and absolutist positions. Different methods have their pros and cons, and no one argument is the be-all and end-all.

    Here’s the landscape, as I see it:

    STV (aka RCV5)
    * No explicit role for parties, so it’s naturally fair to independent candidates.
    * Natural 17% threshold, so naturally limits the number of extreme fringe positions that get represented.
    * Complex ballots and procedures.
    * Arguably, removes/reduces incentives for pre-election coalition formation; could tend to favor single-issue tunnel-vision.
    * Very disruptive to status quo, even beyond what’s necessary to deal with gerrymandering/misrepresentation. A heavy lift to pass in the House.

    Santucci’s OLPR/FMV
    * Simple ballots and implementation (eg, “precinct-summable” tallies)
    * Non-disruptive in the immediate term
    * Raises issue of thresholds.
    * IMO, Probably leads to fragmented parties, vote-splitting, and disruption in the medium term. Even if this isn’t true, if current parties believe it is, they’d veto the idea.

    MMP using existing districts (that is, simultaneously increasing the size of the House):
    * Done right (modified Bavarian), this could include quasi-transfers for below-threshold votes and expending for winning local votes. This means you could have relatively high thresholds without such high wasted votes, with room for intra-party diversity. In other words: relatively non-disruptive to partisan incumbents in the short and medium term.
    * Though ballots and procedures are simpler than STV, they’re less so than one-vote OLPR or PLACE.
    * Goes hand-in-hand with a simultaneous increase in House size. Though that’s actually a good idea in its own right, it still increases the surface area for opponents to attack the idea.

    PLACE voting
    * Relatively simple ballots and summable counts.
    * Avoids wasted votes but still discourages excessive party fragmentation. Encourages healthy forms of organization — grass-roots, but unified by leaders/spokespeople.
    * Non-disruptive to incumbents.
    * Clear lines of accountability.
    * Maximal voter choice and voter power (more than any of the above).
    * New and untested.
    * A complex design which involves several carefully-balanced interlocking ideas. Hard to explain quickly.

    Of the above, I currently think the best chance for Congress to pass “quick fix” along the lines suggested in this post is MMP. But I’d love to be convinced otherwise! This is an important discussion to be having!

    ¹: I say “multi-seat” to help avoid confusion with “mixed-member”.

    1. I think you need 2 define PLACE voting.

      The issue of threshholds 4 Jack’s idea goes with whether a Hare or other quota is used, and I believe Jack is also proposing an increase in the number of seats. This is designed to get rid of the advantage given to the GOP due to the geographical concentration of Dem-leaning voters. But the issue is whether the GOP will be able to hold it together as a party in the coming year, in which case it may be the next center-right party that stands to gain from electoral reform.

      Cui bono? If the GOP holds itself together with the help of megadonors, ‘n bribes to Trump et al., then Jack’s idea wd seem to target giving the Democratic party a way to consolidate its control of the US House. I’m cool with them using a Droop quota, because I think the Hare quota wd have more value-added for state representative elections where de facto segregation is even worse in its effects.

      1. Thanks. The point was (and, I think, is) to find a workable solution to a real-life problem. I know that many people have favorite electoral systems (or sometimes just ballot types). Those debates are really great to have.

  5. Why settle for second best when you have a Fair Representation Act ready to go?

    It would implement STV pretty much as used in Ireland, which should go down well with Biden!

    One point on thresholds. In practice the threshold of first-preference votes to have a reasonable chance of winning a seat in a constitutency is roughly half a quota; and taken across the country parties with much less than that can get their fair share of seats: look at the most recent Irish elction if you don’t believe this – https://en.wikipedia.org/wiki/2020_Irish_general_election.

      1. Ireland in the early 1920s fought the War of Independence, followed immediately by the Civil War. The country was poor, bitterly divided and the British had made a point of withdrawing most administrative resources. Despite that, the Irish managed to cluster existing districts into multimember districts and get on with it. It would be difficult to compare the circumstances of STV’s introduction in Ireland with the current state of the US.

      2. Why not vote-for-one with delegation? I understand that STV could be very difficult to implement and it would be very burdensome to rank many candidates. If 15 candidates (5 D, 5 R, 5 others) ran in a 5-seat district, supporters of the others would probably have to rank lots of candidates to avoid ballot exhaustion. Counting ballots with hundreds or billions of unique rankings would be a hassle.

        With OLPR, supporters of the others would be concerned about whether they’re throwing away their vote with their vote going nowhere if their party isn’t viable.

        With delegation of some sort, a candidate could choose a rankings, full or truncated, for any surplus votes or after elimination. This would allow third parties to become viable without worry of wasted votes, because third party candidates could delegate their votes to a more established party if their party fails to reach the threshold. Counting votes would be easier with only 15 unique ballot rankings.

        1. Paul, thanks for your comment.

          Delegated STV (once known as the ‘Gove system’) is covered in more detail here, by Kristen Eberhard at Sightline.

          As to the point about spoilers/effective ballots under OLPR, Drew raised this in what was, I think, the first comment on the post. I hear you both. My proposal isn’t designed to deal with that (largely theoretical) problem. The purpose isn’t to ‘blow up’ the two-party system. It’s to have seat allocation better reflect the popular vote — a goal, I think, likely to be shared by the people now positioned to impose a PR system.

  6. A year and a half later, I re-read this thread and Matt Shugart’s related thread at Fruits and Votes. I understand the broad strokes of the idea of “nomination districts” (candidates get selected in districts the size of the current SMDs but then run on slates larger multi-seat districts). But, even after reading everything twice, I still find the mechanics hard to follow. I have no doubt that this proposal is easier for election officials to implement. But, damn, in comparison with it STV look easy for voters to understand.

    1. The nomination districts were included primarily to satisfy proponents of “top X” elections. Basing the proposal in some way on single-seat districts also might have helped to accommodate the current regime of majority-minority/minority-influence districting.

      My own preference would be to to return completely to nomination-by-caucus. I can think of few political scientists who would say that nomination-by-primary was a good idea.

      As for any desired “geographic connection,” see the point about ‘bailiwick’ campaigning… and/or recent literature on that connection even under closed-list PR.

      None of this, at least in my view, is meant or even necessary for voters to understand. And I think that a serious examination of STV in practice will show just about as much going on ‘under the hood.’

  7. I think it would be very important for voters to understand how this is supposed to work. Here’s why. When presented with the second-round open list ballot, many voters will instinctively choose from among the candidates nominated in their own “nomination districts”. They will wonder why those names are mixed up with a bunch of names they don’t recognize from adjacent districts, and they will think the ballot looks screwy. But they won’t do anything to find out why. I can’t see how a sizeable minority of voters behaving in this way wouldn’t undermine the goal of proportionality.

    On satisfying the proponents of Top X, I understand the point. But I’m still beyond reluctant to go along, other than recognizing the need to design systems that are equally fair to independent candidates and party-nominated candidates.

    1. Many voters choosing their nomination districts’ picks is exactly what I’d expect. STV ‘bailiwick’ campaigning works similarly. One of the key arguments for OLPR is (and was in the 1930s) that it works like STV does when STV is working well.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.