The Hardball Case for Democrats Abolishing Congressional Districts

How can Democrats respond to Republican attempts at consolidating power and locking out even future Democratic majorities? For example, aggressive Republican gerrymandering may subject America to perpetual minority rule even in the branch of government (the House of Representatives) meant to be most majoritarian in character. But what can Democrats do to stop it?

People often talk about "hardball" options, as if the only reason Democrats aren't acting to defend themselves is a failure of steely-eyed will. Admittedly, that is a vice some Democrats have. But it's not the only issue. A particular problem with many hardball tactics is they invite tit-for-tat retaliation. If Democrats engage in court-packing, for instance, Republicans can just re-pack the courts again the next time they take control of the presidency and Senate. By contrast, one advantage of DC statehood is that it is relatively immune from direct retaliation -- there are no obvious GOP-leaning states that can be admitted in response.

So the ideal hardball tactic is one that Democrats can use (a) in states they control, without Republican (or, perhaps as importantly, Sinema/Manchin) permission, and (b) where Republicans cannot easily respond in kind. And so here's my pitch, more as a thought experiment than anything else: in solid blue states, Democrats should abolish congressional districts entirely, elect all House representatives at-large without any form of proportional representation.

Right now, for example, California has 53 House seats (soon to be 52). Some of these districts are Republican, some are competitive, most are Democratic. Currently, California's delegation comprises 42 Democrats and 11 Republicans. But since California is reliably Democratic at the statewide level, if all representatives were elected at-large (without any nod to proportional representation) we can assume all 53 would be Democrats -- a net gain for Democrats of 22 seats.

Of course, by stipulation I say that Republicans are allowed to retaliate, and so if California passes this law, so does Texas. Texas currently has 23 Republicans and 13 Democrats;* but if all are elected statewide and Texas retains its red hue then all 36 seats would go GOP -- a net gain of 26 for team red (If this makes you think California more gerrymandered than Texas is, recall that Trump won Texas in 2020 by approximately 5.5%, while Biden won California by 29%).  And then there are the states where this doesn't matter, because they're already all-blue (Massachusetts) or all-red (Oklahoma). Switching to at-large would yield the same outcome as the status quo, just without defined districts.

So doesn't it wash out? Not if you play it out, no.

Let's make two simplifying assumptions to start: first, that all House seats in a state-wide at-large race will go to one party (there will be no ticket-splitting), and second, that each state will vote for the party it voted for in the 2020 presidential election. The first of these should under conditions of strong polarization remain true enough (and idiosyncratic exceptions should cancel out). The second is obviously dicier (what if we're in the universe of 2016 instead of 2020?), and I'll address it in more detail in a moment. The result would be basically be the same as the electoral college outcome if we remove DC and the distorting impact of the Senate (recall that every state gets two extra electoral votes from their Senate seats, regardless of population).

Right now, the House of Representatives is Democratic-controlled by a 220-212 margin (with three vacancies). But if all states voted at-large under the above assumptions, the House breakdown after the 2020 election would 253-182 -- a huge Democratic swing.

Now, of course, it is hardly guaranteed that the 2020 election results will replicate themselves in future House elections. Georgia, for instance, went Democratic at the presidential level but had voted for a GOP governor just a few years prior. It would be foolish for Democrats to pin their House majority hopes on Georgia reliably being a blue state and thereby giving all of its seats to Democrats. There is a live possibility, after all, that it goes red, then under my above assumptions all of its seats would go Republican instead. Ditto states like Michigan or Wisconsin or Pennsylvania. In 2016, for example, the House under the above model would have had a 246R-189D split.

So let's complicate the model slightly. In swing states (which we'll define as states where the margin of victory in 2020 was less than 5%), we will assume that the state will not adopt the at-large system but instead will prefer the (typically gerrymandered) status quo as the risk-averse option. For example, right now Florida has a 16-10 Republican advantage in the House notwithstanding that Trump won Florida by just 3.3%. The GOP-controlled Florida legislature might be willing to roll the dice on the at-large system in the hopes of getting all 26 seats; but of course doing so runs a non-negligible risk that they might lose all 26 seats. Better to preserve the status quo where they can, by redistricting, guarantee themselves most seats rather than go for broke. The case is even clearer in Georgia, where Republicans hold an 8-6 advantage in the House delegation in a state Biden won (albeit by a sliver). They're already getting more than they should via gerrymandering, why take a risk and potentially lose everything? In Nevada, Democrats face the same prospect in the opposite direction: they already have a 3-1 lead in the House delegation -- why risk letting the GOP run the table in a good year just to get one more seat?

If only the non swing states use the at-large system (while the swing states preserve the status quo and vote the same as they did now), the 2020 House margin would be 235-200 in the Democrats favor. The forty-two non swing states would break down 186-142 for team blue. The remaining eight swing states are Arizona (4R/5D), Florida (16R/11D), Georgia (8R/6D), Michigan (7R/7D), Nevada (1R/3D), North Carolina (8R/5D), Pennsylvania (9R/9D), and Wisconsin (5R/3D) -- this totals 58 Republican seats and 49 Democratic seats.

Pictured: The 2020 presidential map, with "very close" states greyed out. Note that each state's number of electoral college votes is two more than its number of House seats (House seats plus Senate seats). The 2016 map is below.

What about 2016?

There were even more very close states in 2016 than 2020 -- 11 were decided by five points or less. Of course, states can't necessarily predict in advance that they'll be close (who saw Minnesota coming?). But again, if we assume that only the non-close states would use the at-large (functionally) winner-take-all system, that would start us off with 167 Democrats and 156 Republicans. The eleven swing states were Arizona (4D/5R), Colorado (3D/4R), Florida (11D/16R), Maine (1D/1R), Michigan (5D/9R), Minnesota (5D/3R), Nevada (3D/1R), New Hampshire (2D/0R), North Carolina (3D/10R), Pennsylvania (5D/13R), and Wisconsin (3D/5R). That yields a final result of 223R-212D (thanks to a whopping 67-45 advantage in the swing states). Still a GOP win, but much narrower than its actual 2016 margin of 241-196 (and, in fairness, the GOP -- barely -- won the House popular vote in 2016). Also note that two of those states, Pennsylvania and North Carolina, were forced to revise their lines shortly after the election -- it is likely that the GOP lost enough seats just from those rulings such that, under the reformed lines, Democrats would have been able to win a narrow majority even in 2016.

So this is not a "Democrats always win" solution, by any means. But it does offer Democrats some advantages. For one, it allows Democrats to fully leverage their advantage in larger states where they are leaving more "meat on the bone", so to speak. Many solid red states already have all or nearly all GOP delegations -- switching to the at-large system wouldn't change much in, say, Oklahoma or Utah, which currently have no Democrats in Congress at all and where Democrats in their best year maybe could squeak out one winner. Big solid blue states like California, New York, and Illinois would be rich prizes. Outside Texas, there aren't that many comparable opportunities for the GOP. Ohio would be a solid possibility, but Ohio is already gerrymandered so ludicrously aggressively (11R/3D) that the at-large switch wouldn't actually do that much -- just a six seat swing. Compare that to New York, where going all blue from the current 19D/8R status quo would net Democrats 16 seats.

But the other reason it works is because it neutralizes the specific GOP advantage in gerrymandering swing states. A major reason the House is so close right now is that the GOP has a nine seat advantage in the eight 2020 states that were decided by five points are fewer, even though Biden won six of them. In 2016, the Pennsylvania GOP's "gerrymander of the decade" gave them a 13-5 House advantage in a state that Trump won by less than one percent. That is largely the product of extremely favorable (to say the least) districting lines. Abandoning those lines for an effectively all-or-nothing at-large system would be incredibly dangerous for the GOP. But without going for it in these states, Republicans would be hard-pressed to overcome Democratic advantages in populous, deep blue states like California.

Again, as much as this might seem like stacking the deck, we should note that all that this system really does is make it more likely that the party which gets the most votes controls the chamber that is intended to be most responsive to majority preferences. The above analysis is fancy footwork that boils down to "under this system, the party with the most votes is most likely to win". In 2016, under the modified model where the swing states are risk-averse, Republicans would have narrowly won a House majority after narrowly winning the (House) popular vote. In 2020 under that model, Democrats would have secured a wider House majority after earning a wider (House) popular vote win. This is a good thing.

That said, putting aside its tactical utility as a hardball play that forces recalcitrant players towards more robust democratic solutions, do I think abolishing congressional districts is a good (as in virtuous) idea? That is, would I support it on its own terms, regardless of its usefulness in counteracting GOP gerrymandering? Honestly -- not really. There are good reasons to have politicians represent smaller geographic districts to whom they feel particularly accountable towards -- someone looking out for Fresno or Tacoma or Springfield specifically. An at-large process could still account for that somewhat -- the Democratic "slate" in California could self-consciously include figures from around the state who would hold themselves out as responsive to the needs of a particular community and would take point in responding to local constituent concerns. But there's no doubt there'd be a genuine loss there.

My preference, then (to the extent we're moving this beyond "thought experiment") is for this proposal to be expressly set to sunset at the moment there are uniform federal rules governing redistricting (and forbidding partisan redistricting). Basically, it tells Kevin McCarthy "agree to national rules on voting rights or you can kiss your precious California House seat goodbye". If he agrees to cooperate, lovely. If he doesn't, well, then you make good on the threat.

[Note: It took me about an hour to write this post and then another two to check and recheck my math. I'm not a math guy, so I can't guarantee the math is perfect now -- but I think it should be close. If I made any gaping mistakes, please let me know.]

UPDATE: Turns out that federal law (2 U.S.C. 2c) appears to forbid this, insofar as it requires states establish an equal number of districts to the number of representatives they're entitled to (with each district only electing one representative). I suppose one could try to skirt this by establishing multiple "districts" which overlap the same geographic territory (or better -- detach districts from geography altogether and randomly assign voters to districts). But alas, seems like this thought experiment must stay firmly in the realm of the thought.