You know the Senate is unrepresentative. But it’s worse than you think.
You may have heard that the 51 senators who voted for Kavanaugh represent only 44% of the country. That’s wrong; it’s more like 31%. 39% of voters are represented only by senators who voted against Kavanaugh; and 30% of voters have no real representation in the Senate whatsoever. Here’s why.
In our system, each Senator supposedly represents their entire state. The unabashed partisanship of the Kavanaugh battle puts the lie to that pious fiction. Senators represent the people who voted for them.
Take North Carolina as an example.¹ They have two Republican senators: Richard Burr, last elected in 2016, and Thom Tillis, last elected in 2014. They got 2,395,376 and 1,423,259 votes, respectively. But clearly many people voted for both of them. A reasonable back-of-the-envelope calculation for the total number who voted for either one of them is to add 8% of the smaller number to the larger one,² giving about 2.5 million voters who are represented by either or both of those senators. A similar calculation for the state’s total gives 4.9 million.
I can already hear the objections from some political theorists to this calculation. Our system was built³ as a democratic republic, they’d say; so senators represent all their constituents, not just the ones who voted for them. That’s a nice theory. But if we actually believed in that theory, each state would elect its two senators using a proportional representation method. Under that kind of system, we could guarantee that at least two-thirds of voters, and probably closer to 90%, would have equal representation in the Senate by a senator they actually supported and had helped elect.⁴ In such a system, the idea that Senators are meant to represent the whole state would be theoretically defensible; but not in the current, broken, system.
Or let’s put that in more concrete terms. Say you were a Republican in Massachusetts who opposed Kavanaugh. You couldn’t call Susan Collins and ask her to support him; you aren’t her constituent. And the senators from your own state are paying attention to their supporters, not to you. The same basic story goes if you were a Democrat in Texas who supported him.
Here’s a spreadsheet which applies this calculation to the cloture vote, where Kavanaugh’s 51 votes included all the Republicans, minus Lisa Murkowski (R-AK), plus Joe Manchin (D-WV). By this calculation, 52.1 million people voted for one or two senators who supported Kavanaugh; 41.3 million people voted for one or two senators who opposed him; and 39.4 million people voted for Senate in the last 6 years, but only for losing candidates.
Is this fixable? Unfortunately, any change to the Senate would require a constitutional amendment; almost unthinkable in the current partisan climate. But if you could amend the Constitution, fixing this would be pretty simple. First, amend the Constitution so that both senators from any given state are elected in the same year. Then, pass a simple federal law (not even a constitutional amendment) to use a proportional voting method⁵ to elect those two senators. With current partisan breakdowns, every single state would elect one senator each to the Republican and Democratic caucuses. The Senate would be almost permanently split 50/50, with the vice president — elected nationally—as a tiebreaker. Senate rules would have to be drastically streamlined to avoid supermajority thresholds and to allow absent senators to assign proxies (generally to other senators from their party), but the original intent of the senate — as a deliberative body that defends the interests of the states, not a partisan body that is either a rubber-stamp or a universal veto depending on whether government is divided—would be restored.
As I said, the above solution is unlikely to happen. At best, we may get a few new states (DC and PR?) that help narrow the mean-median gap in partisan and ethnic terms. But something at least as hard as the above is what it would take to have a Senate that actually followed from any coherent philosophy whatsoever, be it the founders’ or anyone else’s.
¹I chose NC because, of the states with two Republican senators, it’s closest to the (population-weighted) median in terms of both partisan breakdown and size.
²The underlying assumption here is that on average 8% of the electorate changes between the elections of the two senators in any given state; and that in states with two senators of the same party, voters do not switch parties between senate elections. How did I get the number 5%? Well, the average voter age is around 48, and the life expectancy at 48 is around 34 years; which would suggest the average duration of being a presidential-year voter is in the ballpark of 50 years. The average time between last election of two sitting senators of the same state is 2.67 years, so that’s 5.3% turnover on average purely from demographic turnover. Add in people moving from state to state, and a bit of a cushion for party-switching voters, and 8% seems like a reasonable floor.
³This very formulation of “our system was built as…” is problematic in this case, of course. Most of the founders (including both Madison and Hamilton) saw the state-centric senate as a distasteful compromise that was worth it in order to get the constitution approved. I think they would have expected us to have adjusted this system several times by now. And indeed, the 17th amendment significantly changed the basis of Senate power, to the point where “original intent” becomes an inherently slippery concept.
⁴If you’re electing N seats, a good proportional representation method can guarantee that N/(N+1) of the voters are equally represented. So with 2 senators per state, that’s 2/3. In the current House, which averages almost 9 seats per state, the average voter could be guaranteed an almost 9/10 chance of getting full and equal representation. And even with just 2 seats, you can get up to 100% representation in those states which happen to have a 50/50 party split.
⁵For reforming House elections, the most practical and viable proportional is PLACE voting. But in this case, for the Senate, you’d only be electing 2 winners at a time, so ballot complexity is not a big issue. In that situation, STV (also called multi-winner RCV) actually works pretty well. (Bucklin Transferable Voting would be a little bit better, but STV is better known and the difference is minor.)