This past April marked the 107th anniversary of the ratification of the 17th Amendment to the United States Constitution. As far as amendments go, it’s not exactly one of the better known — people invoke their First, Second, and Fifth Amendment rights all of the time, but you don’t hear much about the 17th, even if its absence would vastly alter civic life in America. That’s because the 17th Amendment allowed for the popular election of United States Senators directly by the citizens of each state, a right that was previously held by each state’s legislature instead. 

As originally laid out in Article I of the Constitution, it was the state government, and not the state populace, who elected senators. The original purpose of this provision was to draw a clearer contrast between the Senate and the House of Representatives, the latter of which has always been elected by popular vote.1Even this right evolved over time as well. Initially only white, land owning males were granted the right to vote, but this eventually grew to include all white males, and then (ostensibly) all males regardless of race, and then all adults over the age of 21, to almost all adults over the age of 18 today.
By being selected by a state’s legislature, the Framers envisioned that the Senate would become a true upper house composed of, in the words of Constitutional scholar Todd Zywicki, “better men” who would expertly steer legislation when the lowly, directly-elected House became too influenced by special interests to govern properly. Information on how exactly each state selected their senators is scant, but according to the Constitution Center, it began with each chamber of a given state’s legislature nominating a candidate for the Senate. If both chambers nominated the same candidate, then the new senator would be sent to Washington. But if they each nominated a separate candidate, they met in a joint session to compromise on a choice.2Nebraska is the only state that has a unicameral nonpartisan legislature. Presumably whoever won the first vote in the Nebraska Legislature would be elected senator. This system worked fine for the most part, but as time wore on its flaws became more evident. According to UNLV law professor J.S. Bybee, one of Delaware’s Senate seats was left unoccupied from 1899 to 1903 due to a deadlock in the state’s legislature, while in 1897, one-third of the Oregon state legislature refused to take the oath of office due to a dispute surrounding an open Senate seat. These disputes dominated the business of the legislatures; in 1895, the Delaware State Legislature cast 217 ballots, and yet still failed to fill their open Senate seat. Sometimes the seats that were filled were done so illicitly, such as when Montana Senate hopeful William A. Clark bought legislators’ votes, a crime that resulted in his removal from office. Calls for reform started relatively early in the young republic, but came to a head during the Progressive Movement at the turn of the century. After several states found workarounds by holding public non-binding primaries to inform their state legislatures and the House of Representatives began to press for change, in 1911, Congress passed a joint resolution proposing a constitutional amendment for the direct election of senators, and two short years later 36 states ratified the amendment to make it law. 

And yet, there are still some who think that the ratification was a mistake. In 2016, the Utah legislature passed a resolution asking Congress to repeal the 17th Amendment,3 In a nice bit of symmetry, Utah was also the only state to reject the 17th Amendment back in 1913.
and a quick Google search will reveal that there’s a faction of conservative and libertarian thinkers that support giving the power to elect senators back to the state governments. The stated goal of most of these efforts is to restore the balance of power between the federal government and the states, with the added bonus that investing such power in a state’s legislature may even encourage renewed interest in state level elections and, in the words of Utah State Senator Al Jackson, ensure that senators would no longer be “more beholden to special interests than their states.”

Fringe as these opinions may seem, they did get me to thinking about what exactly the Senate would look like if the 17th Amendment were never ratified. Using Ballotpedia’s record of each state legislature’s composition going back to 2014, I created a massive spreadsheet that compared the composition of our current Senate with the composition of each senator’s respective state legislatures at the time, and figured out who would be in and who would be out if the state legislatures still chose the senators. More than just a fun thought experiment, it seemed like the best way to grapple with the arguments of those who would want to see the 17th Amendment repealed. Maybe by modeling what the Senate would look like if that did happen, I would come around to seeing their point of view, or at the very least, have a more substantive reason for disregarding it.

In order to figure out what the Senate would look like in a world where the 17th Amendment never happened, we have to make a few big assumptions, namely: 

1. We must assume that each state legislature elects senators through the same process. We’ll assume that each chamber holds a vote to fill the position, and if the same person wins a plurality of votes in each chamber, they are elected to the Senate. In the event that the chambers select two different people to serve as senator, a special joint session is convened wherein the bodies vote together. Whichever candidate has the most votes after this joint session is elected to the Senate. 
2. This also means that we’re assuming that votes are being cast along strict party lines. That means no West Virginia Republicans crossing the aisle to vote for Joe Manchin or any Maine Democrats voting for Susan Collins. I’d love to poll every state legislator in America and find out who they’d actually vote for, but I’d also like to finish writing this piece before I turn 50, so this assumption is a necessity. One trickle down effect of this assumption is that a joint session is only necessary when the two chambers of a given state legislature are controlled by two different parties. 
3. We’re also assuming that legislators will be voting for the candidates who ran for the Senate seat in question in real life. This is probably the most unrealistic assumption we’re making. In an actual 17th Amendment-less world, outsider politicians with ideologies that lean further to the right or left, such as former Alabama Supreme Court Chief Justice Roy Moore or Vermont Senator Bernie Sanders, would probably be passed over by state legislators in favor of less controversial figures. But since I can’t figure out a way to find out who those less controversial figures would be, those candidates who occupy the furthest reaches of either side of the American Overton window will have to do for now.
4. We’re also assuming that senators are elected after a new state legislature is sworn in. In other words, if a state legislature holds elections every two years, the group of legislators elected in 2018 would determine who fills the Senate seats up for election in 2018. Otherwise, you would have a scenario in which a group of legislators elected in 2016 would be voting to fill a Senate seat in 2018 while their own seats are up for reelection. If states elect their legislatures in years that don’t run concurrent with Senate elections, we’ll assume that the legislature in office at the time of the Senate election will elect a senator (i.e. the Virginia State Assembly elected in 2017 would vote to fill the Senate seat up for election in 2018).
5. There are two independent senators in the 116th Congress: Bernie Sanders of Vermont and Angus King of Maine. Both independents caucus with the Democrats, but only Sanders was endorsed by his state’s Democratic Party. During the 2018 election, King had to face both Republican Eric Brakey and Democrat Zak Ringelstein, the latter of whom was the only major party senate nominee to be a dues-paying member of the Democratic Socialists of America. So while it may seem more likely that the Democrats in the Maine State Legislature would choose to back the centrist former governor, for the sake of simplicity we’re going to assume that their votes will be going to Ringelstein instead. Since no Democrat ran for Sanders’ Senate seat in Vermont, we’ll assume that the Democrats in the Vermont State Legislature will have voted for Sanders.

All of that make sense? All right then, let’s fire up the ol’ spreadsheet and take a look at some tables!

Right off the bat, 84 of the 100 elections (or in the case of the governor-appointed Arizona Senator Martha McSally and Georgia Senator Kelly Loeffler, selections) that make up the current Senate would go the same way no matter if the 17th Amendment did or did not exist. Of those 84 senators elected in both scenarios, only seven faced a split legislature and had to be elected through a joint session:

Even though the end results of these elections are the same as the ones in our universe, the votes still break down in interesting ways. Tom Udall, who defeated Allen Weh in the 2014 New Mexico senate election by 11 points, just squeaks past him by a measly four votes in our model, or just over a 3.5% margin. Senate Majority Leader Mitch McConnell, who in real life won Kentucky’s seat by a comfortable 16 points against challenger Alison Lundergan Grimes, bests her by only six votes here, a little over a 4.4% margin.

What could explain this discrepancy? Well, all of these senators were incumbents at the time of their reelection, so that puts them at a natural advantage. But a more likely explanation seems to be that post-17th Amendment elections are decided by population, while pre-17th Amendment elections would be determined by representation. Think all the way back to your middle school civics class and the Connecticut Compromise. In the popular understanding, during the Constitutional Convention, smaller states like New Jersey wanted states to be represented equally in the national legislature regardless of size or population, whereas larger states like Virginia wanted the number of a state’s representatives to be doled out proportionally, so that those with a larger population had more sway. Eventually, the Connecticut delegation presented a compromise of the two plans that provided us with our current bicameral legislature, consisting of a Senate in which each state gets two senators, and the House of Representatives where each state is allocated a number of representatives based on population.

Even though Americans have been living with the House and the Senate since the Constitution’s ratification, the Connecticut Compromise still remains controversial for its unintended consequences. Former Michigan Congressman John Dingell, the longest serving congressman in American history, argued for the Senate’s abolishment in 2018. His reasoning? By giving Wyoming and North Dakota the same number of senators as California, the Senate provides “sparsely populated, usually conservative states” with an outsized role in shaping legislation in a way that may run counter to the wishes of the majority of the American people. For example, in the 2018 Senate elections, Democratic candidates received almost 20 million more votes than their Republican counterparts. But because senators are elected to represent a population within an arbitrary border drawn decades or even centuries ago instead of districts of roughly equal size, the Democrats still posted a loss of two seats that same year. Whether you agree with Dingell’s proposal or not, it’s hard to argue that smaller states aren’t overrepresented in the Senate, and that this is neither liable to change nor unintended. The House’s method of representation is prone to abuse in the form of gerrymandering, sure, but at least the courts can force districts to be redrawn. No court has the power to redraw a state.

This is all a long-winded way of saying that I suspect the phenomenon described by Dingell is likely also manifesting at the state level, and therefore having an effect on our hypothetical Senate elections. The post-17th Amendment election process allows large, liberal population centers such as New York City and the Twin Cities to tip the scales in favor of Democrats like Chuck Schumer and Amy Klobuchar. But when you reduce the size of the electorate from 19 million people to only a couple hundred, the relative power of those smaller, more conservative districts grows, making the election much closer.

Ernst and McConnell ran into a different version of the same problem. While they may have trouble competing in Democratic enclaves like Iowa City or Louisville in their respective home states, these cities aren’t large enough to swing an election towards their Democratic opponents. But these municipalities’ influence is stronger within the state legislature, allowing them to make a pre-17th Amendment Senate race more competitive.  

So we can already learn a little bit about the pre-17th Amendment electoral process’ advantages and disadvantages by looking at elections that would have been a bit tighter but still ended with the same result. But that’s not why you came here, so let’s get to the good part: What seats would be flipped if we lived in a pre-17th Amendment world?

Before I show you this, I have to ask, if you’re a Democrat, if you really want to see the results. I can assure you that it’s interesting and that my analysis is excellent, but I will not be held responsible for any heart attacks or depressive episodes that occur as a result of looking at the following table. 

You’re still sure you want to see it? Alright, here it is: 

For those of you keeping score at home, 16 seats have changed parties. And of those 16 seats that were flipped, a whopping 13 were won by Republicans, with only three going to Democrats.5And only two of those seats would represent true gains for the party. As I mentioned in our list of assumptions before, Maine’s Zak Ringelstein takes the place of Angus King, an independent who caucuses with the Democrats. The result is a net gain of zero for the party. That gives the Republicans an edge in the upper chamber by a count of 64-36 — in other words, a filibuster-proof supermajority that would also represent the largest Republican Senate majority since Reconstruction. It also leaves the GOP only three votes shy of the two-thirds majority needed to override a presidential veto, so even if Donald Trump were to lose reelection in 2020, they could still make his replacement miserable if they made some gains in this year’s state legislature elections.

As far as why we’ve ended up with such a strong swing towards the Republicans, it’s likely that the same logic we applied to the first table applies to this one as well. The over 1.5 million Philadelphians who may have put Bob Casey Jr. over the top in 2018 outnumber their more rural and conservative neighbors in the event of a simple head count, but when the folks who draw up Pennsylvania’s legislative districts try to balance the two regions, urban centers will have a disadvantage. The same could be said of Democratic losers from Arizona, Ohio, Michigan, and Wisconsin who see the strength of their metropolitan bases blunted in their state legislatures.

The losses of red state Democrats like Montana’s Jon Tester or West Virginia’s Joe Manchin likely have simpler, more qualitative explanations. Tester won by casting his opponent as an outsider and a carpetbagger, whereas Manchin had the benefit of being one of the most Trump friendly Democrats in the Senate. Those qualities helped make them appealing to Republicans and right-leaning independents, but in a straight party line vote they become functionally irrelevant. So repealing the 17th Amendment not only levels the playing field in terms of urban/rural divide, it also flattens out the idiosyncrasies of individual campaigns and candidates as well. Again, we’re making a lot of assumptions here — there’s a good chance that the realities of a 17th Amendment-less world would push a conservative Democrat like Manchin to become a Republican instead — but our results so far point to a Senate whose composition is much more predictable, and much more red.

So we (kind of, sort of) know what the Senate would look like if we repealed the 17th Amendment. But there’s still one last question I’d like to try and answer: would repealing the 17th Amendment be a good idea?

Based on our hypothetical scenario alone, the answer would seem to be an easy yes for Republicans and a strong no for Democrats. But let’s take the politics out of it for a moment and think about whether or not the pre-17th Amendment system is a fair way to decide an election. Those who think we should repeal the 17th Amendment argue that it gives the states a certain degree of power over a federal government that holds an inordinate amount of power over them and would also weaken special interest groups. Those who think we shouldn’t would point to past corruption and inefficiency, not to mention a clear violation of the democratic ideals the country was founded on, as reason enough not too. But I think there’s an even more important question to ask: is the way we choose state legislators — and by extension in this hypothetical situation, senators — fair as well?

Virginia offers a fascinating case study in the way we elect these bodies. Following the state’s 2011 state senate elections, the chamber was split evenly among party lines, with 20 Republicans and 20 Democrats taking office. Just as the Vice President is the President of the United States Senate, so is the Lieutenant Governor the President of the Virginia Senate, and since Lieutenant Governor Bill Bolling was a Republican, that gave the GOP an effective majority of one. And in a world without the 17th Amendment, that means that only one vote, cast by a man who wasn’t elected to the state legislature, would determine that chamber’s nomination and, since Republicans also controlled the House of Delegates, who would represent Virginia in the Senate (in this case, former counselor to George W. Bush and eventual gubernatorial nominee Ed Gillespie). To be fair, under our rules even if Virginia’s Lieutenant Governor at the time was a Democrat, a Republican would still win the seat by virtue of gaining more votes in a hypothetical joint session. But if we step outside of our experiment for a bit, and consider a world where state legislators are allowed to cross the aisle and vote for candidates outside of their party, the prospect of a Lieutenant Governor tipping the balance of a Senate election begins to call the legitimacy of the system into question.

The 2018 Senate election offers an even more chastening example of a state legislature’s questionable ability to accurately represent the electorate. The 2017 Virginia House of Delegates election — which, according to our rules, would elect the body that would elect a senator in 2018 — was perhaps one of the most fraught in recent history. A number of election results were challenged amid allegations that provisional ballots were not counted in certain districts and that some voters had been assigned to incorrect districts entirely, leading to a situation where the very integrity of the election was cast into doubt. The closeness of the elections and their alleged irregularities spawned five recounts, one of which took place in the race for the 94th District, which was being contested between Democratic challenger Shelly Simonds and Republican incumbent David Yancey. The recount showed Simonds defeating Yancey, but the relevant courts declined to certify it, meaning that no winner was declared. The Virginia House of Delegates’ solution for breaking such a tie? The drawing of a random lot, which Yancey wound up winning. As if reducing an election to a coin flip wasn’t bad enough, it was for a seat that determined the balance of power in the House of Delegates, giving the Republicans a one vote majority. So according to our model, the Virginia House of Delegates would’ve elected Corey Stewart — a former chairman of the Prince William Board of County Supervisors perhaps best known for his staunch opposition to the removal of Confederate monuments and symbols and courting of far right groups — to the United States Senate.

Of course, there’s an argument to be made that someone like Stewart — an insurgent who took as many shots at Republicans as he did Democrats — would never have been a viable candidate in a non-17th Amendment world. But even if state legislators would be a moderating force in Senate elections, there’s still no guarantee that they’d act in the best interest of the people. After Democrats Roy Cooper and Tony Evers were respectively elected governor of North Carolina and Wisconsin, each of their Republican controlled legislatures passed a series of laws that weakened their executive powers. This blatant circumvention of a popularly decided election is bad enough, but imagine how it might play out in the context of electing a senator. Sure, Mitt Romney would probably have had little trouble convincing the heavily Republican Utah State Legislature to elect him to the senate seat Orrin Hatch vacated in 2018. But if Romney were beholden to what is essentially a collection of low level party elites instead of the voters of his state, would he have felt free to act on his conscience and  vote to remove Donald Trump from the presidency as he did earlier this year? Even if he had, isn’t it more than likely that the legislature wouldn’t elect him back to the Senate in 2024? We can only speculate, but given how willing the North Carolina and Wisconsin legislators were to push the limits of their authority when they felt threatened, it seems more likely than not that Romney’s vote in February would have cost him his political career, and as such, may not have even happened in the first place.

And, to be fair, Romney may still lose either renomination or reelection four years from now. But the beauty of our current system is that it’ll be a population of over 3 million Utahns, instead of 104 state politicians, that will decide Romney’s fate. It’s true that the wisdom of crowds can at times be anything but wise — for every Roy Moore defeat, there’s a victory for the publicly admonished New Jersey Senator Bob Menendez, or for segregationist senators like John C. Stennis and Jesse Helms who held onto Senate seats long after the Civil Rights and Voting Rights Acts were law. But, barring an expulsion-worthy offense, it should be the voters themselves who determine when a senator no longer deserves their office. The voters should be allowed to break out of the narrow and partisan parameters of the pre-17th Amendment system and feel free to elect an independent candidate like Angus King, or, happened in Alaska in 2010, reject the results of the party primary all together.

After losing the Republican primary to Tea Party Activist Joe Miller,6Yes, the same Joe Miller who ran against Murkowski six years later as a Libertarian. Apparently the guy held a grudge.
incumbent Senator Lisa Murkowski ended up winning the general election through a write-in campaign, a result that would have been impossible (and I will grant, perhaps unnecessary) had the 17th Amendment never been ratified. In such a world, the writer in me would’ve lamented the loss of a great story, that of an electorate choosing to correct its own decision through one of the most directly democratic actions they could take. But the citizen in me would have lamented the fact that the people of Alaska were denied the opportunity to choose between Miller, Murkowski, and their other opponents in the first place. Limiting who elects a senator limits who can become a senator and, were such a system to be reinstated, our civic life and government would be far poorer for it.