Monday, March 14, 2016

Intra-Party Democracy and the Median Voter

by Michael Dorf

Both Democrats and Republicans currently face questions about how small-d democratic their nomination processes are and should be. On the Democratic side, Bernie Sanders supporters worry that even if their man wins more pledged delegates than Hillary Clinton does, her edge in so-called super-delegates could nonetheless give her the nomination. It takes 2,382 delegates to win. There are 712 super-delegates. As the "establishment" candidate, Clinton will end up with the vast majority of them. So, to use some random numbers, suppose that Clint ends up with 670 super-delegates and 1712 pledged delegates. She would just barely win the nomination even though she would have won fewer than 43% of pledged delegates to over 57% for Sanders. That's undemocratic, right?

Meanwhile, on the Republican side, in recent days the party establishment has inched closer to making peace with the possibility that Donald Trump could be the nominee, but depending on, among other things, the outcome of tomorrow's winner-take-all primaries in Florida and Ohio, that could change. There are two scenarios in which the primary winner could be denied the nomination. In the standard scenario, no candidate has enough delegates to secure the nomination on the first ballot, but on a second or subsequent ballot, votes coalesce around a candidate who did not come into the convention with the most first-ballot delegates. In the more far-fetched scenario, the rules committee proposes rules changes that the delegates approve before the first ballot, with the result of freeing up delegates to vote for someone other than the candidate to whom they are pledged.

As readers and I discussed in some back-and-forth in the comments on a post last week, it's hard to tell in advance how loyal delegates will be to the candidate to whom they are pledged. The national rules leave to state-level decision makers questions about just who the delegates are. Depending on the state-level rules and the skill of the campaigns in selecting delegates, a Trump delegate might or might not be loyal to Trump. So even if Trump has a majority of delegates for a first ballot, he might not be able to discipline them sufficiently to prevent a majority of delegates at the convention (including some of "his" delegates) from voting to change the rules to allow defections on the first ballot. If that were to happen, Trump would no doubt cry foul and might even file multiple lawsuits. Even if he were to lose those cases, he would be right that the rules change would be undemocratic, right?

The answer to both questions is "yes and no."

Let's focus on the Democratic example. The ability of super-delegates to vote at the Democratic National Convention undoubtedly makes the nominating process less representative of the will of Democratic primary voters and caucus-goers than it would be if there were no super-delegates. So yes, super-delegates render the intra-party selection process somewhat undemocratic, and the more super-delegates there are relative to pledged delegates, the more undemocratic that process. But when my Sanders-supporting Facebook friends complain that the process is undemocratic, they mean thereby to convey a negative normative judgment: democracy is good; to the extent that the process is undemocratic, it's bad.

But just because a process is undemocratic doesn't necessarily mean it's bad. Because I'm a constitutional lawyer, I think of examples of courts striking down laws that enjoy democratic support. When the Supreme Court invalidated state same-sex marriage bans that had been enacted by democratically elected state legislatures or in statewide referenda, the dissenters were right that the majority had acted undemocratically. But those of us who applauded the result in Obergefell v. Hodges thought that the question whether people of the same sex should be permitted to marry is not the kind of question that should be subject to democratic politics. Conservatives abide this principle too, just for different sorts of decisions. The Obergefell dissenters think that it's appropriate for the Court to displace democratically chosen decisions with respect to such matters as affirmative action and gun control.

Of course, in the foregoing sorts of cases, there is a judgment that some other value (whether it's the equal right to marry or the right to bear arms or whatever) outweighs the interest in democratic decision making. By contrast, the selection of a party's nominee is itself an exercise in democratic politics. Is any countervailing value served by having super-delegates? Or--to use the Republican example--by permitting the members to change the rules so as to deny the nomination to the candidate with the most or even a majority of delegates?

Yes. One could plausibly say that some measures that limit intra-party democracy serve a countervailing interest in the democratic representativeness of the electoral system as a whole. Note that this is NOT the interest in limiting democracy to prevent demagogues, invoked by Ross Douthat in his NY Times column on Saturday. Douthat argues that it is legitimate for a system to be designed to frustrate the people's efforts to rally behind an authoritarian, even if the authoritarian would be their top choice (although Douthat also notes that so far Trump is the choice of only about a third of Republican voters and caucus-goers). That may well be right, but here I'm making a different point: Even in an election cycle in which none of the choices is Mussolini or Trump, undemocratic party nominating procedures can serve democracy itself; they need not be invoked just as a brake on democratic excess. Mine is an argument about the potential benefit of undemocratic nomination procedures even in a choice between acceptable candidates (like Clinton or Sanders).

Because the people who vote in primaries and caucuses skew away from the political center, there is a tendency of primaries and (especially) caucuses to result in the nomination of candidates who have the support of each party's core constituencies but who then do poorly in a general election in which independents and moderates (regardless of party registration) play a larger role. Open primaries, in which independents and even members of the other leading party can participate are one mechanism for moderating this effect, although I wouldn't call an open primary "undemocratic." The allocation of super-delegates is, by contrast, an undemocratic intra-party mechanism. It helps pull the party to the center to ensure both the viability of the party and--when both major parties are doing this sort of thing--the centrism of the president, regardless of who it is.

To be sure, recognizing that there is a legitimate countervailing interest in support of undemocratic intra-party mechanisms is not tantamount to endorsing any particular mechanism. In the current race, after all, there is considerable uncertainty about whether each party's "establishment" candidate (Clinton for the Dems, anyone but Trump or Cruz for the GOP) actually would do better in the general than the standard-bearer chosen by the parties' respective voters. But the possible failure of the party-control mechanisms to achieve their intended effect in 2016 is probably a peculiarity of the current election, rather than an indictment of either the existence of super-delegates or the power within the RNC to change its own rules as it goes along (which is another undemocratic means of a party moderating its nominee).

There are, nonetheless, good arguments of a more general nature for questioning party-control mechanisms, even if one thinks that "undemocratic" isn't by itself a show-stopper. On the GOP side, there is something fundamentally unfair about changing the rules near the end of the game. One could say, I suppose, that because it was always known that the rules could be changed in this way, no one could have reasonably relied on the rules not being changed. But that sort of answer proves too much, because it can always be invoked to defeat reliance interests. The truth is that until very recently, it was simply unthinkable that the rules would be changed before the first ballot, even if the rules regarding rules changes technically allowed it.

The use of super-delegates as a kind of ballast to bring the Democratic Party to the center is defensible in theory, although there is a limit beyond which such super-delegates render the primaries and caucuses a sham. I don't know whether 15%--the current percentage of delegates who are super-delegates--can be said to exceed that limit, but if not, it certainly comes close.

That's especially clear if one thinks that it's unhealthy for the two parties to be too close to one another. The median voter theorem--which is simply Hotelling's theory as applied to politics--will lead the parties, if left to their own devices, to clump in the middle. And just as that is a sub-optimal distribution in the economic space, it's at least arguably sub-optimal politically: Voters faced with a choice between two centrist parties have less freedom than voters faced with a center-left and a center-right party, which is what one should get in a system in which party activists exert significant but not total control over the nominating contests. Thus, seen systemically, it makes sense for the parties to moderate their activists' choices, but only a little.

Finally, I cannot help but note that it is tempting to view contemporary American politics as an almost-complete repudiation of what I've just described, because we have nothing like a center-left and a center-right party. We have a center-left party and an extreme-right party--at least as seen from where I sit (on the center-left). I think the explanation for that phenomenon is that contemporary American ideological preferences do not fall into a normal distribution. They appear to be bimodal (i.e., there is substantial polarization) and the curve on the left side of the center is much more sharply peaked than the curve on the right (i.e., there are many more people on the extreme right than on the extreme left).


Joe said...

The process in place is problematic in that you have a tendency select group of each party choosing each candidate - more so than the general election where often only a segment of the population votes - is not truly a cross section given who votes in primaries and caucuses. The type of race skewers things even further.

So, I think there is some cause -- and 15% doesn't seem too much -- to have a mechanism in place to balance things. And, superdelegates include many people already voted for my the people anyways (e.g., members of Congress) adding further democratic legitimacy in that fashion. If a race involves a small segment of the party voters not really providing a complete representative view of the voters as a whole, it is not in a fashion truly "democratic" to rely on them alone.

Judicial review to protect the right to vote or protect self-expression is likewise "democratic" in the long run.

Ben Alpers said...

A few thoughts....

1) It seems to me that the closest GOP equivalent to the Democratic superdelegates, when it comes to undemocratic selection processes, is not the possibility of a last-minute rule change, but rather the Republicans' unevenly non-proportional delegate-allocation rules. And starting tomorrow, states can run winner-take-all contests in the GOP. If trump wins Florida by a single vote with 34% of the vote, he'll still get 100% of Florida's delegates.

2) I think median voter theory is of a little less use than you suggest here. What superdelegates do is give the Democratic party establishment a thumb on the scale. And that's not always a thumb for the median voter. Take free trade, for example. Free trade enjoys the support of much of the party establishment in both parties and is opposed by a large, bipartisan majority of voters.

3) If we want to select our presidential candidates democratically (I think we should, but whether we want to or not is obviously a separate question from establishing the best way to do so), we need to go to a single national primary that would select the candidates directly, preferably via some system of preferential voting. The long, patchwork primary-and-caucus calendar makes no sense whatsoever. The one argument that is sometimes made in its favor -- that it allows outsiders to win via "retail politics" -- is belied by the history of our modern primary system. Outsiders virtually never win. And the outsider who's winning now -- Donald Trump -- is not doing so via retail politics. (And why is selecting for "retail politics" a good thing?) As with many of the other undemocratic features of our political system -- from the equal representation of the states in the Senate to the electoral college -- one major result of our undemocratic system of selecting presidential candidates is that cities, in general, get a proportionally smaller voice in the process than they deserve in strictly democratic terms. And that is, I think, a very bad thing.

DHMC said...

On the Democratic side, this is a nice thought experiment, as it is highly unlikely that Sanders will beat Clinton in the pre-convention delegate count, even if he accomplishes the highly unlikely feat of stopping Clinton from outright achieving the total delegates needed to claim the nomination. The Sanders folks – and the press – seem to have missed the fact that when Sanders beats Clinton, it is, with only a couple of minor exceptions, by a narrow margin, but when Clinton beats Sanders, she generally wallops him. How will Sanders and his followers react if Clinton goes into the convention with a substantial lead in delegates coming out of the primary/caucus process, but she still needs superdelegates to get over the top? We will then see the Sanders’ folks commitment to democratic – small ‘d’ – principles.

Two important points on the Democrats nominating process:

1) You neglected to mention that all of the races in the Democratic primaries and caucuses allocate their delegates proportionally. That is a far more egalitarian way of running things, and has allowed Sanders to remain in the conversation – the Republicans, by contrast, have winner take all votes starting March 15th. (See

2) Do you not think that perhaps in a piece that discusses the superdelegates of the Democratic Party, you should mention who these superdelegates are? Here’s the full list, with a couple of corrections noted at the top:,_2016#cite_note-2.

The breakdown, from the Wikipedia piece:

++ 20 distinguished party leaders (DPL), consisting of current and former presidents, vice-presidents, congressional leaders, and DNC chairs
++ 21 Democratic governors (including territorial governors and the Mayor of the District of Columbia)
++ 46 Democratic members of the United States Senate (including Washington, DC shadow senators)
++ 193 Democratic members of the United States House of Representatives (including non-voting delegates)
++ 437 elected members of the Democratic National Committee (including the chairs and vice-chairs of each state's Democratic Party)

Given the very small numbers who participate in the primaries, and the incredibly undemocratic nature of the caucuses (see Rick Hasen on this, for instance:, a group of superdelegates who represent elected Democratic office holders does not bother me overmuch – these are people who have been put into office by a far greater number of voters than those who participate in nominating contests. What the Democrats have in their nominating process is, essentially, one part direct democracy (5/6th of the delegates through the primaries and cacuses), and one part representative democracy (1/6th of the delegates). I would be more comfortable is there were not such a large number of superdelegates who were DNC officials, but in thinking about it, I am not overly concerned about a small group (the emphasis being on small) of people who have dedicated their lives to the development of a political party potentially determining the party’s standard-bearer in a general election, if the convention were almost equally divided.

A final quibble concerning your last paragraph – I would argue that the US is divided between an extreme-right party and a center-right party. Compared to European democracies, the Democrats are a lot closer to a mainstream right party than any of the left parties. But that is a different argument for another day. . . .

David Ricardo said...

The math is such that with 15% Super Delegates aligned with one candidate in a two person race, the candidate who has no Super Delegates must get about 59% of the popular vote to overcome the disadvantage of having no Super Delegates. The candidate who has the support of the Super Delegates can win the nomination with about 41% of the actual vote. This seems excessive, that a candidate who wins say just under 59% of the popular vote still does not get the nomination.

So the concept of the Super Delegate seems correct for reasons discussed in the Post and the Comments, but the 15% number just seems to be way too high.

Joe said...

That's something of a worse case scenario where super delegates all go to one candidate.

David Ricardo said...

Uh, according to Real Clear Politics Clinton has 467 Superdelegates to Sanders 26. Welcome to the Worst Case Scenario.