History teaches that, far from being a freakish anomaly, Bush v. Gore is just a particularly visible sign of a problematic phenomenon that is deep-rooted in our nation’s democracy.

The tenth anniversary of Bush v. Gore has not gone unnoticed. There have been personal reminiscences and reflections on its legacy, as well as renewed critiques (and defenses) of the Court’s decision. There has also been at least one academic symposium devoted to a decennial retrospective on the dispute over Florida’s presidential electors. I had the good fortune to participate in this symposium, sponsored by the St. Thomas Law Review, which featured some of the attorneys and public officials involved in the dispute, and I learned details that I had not known previously. (The symposium proceedings were recorded and, as I understand from the Review, will be made public for the benefit of scholarly research. Special kudos to Nate Persily for moderating, and helping the Review to plan the event.)

I want to observe the occasion by looking at Bush v. Gore from an historical perspective, which is also the approach I was asked to take for the St. Thomas symposium. Bush v. Gore is often considered, for better or worse, an anomaly or aberration. Linda Greenhouse has called it a “bad hair day.” Others have characterized it as truly or (if it is possible) especially unique—a situation never to be repeated—and therefore a case that would lack precedential value even if the Court had not explicitly cabined its Equal Protection holding.

I think history, however, sheds a different kind of light on Bush v. Gore. Far from being an anomaly, it is just a particularly recent and glaringly visible sign of a problematic phenomenon that is deeply rooted in our nation’s political culture. This problematic phenomenon concerns our democracy’s legal apparatus for major vote-counting disputes. It has two dimensions—one jurisprudential, the other institutional—and both combined to erupt in the judicial divisions that beset the Court in the case. The jurisprudential dimension is that, as a nation, we are torn between two different ways to view the constitutional right to vote, and we have been so torn from the beginning. One way is to emphasize the constitutional need to protect voters from wrongful disenfranchisement caused by the improper failure to count their particular ballots. The other way is to emphasize the constitutional need to protect the integrity of the voting-counting process from improper manipulation by government officials seeking to favor one of the candidates in the election. These two ways to view the constitutional right to vote can often be in tension with each other, yet both have equally longstanding status in our political pedigree (going all the way back to the Founding), and as a nation we have never attempted to prioritize between or otherwise reconcile the two. Consequently, when a vote-counting controversy emerges over the outcome of a presidential election, we are at sea on what “first principles” to invoke to think through the controversy. Partisans on either side easily can invoke whichever constitutional conception better suits their position in the particular controversy.

The institutional dimension to this problematic phenomenon is that as a nation we also do not know what institution of government we want to have final authority to resolve the jurisprudential debate over how the constitutional right to vote should resolve a particular vote-counting controversy. The Framers of the Constitution did not specify what institution has authority over a vote-counting dispute in a presidential election, and this particular omission is what led to the crisis of 1876 as well as Bush v. Gore. But the institutional uncertainty goes deeper than that. Precisely because the right to vote has a fundamental status in constitutional law (and has always been recognized as having such, even when at the outset the suffrage was significantly limited in scope), there has always been the potential that the judiciary considers as its prerogative the protection of this constitutional right as the judiciary itself thought fit. Moreover, once the Fourteenth Amendment federalized the protection of fundamental constitutional rights against wrongful deprivations by state governments, the institutional uncertainty compounded: now the federal, and not just the state, judiciary would see it as its business to protect the fundamental right to vote against state deprivations (although the federal judiciary’s conception of how best to protect the constitutional right to vote might differ from the state judiciary’s conception, with both conceptions plausible given the unresolved jurisprudential tensions in the context of vote-counting disputes).

I explore the origins of this problematic phenomenon, in both its dimensions, in a just-published piece in a symposium issue of the Indiana Law Review: The Founders’ Bush v. Gore: The 1792 Election Dispute and Its Continuing Relevance. That piece explains how as a nation we got off on the wrong foot, so to speak, and why Bush v. Gore is just a recent and especially glaring misstep. The SSRN abstract of that piece gives a further summary.

But today, the tenth anniversary of the Supreme Court’s 5-4 stay of the statewide recount ordered by the 4-3 decision of the Florida Supreme Court, I want to focus on the particular institutional uncertainty of when the federal judiciary should exercise its supervisory powers over a state judicial system that arguably has contravened the constitutional right to vote in the context of a vote-counting dispute. The issue has no easy answer, which is why Bush v. Gore was an accident (or, for those who view it positively, an appropriate intervention) waiting to happen.

Bush v. Gore is a child, or grandchild, of Reynolds v. Sims, 377 U.S. 533 (1964). That truth is not merely the simplistic point that the majority in Bush v. Gore cited Reynolds as authority for its Equal Protection holding. Rather, Reynolds reoriented the federal judiciary’s relationship to state governments, including state courts, in terms of protecting the constitutional right to vote. In the post-Reynolds era, it is appropriate for the federal judiciary to insist that its conception of how best to protect the constitutional right to vote prevail over a contrary conception adopted by a state’s judiciary. We see this federal judicial attitude in the post-Reynolds decades bubbling beneath the surface in a myriad of lower-salience elections before it explodes in a presidential election in 2000. The most well-known of these harbingers of Bush v. Gore is Roe v. Alabama, which involved the election of Alabama’s Chief Justice.

Prior to Reynolds, the intellectual culture in which federal judges resided was hostile to the kind of comfortable intervention in state election vote-counting procedures that caused Bush v. Gore. This point is best seen by considering Lyndon Johnson’s victory in his 1948 U.S. Senate primary election. As detailed in Robert Caro’s biographical volume, Means of Ascent (part two of The Years of Lyndon Johnson), Johnson won that election because of intentionally fabricated vote totals. His opponent Coke Stevenson attempted to secure federal court intervention, to protect the constitutional right to vote under the Fourteenth Amendment, because in Stevenson’s judgment the state’s judiciary had been captured by pro-Johnson partisans and therefore was unable to serve as a fair and impartial tribunal. In this respect, Stevenson’s position was not dissimilar to Bush’s in 2000. Like Bush’s case, Stevenson’s case went all the way to the U.S. Supreme Court, and an emergency stay was issued, but in 1948 the purpose of the stay was the opposite of the purpose in Bush v. Gore: back then it was to divest the lower federal courts of jurisdiction on the ground that the federal judiciary could not invoke the Fourteenth Amendment to exercise supervisory authority over a state’s vote-counting procedures. As Caro recounts, Justice Hugo Black issued the stay after an unusual in-chambers oral argument, in which Abe Fortas served as Johnson’s lead lawyer. Had the Johnson-Stevenson disputed election occurred after Reynolds, the landscape would have looked entirely different: then Stevenson’s effort to secure federal court protection would have looked routine and compelling. Indeed, on its facts, the Stevenson-Johnson election presents an arguably even easier case for federal court intervention than Roe v. Alabama, 68 F.3d 404 (11th Cir. 1995), which is often taken as an example of when the federal court is right to find a Fourteenth Amendment violation in state vote-counting procedures.

I am not saying here which world is better: pre-Reynolds or post-Reynolds. My point, rather, is that they are so very different, and that we must understand Bush v. Gore as a product of the post-Reynolds world. Anyone who reads Caro’s book thinking that Stevenson was wronged, and wishing that the legal means would have been available to give him redress—means that would exist less than two decades later—must be prepared to accept that in 2000 the Supreme Court, in principle, had the proper authority to stop what it saw as a vote-counting process that had become hijacked by partisanship. Conversely, anyone who objects to the U.S. Supreme Court’s intervention in Bush v. Gore must be prepared to swallow the kind of outright electoral theft of a U.S. Senate election that was perpetrated in Texas in 1948, as portrayed meticulously by Caro, with no ability of the federal courts to intervene. If one is intellectually honest when considering these two celebrated cases, one cannot have it both ways.

As one ponders whether it is better to live in a pre-Reynolds or post-Reynolds world in this regard, consider this additional historical point. The first Justice John Marshall Harlan, often known as the “Great Dissenter” because of his clairvoyant dissent in Plessy v. Ferguson, also penned a dissent that anticipated the majority position in Bush v. Gore exactly 100 years later. The case was Taylor v. Beckham, 178 U.S. 548 (1900), and it involved a dispute over counting the ballots in a Kentucky gubernatorial election. The facts are ugly. One of the candidates was assassinated, and it appeared to many (including Justice Harlan) that the election was being stolen through fraudulent vote-counting. The victim of this apparent electoral theft, obtaining no relief from a reputedly partisan state judiciary, sought refuge in the U.S. Supreme Court. But this was the pre-Reynolds world, and the Court rejected the case on the same ground that Justice Black ruled in 1948: it is improper to invoke the Fourteenth Amendment as a basis for federal court supervision of a state vote-counting dispute. Justice Harlan, however, vigorously dissented. In language that would seem appropriate for a post-Reynolds world, he spoke of the necessity of federal court protection of the fundamental constitutional right to vote.

Here is just the finale of what Harlan wrote in his impassioned Taylor v. Beckham dissent:

“[T]he overturning of the public will, as expressed at the ballot box, . . . in order to accomplish partisan ends, is a crime against free government, and deserves the execration of all lovers of liberty. . . . I cannot believe the judiciary is helpless in the presence of such a crime. The person elected, as well as the people who elected him, have rights that the courts may protect. . . . I am of opinion that the writ of error should not have been dismissed, and that the [C]ourt should have adjudged that the decree below took from Taylor and Marshall rights protected by the 14^th Amendment of the Constitution of the United States.”

Was Harlan wrong? He certainly was not wrong in Plessy even if he was alone at the time. But if Harlan was right in Taylor v. Beckman, then it is hard to say that Court’s December 9^th stay in Bush v. Gore was institutionally inappropriate. If it is proper for federal courts to intervene when they think they see state courts tolerating, or even perpetrating, stolen elections, then the warrant exists for the majority in Bush v. Gore if they think that something of the kind is occurring. We can debate whether what the 4-3 decision of the Florida Supreme Court did amounted to a danger of electoral theft, but then we are back into the jurisprudential dimension of our longstanding problematic phenomenon: there is a genuine tension in a constitutional tradition about how best to handle the kind of vote-counting dispute that the Florida Supreme Court faced.

The truth is that we are unsettled regarding the institutional dimension as we are the jurisprudential dimension. As much as we inhabit the post-Reynolds world, we are ambivalent towards its full implications. We can read Harlan’s dissent in Taylor v. Beckham and be sympathetic, but we also fear overreaching by a federal judiciary. We don’t want to federalize every state vote-counting dispute, and in the ten years after Bush v. Gore we have not yet figured out where to draw the line. Indeed, we only somewhat recognize that any line drawn by the federal judiciary is one ultimately policed by the federal judiciary, and thus there is the inevitability of the federal judiciary having a general supervisory role in these cases even if we purport to limit the exercise of this federal power to a narrow subset of extreme instances.

Bush v. Gore came to us a century after Harlan’s dissent in Taylor v. Beckham and can be seen as the fulfillment of Harlan’s vision of a role for the U.S. Supreme Court to protect the constitutional right to vote, and to have one’s ballot counted fairly, from partisan abuse by state officials. Now ten years after Bush v. Gore, we can begin to see that case as on a historical continuum that starts with the Founding, extend through the Hayes-Tilden dispute of 1876 and on to Taylor v. Beckham at the end of the nineteenth century. A half-century later is the Stevenson-Johnson dispute, which is the culmination of the “no jurisdiction” view adopted by the Taylor v. Beckham majority before the pendulum begins to swing with the advent of Reynolds v. Sims.

Will the pendulum swing back again, because of antipathy to Bush v. Gore or otherwise? Or will we oscillate uncertainly in some kind of institutional “no man’s land,” not knowing what we think of the relationship between the state and federal judiciaries when it comes to vote-counting disputes? And will our institutional uncertainty continue to be compounded by jurisprudential uncertainty regarding the constitutional principles that should apply to these disputes? Unless as a nation we begin to grapple systematically with these two dimensions of uncertainty, I fear that Bush v. Gore, far from being an isolated aberration that we can easily forget, will instead be just our generation’s name for a perennial weakness in our system of democracy.

Here is the SSRN abstract for my article, The Founders’ Bush v. Gore: The 1792 Election Dispute and Its Continuing Relevance, which I am honored to have included in the recent symposium published by the Indiana Law Review:

ABSTRACT:

In 1792, the Founders of our Republic and its Constitution faced their own version of what our generation knows as Bush v. Gore. It occurred in the context of that year’s election for governor in New York. John Jay was challenging the incumbent, and Jay and his supporters believed that they were wrongly robbed of their rightful victory when the state’s canvassing committee – in divided vote that fell along party-line lines – disqualified one county’s ballots because of a technical defect in the delivery of these ballots from the county to the Secretary of State.

Although the 1792 dispute concerned the rules for ballot delivery, rather than hanging or dimpled chads, the same basic jurisprudential debate in Bush v. Gore existed in 1792. Just as Gore argued for avoiding the disenfranchisement of voters as a consequence of faulty electoral machinery, so too did Jay’s side claim that protecting voters from the loss of their fundamental (and constitutional) right to the franchise was justification for excusing a technical breach of the state’s ballot delivery law. Conversely, just as Bush’s legal team claimed that strict enforcement of Florida’s election rules was necessary to preserve the integrity of the electoral process, so too did Jay’s opponents make the identical argument.

Moreover, this debate about how best to operate democracy in the context of a dispute over the results of an important statewide election consumed the attention of the Founding Generation. Jay’s Federalist Papers co-authors, Hamilton and Madison weighed in with their views, as did Jefferson, Monroe, and Burr. Other important, but less well-known, Founders were also involved. For example, Edmund Randolph, the nation’s first Attorney General, forcefully expounded the integrity argument. James Kent, a Jay supporter who would later be known as “America’s Blackstone” for his Commentaries on American Law, condemned the canvassing committee for being structurally biased against Jay’s side in its membership.

This largely forgotten historical episode, with its similarities to the disputed presidential election of 2000, is more than just an interesting curiosity. Rather, the difficulties that the Founding Generation had with its own version of Bush v. Gore – especially its lack of an institution for resolving the dispute that was perceived to be structurally fair to both sides – helps to explain why our own generation had such difficulty with resolving the 2000 presidential election. We have inherited the constitutional system that the Founders bequeathed to us. But that system did not include an adequate mechanism for handling a disputed election for chief executive, whether governor or president. This Article explores the reasons for the Founders’ inability to equip us, or (as shown by the events of 1792) even themselves, with this necessary element of a well-functioning constitutional democracy. As a nation, we will continue to suffer from this inadequacy until we recognize that the Founders did not give us all that we need and thus we take it upon ourselves to add the missing institutional ingredient to the system they gave us.

The dispute over reconciliation in the state’s gubernatorial election is over, but the court points out an issue that could prove troublesome in future close elections.  UPDATE: The Republican candidate has now conceded.

In a unanimous, thorough, and well-reasoned opinion, the Minnesota Supreme Court rules that voter receipts, and not just poll book signatures, may be used for purposes of reconciliation.  This ruling has the effect of eliminating the basis on which Republican Tom Emmer had attempted to challenge reconciliation practices in this year’s election.  Unless Emmer has some other grounds for contesting the election that is not publicly apparent, it would seem that his Democratic opponent Mark Dayton’s victory is a foregone conclusion.  

Still, buried in footnote 4 of the court’s opinion is an unsettled legal issue of potential interest in future elections.  There the court noted that, even though the use of voter receipts is proper, the remains the question of what happens if the number of voter receipts does not match up with the number of ballots cast.  As I noted in a previous post, and as the court itself quoted, the relevant statute calls for the local election officials to randomly withdraw a number of ballots equal to the excess number of ballots cast.  But apparently this statutory requirement is not uniformly followed among election officials in Minnesota.  The court stated:  “In responses filed to the petition, certain local election officials appear to have conceded that they are not removing excess ballots, although they do not concede a statutory violation.”  The court went on to say that it did not need to address the propriety of this local practice, because Emmer’s complaint had focused specifically on the issue of receipts.

It seems beyond the realm of possibility that any failure to follow the statutory requirement of random withdrawal could affect the outcome of this year’s gubernatorial election in Minnesota.  But, as the state supreme court itself has now warned, this issue is there potentially lurking for future elections.  If Minnesota again had an election as close as its 2008 U.S. Senate election, where the margin was just 200 or 300 votes, I presume the state would not want to be in a position where it faced ferocious litigation over the apparent noncompliance with this statutory requirement by election officials in some localities but not others.   That would be a headache that might cause Minnesota to relive the kind of fight that it faced in 2008.  For this reason, it behooves election administrators in Minnesota to ask themselves whether they can take steps to coordinate their local practices with the state’s statutory requirements, so as to avoid this potential situation.

UPDATE (Dec. 8): The Republican candidate for governor, Tom Emmer, has now conceded.  This concession, coming well in advance of the January 3 inauguration date for the state’s new governor, shows that Minnesota was able to bring this disputed election to a successful conclusion.  There was perhaps more saber-rattling this year than the roughly 9000-vote margin of victory would seem to have warranted.  Speculation was that ardent Republicans wanted to use a dispute over the outcome to delay the inauguration of Democrat Mark Dayton, to keep Republican incumbent Tim Pawlenty in office as a holdover for as long as possible.  But in his concession speech, Emmer dismissed use of election litigation for purposes of delay as inapproriate.  Having dodged that possibility this year, Minnesota now can take a collective deep breath and more leisurely consider what risk it may face in the future of such delay tactics and other potential vulnerabilities, including the one identified earlier in this post about the random withdrawal procedure.  Having suffered through an eight-month disputed U.S. Senate election in 2008, many observers in Minnesota wanted to avoid a dispute this time that went into January.   The state accomplished this objective today, but it can still ask itself whether it has the right set of procedures and institutions in place to achieve the same objective if a future gubernatorial election has an apparent margin of victory in the hundreds rather than in the thousands.

It looks as is if, going forward, “reconciliation” will be the main issue in the still-unresolved gubernatorial election in Minnesota.  This post discusses Minnesota law on reconciliation and how it might play out in this year’s election.

In a news conference this morning, Republican gubernatorial candidate Tom Emmer made clear that he is not conceding the race to his opponent Mark Dayton (who leads by approximately 9000 ballots) unless and until Emmer is satisfied on the issue of reconciliation.

What exactly is reconciliation and why is Emmer presently unsatisfied?

Reconciliation is the process by which election officials match the number of voters who cast ballots with the number of ballots cast. In Minnesota, as in some other states, if when preparing to make their initial count of the ballots, local officials find that there are more ballots cast than the number of voters who cast them, these officials are supposed to perform a procedure known as “random withdrawal”: literally, an official is supposed to reach into the ballot box, without looking specifically, and pull out the number of excess ballots.  [In Minnesota, this process occurs on Election Night at the precincts.]

Minnesota’s statute is clear about this requirement: section 204C.20, subdivision 2 (entitled “excess ballots”) details several steps that must occur in the reconciliation process. First, the officials see if any excess ballots were not properly authenticated in advance. If so, these specific ballots are removed and not counted. (If there is no excess to begin with, then these ballots not properly authenticated may still be counted.) After this step is complete: “If there is still an excess of properly [authenticated] ballots, the election judges shall replace them in the box, and one election judge, without looking, shall withdraw from the box a number of ballots equal to the excess. The withdrawn ballots shall not be counted.”

In Minnesota this year, the fight over reconciliation has focused on so far the exact procedures used to conduct the reconciliation process.  The debate has been whether it is permissible to count the number of receipts that voters receive before casting their ballots, as specified by administrative regulation, or instead the process must be confined to what the statute terms as either “the number of signed voter’s certificates” or “the number of names entered in the election register.”  This post, however, focuses not on that issue, but instead on the potential issue of random withdrawal itself.  The inquiry here is whether any failure to perform random withdrawal, after it is settled what counting method is appropriate for reconciliation, could be the basis for litigation over the outcome of an election.  It is an issue that I have explored in previous scholarship.  See Edward B. Foley, The Analysis and Mitigation of Electoral Errors: Theory, Practice, Policy 350, 365-369 (2007).

It is likely that compliance or noncompliance with random withdrawal will fizzle out as a practical matter in connection with this year’s governor’s election in Minnesota.  That’s because, as discussed further at the end of this post, it is extraordinarily unlikely that the number of ballots that should have been withdrawn but were not would be anywhere near enough to make a difference in this race.  Still, absent seeing the actual final evidence on what those numbers are, and thinking about what might happen if Minnesota pulls a trifecta and has another statewide election in 2012 that is close enough to fight about (or if a similar issue should arise in another state), it is worth exploring what might happen if noncompliance with random withdrawal conceivably might have made a difference in a close election.

Allegedly, in this year’s election, some local officials in Minnesota did not perform the random withdrawal procedure, as required by statute. In the past, there has been an indication emerging from research into election administration practices that some officials philosophically don’t like this requirement; they view it as antiquated and contrary to the interests of the voters they serve. These officials believe that they can be confident in the legitimacy of the excess ballots cast even if they cannot be completely reconciled with corresponding evidence of the voter who cast the ballot.  One hears of poll workers saying something along the lines of, “Well, I know Jane Smith voted because she’s a neighbor and I saw her cast her ballot, so even if she did not sign the register and it appears that we are short one voter, we shouldn’t randomly draw one ballot from our precinct, because the extra ballot must be Jane’s.” [Indeed, one local election official was quoted in the Pioneer Press: “You have to remember that I’m in outstate Minnesota where we all know each other. . . .  The wife signs and the husband doesn’t and off they go.”]  One can be sympathetic with the position of the poll worker in this situation, but failure to perform random withdrawal when there remains a discrepancy between the number of ballots and the statutorily acceptable evidence of number of voters does not strictly comply with the requirements of Minnesota’s statutory law.

Minnesotans, of course, have heard this kind of thing before—recently, in the context of their disputed 2008 U.S. Senate election. Back then, the issue was the rules for submitting absentee ballots. It turned out that some election officials liked to be lenient in the enforcement of those rules as well, for the benefit of the voter. Some voters would fail to fill out their absentee ballot envelopes properly, but the officials would still count them if they were confident that the voters and their ballots were valid anyway. In the litigation over the outcome of the Senate race, the courts took a dim view of this administrative leniency: the statutory rule must be strictly followed whether the local officials like it or not. Something similar may happen with respect to the random withdrawal issue this year.

The leading case in Minnesota on the obligation to follow the random withdrawal statute properly is Johnson v. Trnka, 277 Minn. 468, 154 N.W.2d 185 (1967). There, the trial court mistakenly performed the random withdrawal procedure without first removing ballots that were not properly authenticated. Because there were more of these ballots (6) than the excess over the number of voters (2), removing these ballots eliminated the excess and there was no longer a warrant for random withdrawal. The Minnesota Supreme Court, accordingly, reversed.

Earlier in the current fight over the governor’s election, Emmer’s side asked the Minnesota Supreme Court to mandate compliance with the random withdrawal procedure before beginning the mandatory statewide administrative recount of all initially counted ballots. The high court declined to do this, but has not yet stated its reasons. In his press conference, Emmer said he is waiting for the court’s opinion before deciding to make his next move on the reconciliation issue.

The Johnson precedent, because its particular facts involved a different problem, does not specifically say what happens if (after discounting ballots that were not properly authenticated), there still remains an excess number of ballots relative to the number of voters, but the local officials did not perform random withdrawal. In a post-recount judicial contest of a certified gubernatorial election, may the losing candidate demand that the court order the statutorily required random withdrawal? The tenor of the Johnson opinion, indicating that the statute should be followed properly, suggests maybe so. But it is hardly a foreordained conclusion.

For one thing, the judicial precedent in other states is that random withdrawal will not be ordered subsequently in a contest of an election if local election officials can offer good reasons for not conducting the procedure. For example, in Minnesota’s neighbor Wisconsin, that state’s supreme court has held an excess ballot, even if not properly authenticated, should not be removed from the count unless there is separate evidence that it was cast by an unqualified voter. Ollmann v. Kowalewski, 238 Wis. 574, 300 N.W. 183 (1941). That court stated: “It is more likely that the election clerks made a mistake in checking someone who voted . . . . The ballot expressed the will of the person who voted it. To reject it without proof of illegality or fault on his part would disfranchise the voter.” This Wisconsin approach is the opposite of the strict compliance standard embraced by the Minnesotsa Supreme Court in the Coleman v. Franken litigation. But Minnesota’s high court there made a point of saying that the strict compliance standard for absentee voting rules did not necessarily apply to in-precinct voting rules, for which Minnesota apparently adopts a more lenient posture. Thus, the choice between strict and lenient enforcement of the specific random withdrawal rule potentially at issue in this year’s gubernatorial election still seems an open question under the state’s judicial precedents.

Finally, there remains the all-important question of numbers. Assuming that Dayton’s certified victory at the end of the recount is 9000 (to use a round number), presumably Emmer would need to show at least 9001 excess ballots statewide in order to even begin to engage the court with the question of whether to order any random withdrawal that should have occurred but didn’t. Even assuming that Emmer could show a much larger lack of reconciliation—say, 90,000 excess ballots statewide, to jump an order of magnitude above Dayton’s margin of victory—it would seems extraordinarily improbable that performing random withdrawal would overturn that victory. I’m no statistician but my understanding is that those who are would calculate the odds as infinitesimal that randomly removing 90,000 ballots from a certified count of 2.1 million ballots would erase a 9000-vote margin. In this circumstance, then, it is quite possible that the Minnesota judiciary would not put the state’s election officials to the time and expense of performing the random withdrawal procedure where there is no reasonable possibility of changing the outcome of the election.

Still, assuming that random withdrawal did not occur this year in Minnesota as required–an allegation that (as far as I can tell) has not yet been definitely established–one wonders whether it would have been possible to avoid any of these legal issues on what if anything to do to remedy such a shortcoming.  Perhaps if the election officials had just performed random withdrawal as required by the statute, rather than attempting to improve on the administration of the electoral process according to their own lights (assuming for the moment that any such behavior occurred in Minnesota this year), any potential legal issue regarding compliance with the random withdrawal rule could have been avoided. If the statute on random withdrawal needs to be reformed, that is a task for the legislature to take up. Just as with updating antiquated rules for absentee voting, problems emerge if local officials deviate from the existing rules on their own, rather than getting the rules fixed for everyone statewide.

[Updated December 5, 2010]

As we watch events unfold this year concerning the unsettled control of the state senate in New York, and the several elections on which that control depends, let’s hope that the current resolution contains nothing like the ugliness that afflicted a similar episode in 1891.

Thanks to the sleuthing of Justin Levitt, I now know that control of the state senate in New York depends on the outcome of three undecided races. The New York Times has the details.

Of course, it is still extremely early in the process, and no one knows how it will turn out, but the basic facts bear some resemblance to one of the lowest moments in New York’s electoral history. During an episode known as “The Stolen Senate” (as described in Chapter XXX of this book), partisanship ran amok in 1891 when control of the state senate similarly depended on the outcome of several close and disputed elections.

The historical details are too convoluted and technical to describe in this “blog” post. My co-author Steve Huefner and I will cover it more in depth in the book on the history of disputed elections in the U.S., which we are currently writing. But the following brief synopsis, necessarily oversimplified for present purposes, will give you a flavor of what happened.

Back then, although four state senate seats were disputed, the fight ultimately boiled down to just two of them. If the Republicans prevailed in either of these two elections, they would control the state senate. The partisan split in the chamber was 15-14 in the Republican’s favor with only these two seats unresolved. Conversely, if the Democrats prevailed in both, they would have control.

One issue, involving Onondaga County (where Syracuse is located), concerned an error made by local election officials: they sent the official ballots designated for one precinct to another precinct by mistake. This election was one of the first in New York involving the so-called “Australian” ballots, in which the officials were responsible for printing and distributing them. Evidently, being unfamiliar with this responsibility, the election officials had trouble in its implementation. As a practical matter, however, the error did not prevent voters for making a choice in their local senate election: even though these voters received the ballot for the wrong precinct, the name of their senate candidates was on the ballot, and they were thus able to use the incorrect ballot to make their choice.

Based on the returns, if these ballots were counted, the Republican candidate for this seat would win. Not surprisingly, the Democrats challenged these ballots on the ground that they were improper under the state’s new election law adopting the Australian ballot. In a 4-3 vote of the state’s highest court, the Democrats prevailed on this argument. The judges in the majority happened to be all Democrats, and they adopted a “strict compliance” standard for the new Australian ballot statute. The dissenters, two Republicans and one Democrat, argued vociferously for an interpretation of the statute that would prevent the disenfranchisement of innocent voters as a result of official error. “Assuming that [the statute] is susceptible to two constructions—one preserving the right of the voter, and the other forfeiting it—can it be doubtful which construction ought to prevail?” The dissenting position was a version of what Rick Hasen calls “the Democracy Canon” of statutory construction for disputed election cases, but that position did not carry the day, arguably due the partisanship of the majority in that case (although of course the judges in the majority articulated what they claimed were important policy justifications for strict enforcement of the new anti-fraud protections of the Australian ballot).

This 4-3 judicial victory in the Onondaga County seat brought the Democrats to a 15-15 tie in the state senate.

The resolution of the other outcome-determinative disputed election, in Dutchess County (where Poughkeepsie and other Hudson River Valley communities are located), was even more troubling. This dispute also concerned the new Australian ballots, but a different type of problem. Here, the problem was essentially a smudge that the printing process had left on the ballots. The Democrats claimed that these smudges were identifying marks that invalidated the ballots. If these ballots were invalidated, the Democratic candidate again would prevail. But this argument was too much for any of the Democrats on the state’s highest court, which unanimously ordered the certification of returns showing that the Republican had prevailed. (Because of the complicated judicial procedures involved in the case, and the posture in which it arrived at the court, the order was technically for the certification to occur once it arrived from the county to the Secretary of State.) The state canvassing board, which was controlled by the Democrats, however, refused to obey this judicial decree. Instead, the board certified a return that excluded the disputed ballots, and on the basis of this certification, the Democrats took control of the state senate. (They accomplished this obstruction of justice by falsely claiming that the judicially-approved return never arrived from the county to the Secretary of State’s office.) Later, the state’s highest court held members of the state canvassing board in contempt of court for this conduct, and it was quite a scandal in the state. New York adopted a new constitution in 1894 that attempted, not entirely successfully, to reduce the role of partisanship in the counting of ballots. In any event, these subsequent proceedings did not undo the successful partisan theft of the state senate that had occurred based on the combined outcome of these two disputed seats.

As we watch events unfold this year concerning the unsettled control of the state senate in New York, and the several elections on which that control depends, let’s hope that the current resolution contains nothing like the ugliness that afflicted the episode of 1891.

As I look over the undecided statewide races, this is what emerges for me this morning:

Connecticut is the only state that I would label as “yellow alert” so far, using my alert system, given that litigation has already occurred to extend the polling hours in Bridgeport, and more litigation is anticipated.

Otherwise, the processes are proceeding as they should—my “green” category.

Colorado potentially deserves a “blue” because of the Denver Post’s report of the large number of uncounted provisional ballots that could affect the outcome. I can’t yet tell whether any of the other states (besides Alaska) similarly deserve “blue” rather than “green” status. The key difference in my mind between “green” and “blue” is that provisional ballots by definition have an uncertain eligibility status and need eligibility review before being counted. The mere fact that some ballots, like absentees, still need to be counted is not enough to shift a state from green to blue. UPDATE: The Denver Post has called the Senate race for Bennet.

For a similar reason, Alaska deserves a “blue” since the write-in ballots need to be removed. I don’t think Alaska deserves a yellow based on the previous litigation, since it was resolved without apparently affecting the result. If there is new litigation, that would convert Alaska back to yellow.

Another interesting part of the narrative: the states with the three big recounts of the 2000s — Florida, Washington, Minnesota — all have undecided statewide races at the moment (although that might change quickly with Florida–UPDATE: Florida is, indeed, settled with a concession from the Democrat). These states can show that they specifically have improved their processes, after having been put under the microscope recently.

Finally, with respect to all the undecided races, these states have an opportunity to be “laboratories of democracy,” to invoke the famous Brandeis phrase. It will be interesting to watch how the New England states differ in handling their close races. The same holds with a side-by-side comparison of Oregon and Washington. The Midwest can compare Minnesota and Illinois–states with very different rules and political cultures.

On Election Day, as we contemplate the potential process of resolving races “too close to call” tonight, let’s not jump to any premature conclusion that there were voting issues. Just because a race is too close to call is not an indication there are any problems. The increased use of absentee and provisional ballots in the past decade means we have to sometimes wait for these ballots to be counted before a winner can be declared. And, a recount is a process, not a problem. After a recount is conducted, if there is litigation, that is when we begin to become concerned.

Folks have been quoting the Election Administrators’ Prayer lately: “Lord, Let This Election Not Be Close.” You can understand why: there are widespread predictions of significant statewide races—U.S. Senate or gubernatorial—that could be close enough to require recounts.

While we wait for the early returns that will either bear out or refute these predictions, it is worth keeping the following points in mind:

First, a recount is a process, not a problem. The mere fact that an election requires a recount is, by itself, not a bad sign at all. All it means is that the particular election was extremely competitive, a sign of a vibrant democracy. This point especially applies to statewide races, which necessarily involve the state’s entire electorate and are not complicated by the potential problem of gerrymandering. (In other words, one close congressional race in a state with ten or twenty congressional districts is not necessarily a good sign at all.)

A recount means that the two top vote-getting candidates were almost equally successful in appealing to the average voter who, going into the campaign, was open to persuasion from both sides. Democracy at work. Recounting the ballots is just an indication that the legal system considers elections important enough to take the time and effort to double-check the results, in an effort to make sure that they are correct.

Second, before there can be a recount, it is necessary to complete the initial count. The public is starting to get used to the fact that we don’t always have a complete initial count of all ballots cast in an election on Tuesday night, after the polls close. Two developments over the last decade have accentuated this tendency. One, the existence of provisional ballots, mandated by the Help America Vote Act of 2002, which give voters an opportunity to cast a ballot even if there is question about their eligibility. Because it takes at least a few days to verify whether or not these provisional ballots are indeed eligible for counting, it is impossible to declare a winner in an election where there are enough provisional ballots to conceivably make a difference. Two, as more states increasingly rely on absentee voting—and as they permit absentee ballots to arrive after the polls close as long as they are postmarked by election day—these last-arriving absentee ballots can also make a decisive difference in the outcome.

But neither of these developments are a sign of a malfunction in the election system. On the contrary, there are expected and appropriate byproducts of the policy choices to require provisional ballots and to permit election day postmarking of absentee ballots. If and when the public hears that a statewide race is “too close to call” because of yet-to-be-counted provisional and absentee ballots, the public should not immediately jump to the conclusion that something must be wrong and that there is a danger of the electoral will being nullified in the ballot-counting process. Instead, by itself, the announcement that “calling the race” requires waiting for the counting of these ballots is a sign of the system working properly—and that democracy is capable of identifying a winner when the electorate is closely split between the two leading candidates.

Of course, waiting to start a recount until all the provisional and absentee ballots have been counted initially does delay the process of identifying the winner for at least a couple of weeks. A public tired of a long season of campaign ads, as well as perhaps several weeks of early voting, may be anxious for a result in a major statewide race. Still, even more than answer, the public wants the count to be correct, and thus the public will be patient for the results of a recount as long as it remembers that the recount itself is not a sign of the system going awry, but an integral part of the system itself in order to properly handle the exceptionally close races.

Third, the commotion of the candidates and their supporters is not a sign of system weakness. We can be sure that, where there is a recount, there will be a lot of noise coming from both camps in an effort to sway public opinion towards one side or the other. This noise will include allegations of improprieties and irregularities, as well as pious comments about “counting all the votes” or “the need to bring closure to the election” (you can easily guess which side will say which, depending on who is ahead after the initial count). But the public would do well to filter out this noise and focus instead on the statement of the officials responsible for conducting the recount and bringing the election to a conclusion.

Consider the analogy of the noise among the crowd in a stadium at an important sporting event. If the game has gone into overtime, and a championship is at stake, of course the fans on both sides will be shouting. They are excited and rooting for their team. In the heat of the moment, they may even shout invectives at the referees. But observers who are not particularly partial to one team or the other—or even fans who are when they get a chance to calm down—can recognize if the refs are basically officiating the game fairly. And if they keep their eyes on the refs as they watch the end of the game, they know that all the noise in the stands from the fans on both sides won’t make a difference to the propriety of the outcome. The game will be fair, whatever the noise, because of what happens on the field, not in the stands.

There are different ways to express this key point. ALLEGATIONS of improprieties do not by themselves prove the EXISTENCE of improprieties. Or, as I like to say: in the context of elections at least, where there is smoke does NOT necessarily mean there is fire.

The candidates and their most ardent supporters have the incentive to “make allegations first, attempt to substantiate them later.” The public as a whole, however, should not be bamboozled by these charges. Wait and see what the refs have to say about them—and see if the refs conduct themselves properly, in a way that makes the competition a fair one.

This brings us to our most important point.

Fourth, if there is litigation over the outcome of an election, watch for whether the judges agree or disagree about what the result should be under the law.

The existence of lawsuits over the counting of ballots after the polls close is analytically distinct from the existence of a recount, and it is very much worth keeping this distinction in mind as events unfold. Unlike recounts, lawsuits are something of a warning sign. It does mean that one side or the other thinks there have been problems that are potentially outcome-determinative—not merely a close vote that deserves recounting.

Remember, of course, that allegations of problems are just allegations—unless and until a court accepts them as proven facts based on adequate evidence. The filing of a lawsuit is just a formal form of alleging improprieties. The mere filing does not prove them true. Moreover, candidates and their attorneys are quicker to file lawsuits in recent years than in the past, and thus a kind of discount factor can reasonably be applied to the mere filing of a lawsuit in the aftermath of balloting.

If there is this kind of lawsuit, it all comes down to how the judiciary handles it. If several judges look at the lawsuit, at both the trial and appellate levels, and especially if the judges who do so come from different political backgrounds, and they all reach the same conclusion that the lawsuit lacks merit, then the public essentially should treat the result as if the lawsuit never happened. In other words, it was just a very close election; one side tried to prevail by going to court, but they couldn’t make it happen. The system worked. The outcome can be trusted. The situation is essentially equivalent to a recount without a lawsuit at all.

If conversely the judges all agree that the lawsuit HAS merit, then the situation is somewhat different, but not critically alarming. The fact that the lawsuit was correct in alleging problems in the vote-counting process means, regrettably, that there WERE problems that required the court’s attention. It would have been better, obviously, if there hadn’t been the problems in the first place. But if the judges, regardless of their political backgrounds, all agree what the problems were and what should be done to remedy then, then the public can be confident that the election system has well-working judicial “safety-value” to protect the system from problems that can (and, again regrettably, do) arise. The public can trust the outcome of the election in this situation—indeed, more so than if the judiciary had not acted to remedy the problem.

It is only when the judges split in their rulings that we have cause for anxiety. At the very least, a split among judges means that the law or evidence was unclear and thus it was possible for judges to go either way. In this situation, we can’t completely trust the results of the election, even if the judges acted in good faith to the best of their judicial abilities, because how do we know which judicial position was the correct one? If one good-faith judicial position says Candidate A should win the election given the law and evidence, but the other good-faith judicial position says Candidate B should win given the same law and evidence, then by definition the election is being decided by the judges and not just the voters. Moreover, there is inevitably some arbitrariness to this split judicial decision, no matter how well-reasoned both good-faith judicial positions are articulated. How come there were more judges who adopted the one point of view rather than the other? The other candidate would have won this important election if it happened that more judges adopted the alternative good-faith position. That luck-of-the-draw is inevitably unsettling to an electoral system and in some basic sense inconstant with the democratic idea that the voters themselves should determine the outcome of the election.

Of course, the situation is even worse—much worse—if the split among the judges tracks their differing political backgrounds. If Democratic judges rule for Democratic candidates and Republican judges rule for Republican candidates, as disturbingly happens from time to time, then there is no trusting the result. In this situation, the litigation neither validates nor corrects the result of the recount (or initial count), but simply indicates that partisan judicial fiat determines which side prevails. If a system of election integrity monitoring had a set of “alert levels” comparable to those for terrorist warnings, this situation of a partisan split among the judges responsible for deciding litigation over the counting of ballots would qualify for the highest Red Alert level.

But here on Election Day, as we contemplate the potential process of resolving races “too close to call” tonight, let’s not jump to any premature conclusion that we are going to have anything like this kind of Red Alert situation. Let’s hope instead that the process unfolds just as it is supposed to: with the counting of provisional and absentee ballots and the conduct of recounts, if necessary. In the absence of any litigation, there wouldn’t be the need for an heightened alert level. But even if litigation occurs, adjust the alert level only to yellow until we see how the judges handle it. If they split, but not along partisan lines, that warrants an orange-level alert. Let’s save Red Alert for a partisan judicial split. If we do end up seeing it, then let’s call it out: Red Alerts do deserve dramatic attention. But let’s be sophisticated and nuanced enough to be able to distinguish those situations that deserve Red Alert from those that do not.