The risk moreover is increasing: as if Ohio, already prime target, is expanding the size of the bull’s-eye.

Today, December 17, is the date the presidential electors of each state meet to cast their official votes for president. No drama surrounds the event this year, because there have been no vote-counting disputes between November 6 and now that could affect the outcome of the presidential election. To invoke a phrase that has become familiar, the margin of victory was comfortably beyond the margin of litigation.

But, as 2000 showed, it may not always be so, and some numbers from this year’s election indicate that, of all the presidential swing states, Ohio is the most vulnerable to a ballot-counting dispute that would delay determining the Electoral College winner. As will be detailed below, the swing states fall into three categories in terms of their level of risk–low, medium, or high–with respect to a disputed presidential election.  Ohio stands alone in having the highest risk.

Back on November 6, while waiting for the initial vote tallies from the battleground states, I tried to calculate what the margin of litigation might be. Or, to state the point somewhat differently, I tried to determine the deficit by which a presidential candidate might be behind in an outcome-determinative swing state, based on the initial count of ballots on Election Night, and yet still fight on in the hope of being able to overcome that deficit during the canvass of returns, leading to that candidate’s being pronounced the winner in the official certification of the result.

Focusing on Ohio, which Nate Silver (among others) had predicted would be most likely to be the “tipping state” this year (the one which would determine the winner of the Electoral College)—and which, in fact, had been the decisive state in 2004—I reviewed the relevant numbers from 2004 and 2008. Between Election Night and official certifications, John Kerry had managed to reduce George Bush’s lead in Ohio by only some 17,000 votes, from about 135,000 to a little under 120,000. No wonder, then, that Kerry conceded the morning after Election Night, after his campaign did the math. He could not come close to overcoming his Election Night deficit and therefore it certainly was not worth the fight. Had his Election Night deficit been 20,000 votes, however, it would have been worth fighting on during the canvass, as it turned out that he gained almost that amount without even fighting for it. And since on Election Night (or the morning after), he could not know exactly how much he might make up if he pursued the task vigorously, he reasonably might have perceived the 2004 presidential election to be within the proverbial “margin of litigation” if he had been behind by 30,000 or even 40,000 votes.

On election night in 2008, Obama was way ahead in Ohio, by about 205,000 votes, on his way to his Electoral College landslide. Interestingly, however, he increased his lead in Ohio between Election Night and official certification by about 55,000 votes, much more than the 17,000-vote gain that Kerry made during the equivalent canvass in 2004. Thus, hypothetically, if the 2008 presidential election had all come down to Ohio in the same way that it did in 2004, and if Obama had been down in Ohio on Election Night by 50,000 votes, he easily could have made up that rather large deficit—since he did even better than that during the actual 2008 canvass without lifting a finger. Indeed, if Obama had been behind in Ohio on Election Night in 2008 by 60,000 votes, or perhaps even more, and if Ohio had been the decisive state that year, the election still would have been within the margin of litigation, as Obama would have had a reasonable shot at overcoming a 60,000-vote deficit.

This year it turns out that Obama expanded his lead in Ohio between Election Night and official certification even more than he did in 2008, by some 66,000 votes, again without even trying. According to numbers printed in the Washington Post on November 8, Obama led Romney in the initial returns (with 100 percent of precincts reporting) by 100,142 votes. By the time Ohio’s Secretary of State posted the official certified result, Obama’s victory over Romney in the state had increased to 166,214.

Thus, the margin of litigation in Ohio has expanded again. If Ohio had been the decisive state this year (as it was in 2004), and if Obama had been behind on Election Night by 60,000 votes, he easily could have made up that difference—as he yielded even more votes than that during the actual canvass this year! Had he been behind by 70,000 or even 80,000, it still would have been worth a fight, since it would not have been unreasonable for him to think that he had a decent shot at making up even that large a deficit. Back on November 6, when contemplating the possibilities, I had characterized making up an 80,000-vote deficit as a “long shot.” But my assessment had been based on what happened in 2008, when Obama picked up “only” 55,000 votes between Election Night and official certification. Knowing now that Obama actually picked up 66,072 votes during the canvass this year without any fight at all, I still think the odds would have been against him if he had been down by 80,000, but the odds now seem somewhat shorter than I originally estimated. The margin of litigation in Ohio, in other words, is even larger than I believed on November 6.

What explains this expanding margin of litigation in Ohio? Funnily enough, it is not an increase in the number of provisional ballots. Ohio’s Secretary of State has not released a final official total of the number of provisional ballots cast in the state this year. But unofficial reports on Election Night put the number around 200,000, which would make it down slightly from the 206,859 provisional ballots in 2008. To be sure, Ohio had a very large number of provisional ballots in both 2008 and 2012, which likely made it possible for Obama to add so much to his victory each time. Still, there must be some other explanation for Obama’s gain in the canvass being 11,000 votes larger this year than in 2012 (with a similar number of provisional ballots cast). It is very much worth trying to figure out this explanation, since reversing this trend—and thus reducing the margin of litigation in Ohio—should be a high priority for anyone who wishes to limit the risk that a future presidential election becomes mired in an “extra innings” dispute, spanning from after Election Night through the canvass and leading up to the official certification required for the Electoral College meeting in that state. (It is possible that a change in the rate at which provisional ballots are counted, or some other demographic shift affecting the counting of provisional ballots—despite the relatively constant rate at which they were cast in both 2008 and 2012—might be a factor in explaining Obama’s increased gain in 2012 from initial returns to official certification. But we cannot do more than speculate about this until we see the full set of numbers concerning provisional ballots in Ohio this year.)

It turns out, moreover, that Ohio is something of an outlier among battleground states in having such a large margin of litigation. Based on the numbers in the same November 8 edition of the Washington Post, Election Law @ Moritz Research Fellow Jennifer Hart has calculated the difference between Obama’s initial lead and his certified victory in each of the battleground states. This difference is largest in Ohio—again, 66,072 votes— and only Michigan comes close, at 56,130. Here’s the full list, in descending order:

Obama’s gains from initial returns to official certification:

In addition to these ten states, Obama also gained 5,461 votes between initial returns and official certification in North Carolina, but there he was decreasing Romney’s lead rather than increasing his own. (All these numbers come from the last column of the spreadsheet prepared by Jennifer Hart, which is also available here.) What to make of the fact that in none of these states did Romney gain ground from the initial returns to final certification? That’s a point to consider after addressing some others.

First, these numbers reveal that there is a group of battleground states with fairly low margins of litigation. These states are Wisconsin, Iowa, Nevada, and New Hampshire, as well as North Carolina. A Democratic candidate for president would need to be behind the Republican in the initial returns by less than 10,000 votes in order to have a reasonable shot at making up the difference and pulling ahead during the canvass. These states, in other words, conform to the traditional expectation among recount lawyers that unless the gap between the two candidates after all the returns are in on Election Night is exceptionally narrow, it is very difficult for the trailing candidate to come from behind and, through lawsuits or other forms of disputation, overtake the leading candidate.

Second, however, there is another group of states with a somewhat expanded margin of litigation. These states are Florida, Colorado, Pennsylvania, and to an even greater extent Virginia. In these states, a Democratic candidate could be behind on Election Night by 30,000 votes—or, in the case of Virginia, even 40,000 votes—and still have a realistic chance of overcoming that deficit during the canvass. One could look at these numbers in two ways, either a “glass half empty” or “glass half full” perspective. Seeing the glass half empty, one might say that these battleground states face a significant risk of a disputed presidential election. If any of these states becomes decisive to determining the winner of the Electoral College in 2016, and if the Democratic candidate is behind the Republican by only 30,000 votes in the state on Election Night, then the state will confront an “extra innings” dispute during the canvass as the two presidential candidates conduct a ballot-by-ballot battle up to final certification of the winner.

Conversely, seeing the glass half full, one could point out that these swing states at least are not as vulnerable to a disputed presidential election as Ohio is (or Michigan, which is less likely to be the “tipping state” that determines the winner of the Electoral College). The Election Night lead for the Republican candidate would need to be below 50,000 in any of these four states (again, Florida, Colorado, Pennsylvania, and Virginia) for the Democratic candidate to have a reasonable chance of prevailing in the end. A statewide lead of less than 50,000 is extremely close—not as close as less than 10,000, but still very, very close. (This year, only in Florida and New Hampshire was the gap between Obama and Romney in the initial returns less than 50,000 votes, and in Florida only just barely: 47,016. In New Hampshire, it was 39,634.) Thus, these four states may face an elevated risk of a disputed presidential election, but from the “glass half full” perspective the margin of litigation in these states remains fairly narrow.

Not so in Ohio. Unlike the other perennial battleground states in presidential elections, Ohio has a uniquely large (and curiously growing) margin of litigation, and thus faces a uniquely higher risk of being the site of a disputed presidential election. Ohio is unique in this respect, not only because the difference between its initial returns and certified result is the largest of any battleground states, but as between Ohio and Michigan—the two battleground states with the largest margins of litigation—Ohio is much more likely than Michigan to be the single state that determines the Electoral College winner (as, indeed, Ohio was in 2004).

Ohio thus currently stands in a different posture than Florida, Colorado, Pennsylvania, and Virginia—the other presidential swing states with a somewhat elevated risk of hosting a dispute over the counting of their presidential ballots. Whereas in these states the Election Night gap between Republican and Democratic candidates would need to be less than 50,000 for the state to be within the margin of litigation, in Ohio the gap could be much larger, perhaps as high as 80,000 or even 100,000 votes. Moreover, it is important to understand the significance of the fact that the margin of litigation in Ohio is growing: whereas in 2004 Kerry could not have expected to overcome an Election Night deficit of 80,000 votes, in 2012 Obama reasonably could have thought that he would have had a decent chance to surmount even this large a deficit (had he needed to). It is as if Ohio, knowing that it is a prime target of a potential disputed presidential election, is expanding the size of the bull’s-eye.

Presumably, Ohio would be better served if it could figure out a way to shrink the size of this bull’s-eye and at least be in line with the other group of “mid-risk” swing states (Florida, Colorado, Pennsylvania, and Virginia), even if it cannot narrow the scope of its margin of litigation all the way down to the “lowest-risk” group (Wisconsin, Iowa, Nevada, and New Hampshire, as well as North Carolina). One way to think about this particular point is to say that Ohio at least should try to return to the level of risk it faced in 2004, when its margin of litigation was more comparable to the current mid-risk group of states. Indeed, Kerry’s 17,000-vote gain from initial returns to final certification was lower than Obama’s gain in any of these mid-risk states (the smallest of these being in Pennsylvania, at 21,975). Ohio made sweeping changes to its voting laws after the 2004 election, and in light the huge gains that Obama made in both 2008 and 2012 between initial returns and official certification, Ohio may wish to reexamine some of those changes.

Decreasing the number of provisional ballots is a step Ohio could take to reduce its margin of litigation. But Ohio likely will wish to consider other possible measures as well. Ohio already had a fairly large number of provisional ballots in 2004 (roughly 150,000), when its margin of litigation was much lower than in 2008 and 2012. And, again, the volume of provisional ballots in Ohio does not explain the significantly increasing margin of litigation from 2008 to 2012. Further study is needed to shed more light on this issue.

There is one more point worth making about these numbers. In all of these states, as already indicated, Obama was the candidate to gain ground between the initial returns and official certification. In other words, in not one of these battleground states was the movement from Election Night to official certification beneficial to Romney. This uniform asymmetry is potentially important. For one thing, it means that the margin of litigation on Election Night is different for Democratic and Republican presidential candidates. If a Republican presidential candidate is behind in Ohio on Election Night by 50,000 when all the initial returns are in, he cannot reasonably expect to be able to make up this gap by fighting on (as Karl Rove eventually realized, albeit in a rather confused public display). By contrast, given Ohio’s numbers in both 2008 and 2012, a Democratic candidate could expect to easily make up a 50,000 deficit on Election Night. Putting the point differently, a Republican candidate should hope to be up in Ohio by more than 50,000 votes on Election Night in order to still be ahead at the finish line of official certification.

This asymmetry may have reverberations. The conventional wisdom among recount lawyers is that the rules that govern the resolution of electoral disputes is one area in which Democrats and Republicans cannot predict ahead of time which rules they would prefer. Rather, the candidate who is ahead will always prefer rules that make it more likely to stay ahead, and the candidate who is behind will always prefer rules that make it more likely to pull ahead—and Democratic and Republican candidates will have whatever preferences make sense depending on whether in any given election they happened to be ahead or behind. (The dispute over Minnesota’s 2008 U.S. Senate election, between eventual winner Al Franken and the previous incumbent Norm Coleman, is a good example of this observation.) This agnosticism of which rules to prefer ex ante, before knowing whether one is ahead or behind, is what has led some to believe that the resolution of vote-counting disputes might be one area of election law where it is relatively easier to develop a bipartisan consensus on appropriate rules ex ante. But perhaps this assessment is potentially challenged by the asymmetry in this year’s numbers concerning the uniformly Democratic gains in swing states between initial returns and official certification. If the canvassing process always helps Democratic presidential candidates in swing states, would that dynamic cause there to develop strategically partisan preferences about what rules should govern the canvassing process?

It is premature to attempt to address this conjecture. As indicated earlier, we do not know yet why the process from Election Night to official certification has this apparently asymmetrical effect, either in Ohio or elsewhere—and we don’t yet know why it is growing in Ohio. Thus, more study needs to be done to figure out what is going on, and how potentially to address it.

In the meantime, however, one thing does seem to be somewhat clear: Ohio, being the outlier in these numbers, ought to have high on its own agenda an examination of potential ways it can bring itself more in line with other presidential swing states.

The media is reporting this morning that France’s conservative opposition party has become mired in a vote-counting dispute over the selection of the party’s new leader.

What I’m able to gather is that the “electorate” is rather small: about 300,000 party members entitled to vote (of whom only about 60% cast ballots), with the current count separating the two candidates by about 200 votes or so. Both sides apparently claim “irregularities” with some news stories referring to allegations of “fraud”.

Most interestingly (from my perspective), there is apparently a “commission” with jurisdiction to supervise the vote-counting and resolve the dispute. Some news stories describe the commission as “independent,” but this Wall Street Journal report describes it as a “party commission”: I’m curious to learn more about the status and composition of this commission—and whether this dispute will serve as a test of the ability to a commission to handle a dispute of this nature.

For several years now, I have thought that France might be the most interesting and relevant comparison to the United States in terms of institutions and procedures to handle vote-counting disputes, because it has more of a presidential system than the English-speaking parliamentary systems with which the U.S. is often compared—Canada, Australia, Britain, etc.

Let us hope that there is bipartisan support for that sentiment.

Correction: the President was originally quoted as saying “we need to fix that”; the quote has been changed to be exactly accurate.

At the beginning of his reelection victory speech, President Obama referred to the extraordinarily — and unacceptably — long lines that many voters in this year’s election faced.  Then, he added: “By the way, we have to fix that.”

Let us hope that there is bipartisan support for that sentiment.

There will be ample time for reflection and examination on what went wrong this year regarding the voting process, and what aspects of it work properly.  But for now it is enough to say that there is an overwhelming, and widespread, impression that “surely we can do better” — to echo Alex Keyssar’s eloquent statement today on the New York Times website.  As he said, it is a serious problem that “[a]ll of us now expect elections to be not only competition between candidates and parties but also conflicts over voting rights and election procedures.”

Rick Hasen has expressed the hope that the nation won’t ignore the significant structural weaknesses in our electoral system, now that the presidential election is “over” (except, of course, for all the official processes that remain, from the canvassing of returns to the meeting of the Electoral College, and the congressional reception of those Electoral Votes).  I share Rick’s hope, and look forward to doing the hard work necessary — through the American Law Institute’s election law project and otherwise — to help develop principles and proposals for what an improved electoral system would entail.

Meanwhile, there remain races other than the presidency that are still “too close to call,” and it will be important over the next days and potentially weeks to watch the processes by which they are officially determined.

It’s about the lead between the candidates and the volume of provisional ballots.

Here is how I’m thinking about the two Ohio numbers I will be looking for tonight, especially if Ohio ends up being the key to winning the presidential election.

The first number is the lead that one presidential candidate has over the other based on the initial returns of ballots counted tonight.

The second number is the volume of provisional ballots (as well as other uncounted, but potentially countable, ballots) that will be evaluated for eligibility over the next 10 days.

The key issue is the possibility that the trailing candidate, based on the first number, can overcome that lead by netting enough extra votes derived from the second number.

In 2004, Kerry trailed Bush by about 135,000 votes on Election Night, and there were about 155,000 provisional ballots. When it came time for the final certified count, Kerry narrowed Bush’s lead by only about 17,000, dropping the margin to a little under 120,000. We can think of this as Kerry achieving about an 11% “yield” from the provisional ballots in terms of his ability to narrow Bush’s lead.

By contrast, in 2008, Obama led McCain on Election Night by about 205,000 votes. Coincidentally, there were also about 205,000 provisional ballots that night. One big difference from 2004 was how much Obama was able to increase his lead by the time of the certified count. Obama expanded his lead by about 55,000 votes, to about 260,000. Given these numbers, Obama’s “yield” from the provisional ballots was about 27%, or more than double what Kerry’s had been.

I’m not quite sure what explains the difference in the “yield” from provisional ballots between the two Democratic candidates. I don’t think it is the rate at which provisional ballots were counted, because the rate was fairly similar those two years. I don’t see how it could be just the amount of provisional ballots cast, since why wouldn’t the “yield” rate be constant, whatever the volume of provisional ballots (either 150,000 or 200,000)? Perhaps it is some exogenous factor, concerning absentee ballots or something else.

But whatever the explanation, we need to contemplate the possibility of a different “yield” rate this year. Provisional voting is new enough, in the aftermath of the Help America Vote Act of 2002, that this year we might end up seeing a third “yield” rater (higher or lower than the other two, or somewhere in between), just as we may see a different volume of provisional ballots (higher or lower, or somewhere in between, the two benchmarks of 150,000 and 200,000).

Also, we need to factor in a candidate’s potential aversion to conceding the election too quickly. Suppose, for example, that President Obama is behind at the end of Election Night by 50,000, with the same number of provisional ballots cast as in 2008. Presumably, it would be foolish for him to concede under this circumstance, since he added 55,000 to his lead in 2008. If he has the same “yield” rate, he would be able to overcome a 50,000 deficit.

Now suppose, however, that Obama is behind by 80,000 with the same number of provisional ballots (200,000). He would have to substantially increase his yield rate from 2008, to 40%, in order to overcome this deficit. It seems like a long-shot, but then in 2008 (as indicated above) he more than doubled the “yield” rate from Kerry’s in 2004. Perhaps it would be a “hail mary” to get it up to 40%, but would he want to concede before letting the process of evaluating provisional ballots play out? That’s a political judgment he would need to make, and I’m in no position to assess the politics of it. But it seems like the kind of consideration that is relevant in thinking about the numbers. And, of course, if there are more provisional ballots this year than in 2008—say 250,000, instead of 200,000—then he would only need to increase his “yield” rate to 32% to overcome a deficit of 80,000. That still might be hard, but moving from 27% to 32% might be plausible enough to make a concession seem premature in that circumstance.

Therefore, I’ll be looking at the lead and the volume of provisional ballots, and assessing what “yield” rate would be necessary to overcome that lead, given the specific volume of provisional ballots.

Of course, the circumstances are different if Obama has the lead in the initial returns. Given the propensity of provisional ballots to favor Democratic Party candidates, it would seem much more difficult for Governor Romney to overcome an Obama lead by relying on provisional ballots. He might be able to make up a relatively small gap by relying on uncounted (and potentially countable) absentee ballots. Even so, he presumably needs to hope that he has a comfortable lead in the initial returns tonight; otherwise, he faces the possibility that Obama can catch up by relying on provisional ballots.

A late-breaking controversy has emerged over provisional voting in Ohio. It concerns the official form used to indicate whether or not the provisional voter has shown the poll worker a valid identification. Here’s a preliminary analysis.

The form contains a space for the provisional voter to write down the last four digits of the voter’s Social Security Number (SSN), if that is the form of ID the voter wishes to use. Alternatively, the voter may write down his or her Ohio driver’s license number (DLN), and the form provides a separate space for that.

Beyond those two methods of identification, the form also contains several boxes the provisional voter may select depending upon which other type of identification the voter presents. There is a box for “military identification card”; one for current utility bill, bank statement, or other similarly acceptable document; and another for a government-issued photo ID. Finally, there is also a box for a voter who does not possess any of the permissible types of ID and who fills out a separate form swearing so.

The controversy concerns what happens, when it comes time to determine the eligibility of a provisional ballot, if none of these boxes are checked—and the SSN or DLN spaces are also blank—and the provisional voter has failed to submit a valid type of ID by November 16. The form itself warns the voter, albeit in somewhat smaller print: “If you do not check one of the following boxes . . . the board of elections will conclude that you did NOT show ID to your [poll worker] and you must show ID at the board of elections during the 10 days after the election for your vote to be eligible to be counted.” The form also declares sideways, at its left-hand edge, in larger bold type: “MANDATORY INFORMATION REQUIRED FOR YOUR BALLOT TO COUNT.”

In addition to the language of the form itself, the Secretary of State has given the local boards of election a directive on what to do when evaluating the eligibility of provisional ballots. This directive was released just this past Friday, November 2—a reflection of the fact that the rules for counting provisional ballots in Ohio have been the subject of ongoing federal-court litigation for months (indeed years), and that one aspect of this litigation was resolved only last Wednesday, October 31, in a ruling by the U.S. Court of Appeals for the Sixth Circuit.

This directive states that the local board must reject a provisional ballot if “the voter did not provide identification on the provisional ballot affirmation” and did not give to the board by November 16 the missing ID. The directive does not speak explicitly in terms of the boxes (and SSN and DLN spaces) on the form being blank, but the reasonable implication  is that the way the board determines that “the voter did not provide identification on the provisional ballot affirmation” is by seeing that the boxes (and spaces) are blank. Therefore, if the boxes (and spaces) are blank, and the voter did not give the board a valid ID by November 16, then the directive instructs the board to reject the ballot.

A group of plaintiffs who have sued Ohio over its provisional voting rules are now complaining that the form and the directive put the responsibility on the voter, rather than the poll worker, of recording which type of ID the voter uses, by checking the corresponding box (or filling out either the SSN or DLN space). The plaintiffs contemplate a situation in which the voter actually presented to the poll worker a valid type of ID, but did not check any of the boxes (or complete either the SSN or DLN space), and neither did the poll worker. In this situation, the plaintiffs seem to imagine, the voter will not think he or she needs to get ID to the board within 10 days (even though the form warns that failure to check a box has this consequence), and thus the ballot will end up being rejected although the voter actually did present a valid ID. The plaintiffs want instead the poll worker to bear the obligation in this situation to make sure that the box is checked (or one of the two spaces is completed) and, if it is not, for the ballot to count even though the board has no evidence that the voter actually presented to the poll worker any ID.

The plaintiffs point to a state statute, Ohio R.C. § 3505.181(B)(6), which could be read as requiring the poll workers to make sure that one of the boxes is checked (or either the SSN or DLN space is filled out). But plaintiffs have brought their new complaint to federal, not state, court. Insofar as the plaintiffs are arguing that Ohio’s Secretary of State is violating Ohio law, the Sixth Circuit previously has made clear that the proper forum for that issue is state, not federal, court.

Conceivably, there might be two possible bases in federal law for plaintiffs’ new claim. One would rest directly on the Fourteenth Amendment of the U.S. Constitution. The other would rely on a consent decree that the plaintiffs secured with the previous Secretary of State, Jennifer Brunner, on April 19, 2010.

I do not see how a claim resting directly on the Fourteenth Amendment could succeed in light of the Sixth Circuit’s ruling on October 11. There, the appeals court explained that it was not an unconstitutional burden to require provisional voters to print and sign their name on the same form now in dispute. Given this, there seems no chance that the Sixth Circuit would consider it an unconstitutional burden to require provisional voters, rather than the poll workers, to check the relevant ID box (or fill out either the SSN or DLN space) on the same form. Indeed, in its October 11 opinion, the Sixth Circuit included the requirement to “provide identification” as one of the form’s three “rather simple instructions,” although the court observed that the ID requirement was not under challenge at the time.

Explaining how the plaintiffs might be relying on the consent decree is more complicated. But first it is worth pausing to consider the timing of the plaintiffs’ new complaint. The form, with its boxes for voter to check (or spaces to fill out), was adopted on January 4 of this year. It was used in Ohio’s presidential primary in March, as well as in a special election in August. If the plaintiffs believed that the form violated the 2010 consent decree, they should have raised this issue long before last Thursday, November 1, when they went to court about it. As the U.S. Supreme Court has explained, as has the Sixth Circuit, last-minute federal court filings to alter the rules for casting and counting ballots are disfavored, most especially when they could have been brought much earlier.

To be sure, the Secretary of State’s new directive did not exist until Friday (in large part, as already indicated, because of all the antecedent litigation). Even so, Thursday’s filing shows that the plaintiffs’ new complaint concerns the form at least as much as the directive and thus could have been raised in January or at least over the summer, when the plaintiffs presented other issues concerning the enforcement of the consent decree. Moreover, the language of Friday’s new directive does not significantly alter the basic fact that it is the form itself that puts the onus on the voter, rather than the poll worker, to make sure that an ID box is checked (or SSN or DLN space is filled out)—and, if it is not, then the ballot will be rejected unless valid ID is supplied within 10 days after Election Day.

Assuming that it would be appropriate at this late date for the federal court to reach the merits of plaintiffs’ claim that the form (coupled with the directive) violates the consent decree, several subsidiary issues emerge.

First, the consent decree makes plain (in multiple places) that the scope of its substantive provisions is limited to those provisional voters who use “the last four digits of [their] social security number” rather than a different type of valid ID. Section 5.a.vi of the decree, for example, states that a ballot must count if, among other things, “[t]he provisional ballot affirmation includes the last four digits of that voter’s social security number, which is not found to be invalid.” Section 5.b. generally provides that a local board of elections “may not reject a provisional ballot cast by a voter, who uses only the last four digits of his or her social security number as identification, for any of the following reasons” (italics added). One of those following reasons—the one relied upon by the plaintiffs in their new complaint—is subsection 5.b.vii: “The poll worker did not complete or properly complete and/or sign the provisional ballot application witness line and/or the provisional ballot affirmation form, except for reasons permitted by the governing statutes.”

The immediate difficulty is this: how is one supposed to know that a provisional voter used the last four digits of his or her SSN, and therefore is within the scope of the consent decree’s protections—rather than another acceptable type of ID, and thus is outside the scope of the decree’s protections—if both the SSN and DLN spaces on the form are blank, and all the boxes corresponding to the other alternative types of ID are all unchecked? To count ballots when no type of ID has been indicated on the form would seem to go well beyond the decree’s limited applicability. The key point here is that the provisional ballot form seemingly must show that the voter used the last four digits of his or her SSN to get the benefit of the decree, and yet the circumstance we are contemplating is that the form does not show this.

Second, the specific subsection upon which plaintiffs rely, 5.b.vii, does not explicitly refer to a poll worker’s obligation to complete the ID portion of the ballot affirmation form. Rather, as quoted above, this subsection’s language is more general in nature, referring to the poll worker’s failure to “complete or properly complete and/or sign the provisional ballot application witness line and/or the provisional ballot affirmation form.” The form itself contains a separate section for the poll worker to fill out. It is below the section that the voter completes, including the ID section. The poll worker’s section contains the italicized heading: “Failure by the [poll worker] to complete this section will not affect whether or not this provisional ballot is counted.” Thus, it is possible to construe subsection 5.b.vii of the consent decree as applying specifically to this section of the form—and not applying to the voter’s failure to mark down the presentation to the poll worker of a valid ID.

Third, the Sixth Circuit made clear in its October 11 opinion that the consent decree risks violating the Equal Protection principle of Bush v. Gore insofar as it protects provisional voters who use the last four digits of their SSN rather than a different type of valid ID. Consequently, it would seem necessary to expand the scope of the consent decree—beyond its originally intended limited applicability—to cover all provisional voters regardless of the type of ID presented, in order to avoid the Bush v. Gore problem. But as the district court itself recognized when it refused to extend the scope of the consent decree with respect to the issue of missing names or signatures, the court has no basis for extending the decree if doing so is not necessary to address a potential constitutional violation. Asssuming (for reasons stated above) it is not unconstitutional to require the provisional voters themselves to mark on the form the type of ID they supplied, then to extend the consent decree beyond its original scope seems unwarranted. It would cure the Bush v. Gore problem just as much to retract (or interpret) the scope of the decree, so that it provides no relief to any provisional voter who fails to mark down the type of ID supplied. Retracting the decree in this way is what the district court did with respect to the issue of missing names and signatures, and it would seem that the same approach is called for with respect to this new issue of a missing indication on the form of ID that allegedly was in fact presented to the poll worker.

For all three of these reasons, it seems problematic for the federal court to grant the relief the plaintiffs now request, even if it were appropriate for the court to reach the merits of plaintiffs’ new claim. Ordinarily, I am hesitant to evaluate the merits of a pending issue before a court until I’ve had the opportunity to read and reflect upon the full briefing of the issue, from all sides to the litigation. In the present situation, there are more briefs to come this week from both plaintiffs and the Secretary of State (and perhaps others), according to a schedule set by the court. But because there has been considerable media attention to this issue over the weekend, I thought it might be helpful (for members of the public trying to understand what is going on) to offer this preliminary analysis of the issue—even though it is necessarily tentative in light of the still-unfolding nature of litigation.

Two more thoughts: First, the briefing schedule set by the federal district court means that the resolution of this new issue will not occur until after ballots are cast on Election Day. In my judgment, that is regrettable. A basic principle of voting-related litigation, widely recognized among election law scholars, is that it is better to resolve disputes over the ballot casting-and-counting process, if at all possible, before rather than after Election Day. The nature of this new issue is such that, even though it arose for the first time on Thursday and Friday of last week, it would have been preferable to settle it on Monday, November 5, rather than extending it into week. It is true that the effect of any ruling will take place at the time the ballots are counted, rather than cast, but still it would be preferable to settle the counting rules before the ballots are cast. (The plaintiffs may not be asking for a change in the form itself, or in the conduct of voters and poll workers on Election Day, but they are asking for a change in the counting rules as a consequence of Election Day conduct by voters and poll workers. The counting implications of casting events should be determined, if feasible, before the casting as well as the counting.)

Second, for those trying to assess the potential significance of this particular issue, especially in the context of the presidential election, it is worth mentioning two numbers. In 2008, Ohio rejected 1990 provisional ballots for flawed or missing ID. Also that year, Ohio rejected 2201 provisional ballots for a flaw in printing or signing the provisional ballot envelopes. These numbers give some kind of idea about the potential size of problem this year if provisional ballots are rejected because there is no indication on the form that the voter presented a valid type of ID.