Out for the count

The decision by a court to name Felipe Calderón president of Mexico brings back memories of President Bush's first election win. Link.

September 7, 2006 12:05 PM | Printable version

Felipe Calderón has been named president of Mexico, by a court, much as George Bush was named President of the United States, by a court.

But did he win the election? We do not know. The court's decision does not establish this, any more than the Bush v Gore case established that Bush won his first election - which, as we now know, he did not.

In both cases, the truth could have been known in time. But it was not. And that is because one side - in the legal struggle, the winning side - refused and resisted a full recount of the votes.

In both cases, we can be sure that if plaintiff and defendant had been reversed, the decisions would have been unanimous the other way. If Bush had been for it and Gore against, the US supreme court would have voted 9-0 for a full recount in Florida. If Lopez Obrador had led the count by a minute fraction, and if it were Calderón charging irregularities and fraud, then the Mexican electoral court would have recounted all the votes.

On July 16, I summarised in this space the principal irregularities found in the preliminary report of the vote (the PREP), in an analysis by the physicist Luis Mochán. They were: (1) an improbably stable absolute differential between Calderón and AMLO, throughout the count; (2) an improbably low AMLO total in the first 10,000 boxes, which were not included in the PREP, suggesting that he might have started the night with negative votes; (3) obvious backtracking and mistabulation in the late hours of the count, and (4) a non-normal distribution of the differentials across precincts between Calderón and AMLO, suggesting that in many close precincts small numbers of votes could have been shifted to Calderón.

Since then, not one of these anomalies has been explained. The court decision did not address them. Instead, we have seen an intense propaganda effort, aimed at establishing that the Mexican electoral procedures were, and are, incorruptible. A former high Mexican official with a well-deserved international reputation as a democrat - since it was a private communication and he is a friend I will not name him - told me that they were "bulletproof".

Yet the quality of the procedures is the easiest thing in the world to test, and Luis Mochán has now made a new analysis, which tests them. His test concerns a very simple, very basic number. It is a number that, in an, impeccable process, should be known for sure. How many votes were cast in this election?

Mochán points out that there are four separate ways to count this number. And remarkably, all four counts were not only used, but made available in the reporting of the preliminary count. Therefore, we have the capacity to check on the perfection of the counting process.

The first way to do this is by counting the number of voters who signed in, receiving an official stamp by their name on voting list. The number of stamps equals the number of voters. This count was made.

A second way, is to count the number of ballots received by the voting station, less the number left over at the end of the voting process. This was done.

The third option is to count the total number of ballots cast, before separating them into piles and recounting who they were cast for. This, too, was done.

Finally, you can count the number of votes cast for each candidate, plus the number of write-in votes, plus the number of blank ballots. Add these together, and you should have the total number of ballots cast. Obviously, this also was done, in every polling place.

In a perfect process, all four methods should yield exactly the same result. And while some human error is normal, and to be expected, error inherently means that the process was not perfect. Errors, of course, occurred; the process was not perfect. The real question is: how substantial were they?

Mochán 's analysis yields the following figures, among others:

Among 130,488 domestic polling places, the records for 13,201 (just less than 10%) are missing altogether from the preliminary count.

Among the 117,287 records that are present, 24,148 are incomplete, missing one or more numbers. The incomplete records correspond to 21% of the polling places reported. While these mainly reflect blank votes and write-ins, and do not directly affect the election outcome, the missing data makes it impossible to know exactly how many total votes were cast in those precincts.

Among the 109,134 records for which one can compare directly the number of ballots in the box against the number of ballots received and the number left over, there were 17,465 records where the number of ballots deposited was larger, with the difference equal to 788,077 votes. There were also 32,758 records where the number of ballots in the box was less, by 716,489 votes. In total, this type of error occurred in more than 50,000 ballot boxes, that's more than 46%, with a grand sum of 1.5m missing or excessive votes.

In 97,790 records where one can compare the number of citizens voting with the number of votes counted, there were 22,419 where the number of votes counted exceeded the number of citizens voting, and 22,391 where the number of citizens voting was less. In total, this type of error affected 46% of the ballot boxes, and involved nearly 1.8m votes.

In 107,425 cases one can compare the number of ballots deposited and the number of voters signing in. Discrepancies between these two numbers affect 41% of the boxes, and about 2.35m ballots.

The bottom line of this back-breaking analysis is that the preliminary vote count was affected by basic inconsistencies in tens of thousands of boxes, with the inconsistencies cumulating to millions of votes. This, in an election decided, by the official count, by less than a quarter of a million votes.

Were the problems corrected in the official, or "district count"? One cannot tell, because the detailed data made available in the PREP were not published for the District Count. Yet we know that the number of ballot boxes recounted, vote by vote, in the District Count was much lower than the number showing inconsistencies in the PREP. Thus we have no reason to believe that the District Count was any better.

Nothing here establishes actual fraud. The errors discussed above may have been entirely innocent, for all we know. But they are so large that they rule out accepting the diktat of the court, as to the winner of the election. In the final analysis, we do not know who won Mexico's presidential election. And without a full recount of all the ballots, no one will ever know.

Mochán 's analysis contains the following severe summary of the facts:

"(1) The PREP presented obvious errors; (2) These errors show that [Mexican election] officials have the capacity to interfere with the computers that made the reports, adding, eliminating and modifying data; (3) The computer system, or at least that part in charge of reporting the results from the PREP, is not robust and it may be interfered with; (4) The absence of an explanation of these anomalies and the enormous propaganda pretending to induce the notion of a perfect process can not but produce distrust about the other stages in the election process."

In its decision, the Mexican election court compounded this problem with a type of reasoning quite familiar in the United States, where it cropped up over the Ohio vote in 2004. Yes, they agreed that seals were broken on the voting packets. Yes, they agreed that Lopez Obrador suffered illegal propagandistic attacks. Yes, they agreed that the private sector and Vicente Fox had participated illegally in the election. But because it could not be proved that these illegalities were sufficient to turn the election, they accepted the declared result.

This is a bogus argument. Where massive illegality afflicts an election campaign and a voting process, the result is not legitimate. It therefore cannot be accepted as legitimate. In Ohio in 2004, there was undeniable, massive partisan interference in access to voting machines - which I witnessed with my own eyes - and many other improprieties in the count. It doesn't matter whether Bush won Ohio or not. He won it - if he won it - illegitimately, by techniques that amount to unprosecuted crimes. Therefore, his second term is as illegitimate as his first. The same will now be true of Felipe Calderón.

Meanwhile, American democrats have much to learn from our Mexican friends, who have been fighting for democracy far more toughly than we did, in 2000 or 2004.