Monday, August 27, 2007

Bob Riley and the Mysterious Votes

While we are examining ethical issues surrounding Alabama Governor Bob Riley, let's look at the strange election that put him in office in the first place.

Several journalists have reported on a paper by Auburn University professor James H. Gundlach, analyzing possible electronic manipulation of vote totals in Baldwin County during Alabama's 2002 governor's race. Scott Horton, of Harper's referenced Gundlach's analysis here.

Steve McConnell, a reporter for Gulf Coast Newspapers, included an interview with Gundlach in a recent in-depth article about the Riley-Don Siegelman race of '02.

While Gundlach's work is fairly well known, I'm not sure it is well known that his paper can be read on the Web. The paper is available in PDF format here.

Some highlights of Gundlach's work:

* He says three factors raise suspicions about the returns: (1) An unusually large increase for the 2002 Republican candidate (Riley) over the 1998 GOP candidate (Fob James); (2) The one-third reduction in Siegelman's vote total from the first to the second report; (3) Computerized vote tabulation, which provides no method for producing two different results, save human intervention.

* Gundlach states: "When Baldwin County reported two sets of results, it was clear to me that someone had manipulated the results. There is simply no way that electronic vote counting can produce two sets of results without someone using computer programs in ways that were not intended. In other words, the fact that two sets of results were reported is sufficient evidence in and of itself that the vote tabulation process was compromised."

* Gundlach has a theory as to what happened. "My hypothesis is that someone was moving a little more than 3,000 Baldwin County votes from Siegelman to Riley by calculating a fifth of Siegelman's votes in each voting district, rounding it to a whole number, adding the resulting value to Riley's votes in that district and then subtracting that number from Siegelman's vote. However, instead of subtracting the calculated number, they added it to the vote for Siegelman. This is a common error created by using copy and paste to produce the invisible formulas for cells of spreadsheets. The result was a first report of county vote totals that had percentage distributions close to what was expected but a total vote that was much higher than expected. Once they went back and fixed the procedure so that it performed as they desired, a reasonable total vote and Riley winning the election, the difference between the first and second reporting of Siegelman's vote was twice the number of electronically shifted votes. If what I hypothesized happened, then the total votes for Baldwin County was 27,866 for Riley and 15,283 votes for Siegelman. This would have produced state totals of 669,039 for Riley and 671,652 for Siegelman."

* How could the vote be manipulated? Gundlach offers four scenarios and focuses on one in particular. "The fourth approach, and the one I would take if I were to do it, would be to install an 802.11 card on the tabulating computer, along with enabling software, and use a similarly equipped laptop in a nearby room to modify the data files immediately after they were read from the cartridges. This would simply require access to the tabulating computer at some time before the election to install the card and after the election to remove the card."

I'm sure some would like to call Gundlach a crank, or worse. But the main raises some serious issues, and I know of no one in authority who has taken a look at the key points he makes.

11 comments:

  1. Gundlach's problem is that he does a statictical analysis instead of actually looking at what happened. See the following from the Nov 8, 2002 New York Times to see what happened. Note: If siegelman's vote total for Magnolia Springs had stood there would have been at least 5,700 more votes than voters that day.

    THE 2002 ELECTION: ALABAMA;
    Full Recount Asked in Race For Governor Of Alabama

    BYLINE: By JEFFREY GETTLEMAN

    SECTION: Section A; Column 6; National Desk; Pg. 26

    LENGTH: 717 words

    DATELINE: MONTGOMERY, Ala., Nov. 7


    Down by a tally of 3,195 votes, Gov. Donald Siegelman refused today to swallow defeat and instead demanded a statewide recount.

    The project could take weeks, maybe even months, stretching out the awkwardness of two men, Mr. Siegelman, the Democratic incumbent, and Bob Riley, the Republican challenger, each pressing ahead as if he were the next governor.

    Mr. Siegelman said he was not seeking the recount for himself, "but for the people."

    At a news conference tonight, he said: "This is not about Don Siegelman. This is not about Bob Riley. This is about the people of Alabama and what's best for them."

    The Riley campaign was not moved. "Governor-elect Riley is going ahead with his transition team," Mr. Riley's campaign spokesman, David Azbell, said. "Bob Riley is the next governor of Alabama. Period."

    Lawyers with the Riley campaign later said that they did not think the governor had grounds for a statewide recount and that they would contest it,
    Alabama officials said that any voter could ask for a recount, county by county, as long as the voter puts up a security bond to cover the cost, should the results remain the same. The cost is expected to be around $5,000 a county, to be paid by the Siegelman campaign.

    Mr. Siegelman plans to use Democratic voters in each of Alabama's 67 counties to protest the vote. He is behind by about two-tenths of 1 percent of the 1,361,873 votes cast.

    The trouble started election night in Magnolia Springs, a town near the Gulf of Mexico, where unofficial poll numbers were released to the news media that erroneously put the incumbent ahead. The mix-up was straightened out a few hours later, but not before Mr. Siegelman declared himself winner, proclaiming, "How sweet it is!"

    State election officials said there was no way he could have won the 7,000 votes he was at first reported to have gained from the Magnolia Springs precinct. Only 1,300 votes were cast there.

    "If Siegelman's guys are looking for votes, they'll have to go fishing," Charles E. Grainger Jr., legal adviser to Alabama's secretary of state, said.

    Each man has been acting very much like governor. Mr. Siegelman was consoling residents today hit by a tornado in the state's southeast corner. Mr. Riley was in Tuscaloosa, meeting with University of Alabama trustees, talking about his transition plans.

    Montgomery, the state capital, seemed the most confused.

    "Congratulations Governor Siegelman!" read one banner across the street from the Capitol. Next door, stretched an even bigger display: "Congratulations Bob Riley!"

    Starting tomorrow, the Siegelman campaign will begin asking for the recounts, county by county. Under Alabama law the results of the recount cannot change the outcome of the election unless the Legislature intervenes. The earliest that would be is Jan. 14, when lawmakers return for the next session. Democrats control both the House and Senate in Alabama.

    Siegelman campaign officials said they would not contest the election in front of the Legislature if the recount affirmed Mr. Riley's lead.

    They also said they had no plans to go to court.

    "We're not challenging or contesting this election; all we're asking for is a recount," said Joe Espy, a lawyer with the Siegelman campaign. "This is not going to be another Florida."

    There may be other avenues, though, the Siegelman camp can pursue.

    There were several reports of improper absentee ballots, mostly from Republican counties.

    In Birmingham, someone posted fliers in largely black areas telling voters they could not vote unless they first paid their fines and their rent. Democratic operatives said those fliers cost Mr. Siegelman hundreds of votes.

    Mr. Siegelman, a lawyer and career politician first elected governor in 1998, won most of his votes in cities. It was there that voters stuck by him, despite accusations of ethical lapses and a federal investigation of his personal finances.

    Mr. Riley, formerly a little-known three-term congressman from largely rural east-central Alabama, polled well in rural areas like Magnolia Springs and surrounding Baldwin County. That was one reason the Siegelman camp suspected that something funny had happened to the 7,000 votes on election night.

    Local officials said that was not true.

    "If there is a recount, they will not change," Probate Judge Adrian Johns said.

    ReplyDelete
  2. Excellent explanation from the Birmingham News.

    Birmingham News (Alabama)

    November 10, 2002 Sunday

    EARLY PRINTOUT WAS ONLY SOURCE OF WRONG TALLY

    BYLINE: JOHN ARCHIBALD and BRETT J. BLACKLEDGE News staff writers

    SECTION: NEWS

    LENGTH: 1662 words



    The confusion surrounding Alabama's gubernatorial election stems from a problem with a computer's summary of votes in one county precinct, not from problems with the actual votes recorded in election machines, according to Baldwin County election records and those familiar with the voting machines.
    The error - which made it seem on Tuesday night that Gov. Don Siegelman received 6,334 more votes in Baldwin County than he actually got - occurred only on a printout designed to summarize the vote totals for interested political operatives and the news media. It did not affect the votes cast or ultimately counted.

    A problem in the way officials downloaded vote data from a computer cartridge - which contained correct vote totals at the precinct level - triggered the mistake and "fried numbers" in the summary sheet.

    It wasn't an unusual error, county election officials across the state said. In fact, Siegelman's vote totals were not the Election, Page 10A 1A only ones misrepresented on the summary sheet. A local state Senate race also had incorrect totals on the sheet, but there were accurate counts on the more detailed precinct-by-precinct report.



    The real problem surfaced because Baldwin officials released the summary report before double-checking the printout's accuracy.
    The mistake was noticed quickly by U.S. Rep. Bob Riley's campaign and other election onlookers who saw the county's more detailed vote totals, by precinct, on Baldwin's Web site. Baldwin officials met Wednesday morning to correct the error on the summary sheet.

    There was never a recount, and votes were never altered, said David Pimperl, director of communications and information services for Baldwin County.

    But with the dissemination of the faulty summary report, the damage had been done. On election night, Siegelman seized on the report, which showed he received 19,070 votes in the county, enough to put him over the top in his bid for re-election. The actual vote total on the cartridge was 12,736, Pimperl said.

    The change in reported ballot totals prompted Siegelman and his aides to question why votes cast for him "disappeared." Siegelman said Wednesday morning he believed that someone returned to the Baldwin County courthouse after everyone else left election night and recounted the votes.

    "Maybe they didn't disappear, but it sure seems like they did," Ted Hosp, Siegelman's legal adviser, said during a television interview Friday.

    The confusion over the differing vote totals and how they were produced has created one of the fiercest election battles in Alabama and has prompted Siegelman's call for an unprecedented recount of every ballot in the state. He based his request on discrepancies in the Baldwin totals, saying they suggested a problem with how ballot totals were recorded on Election Day in voting machines.

    But Baldwin election officials have said that because the actual record of the votes in the machine was never at issue, they regarded the faulty summary sheet as one of many unofficial and incomplete tallies of results distributed by the state's 67 counties on the night of the election. The counties produce the unofficial summaries for the media and candidates to monitor election results, but officials always warn that the summaries likely will change by the time the results are certified later in the week.

    "It was just an unofficial report handed out to give people some information," Pimperl said. The official record of votes was canvassed from all Baldwin County machines and certified the day after the election. Reporting error

    County election records appear to support the contention that the discrepancy in vote totals was more a problem with reporting summary results and not a problem with the actual casting or counting of ballots.

    The difference between an incorrect summary of accurate vote counts and actual problems with ballots is significant, particularly since Siegelman and others compare the confusion with problems in Florida during the 2000 presidential election.

    In Florida, actual ballots were not recorded properly because of the way they were punched by voters. In Baldwin County, the ballots cast by voters were stored correctly in electronic machines, but the computer's internal calculation of those votes produced for the summary sheet was in error, local election officials said.

    Mark Kelly, a representative for Election Systems and Software, the vendor that manages the election computers in Baldwin, explained the error this way:

    "It's a little fancier version of what would happen in the old days when people would gather in a courtroom for the results to be read," he said. "Somebody would call the numbers out, and somebody else would punch them into a calculator. I might call out 1,360, and somebody might enter 1,630 in the calculator. When it was summed up and written on a chalkboard, the total would be wrong, but it wouldn't change the vote.

    "It was something like that, only fancier," he said. "There was not a recount. The votes were counted by machine, they were counted on the spot. There is no more counting of ballots."

    A number of reports, such as the faulty summary, are generated from the raw data that is collected at the precinct level and transferred to the county's election database.

    Magnolia Springs

    Baldwin's troubles began in the Magnolia Springs polling place, a precinct where 1,290 people cast votes for governor Tuesday. Of those, 342 voted for Siegelman and 910 for Riley.

    At the precinct that night, election officials took the computer cartridge on which the ballot data from machines was stored and inserted the cartridge into the computer. That computer read the cartridge correctly, Kelly said. The cartridge, which records all the votes but does not have the capacity to alter them, was then taken to the sheriff's department with all the others.

    The cartridge, something similar in appearance to an eight-track tape, apparently was misread by another computer at the sheriff's office to produce the summary sheet. The misrepresentation of vote totals was printed on the sheet and distributed.

    It was that summary sheet containing the miscalculation that Siegelman re lied on to claim victory early Wednesday and to cheer, "How sweet it is."

    And it was that summary sheet, complete with the miscalculated vote totals, that Baldwin election officials distributed late Tuesday night to the candidates' representatives and the media, notably The Associated Press, a news cooperative that distributes election results to media organizations across the state. The AP used the faulty numbers on the sheet to declare Siegelman the winner about an hour after the governor had declared himself re-elected. AP pulled its declaration later Wednesday.

    Baldwin County officials produced a detailed listing on election night of every vote cast for each candidate in every race, by precinct, on Baldwin's Web site about the same time they distributed the inaccurate summary sheet to the media and poll watchers. That report was created from the same voting data stored in the county election computer that was used to produce the inaccurate summary of votes for the media and candidates.

    Election officials also produced paper printouts, created from votes stored in each precinct machine, that showed the actual vote counts that matched original totals calculated at each polling place. Fact and fiction

    Siegelman and Riley have both said the votes were counted and recounted in Baldwin since Tuesday. They're both wrong, county election officials said Friday. The only count taken and recorded in the machine occurred when voters fed their ballots into the machine on Election Day.

    Siegelman's advisers acknowledge now that they have no specific reports or evidence of another ballot count occurring after the courthouse emptied on election night. But they say the mere fact that a miscalculation occurred at all should be enough to justify a recount in every county. And they said it was suspicious that there were no misrepresented numbers in other Baldwin races.

    But the summary sheet also contained errors in the county's state Senate District 32 race, which was printed directly across the page from the vote summaries for the gubernatorial race.

    In the Senate race between Libertarian Richard Medicus and Republican Bradley Byrne, the summary report tallied 3,431 votes for Medicus, 32,743 votes for Byrne, and 13,935 write-in votes. The precinct tapes, and the numbers later certified, showed the same numbers for Medicus and Byrne, but only 235 write-in votes.

    "That tells you right there that you've got fried numbers," Kelly said of the summary sheet totals.

    Also, examinations of Baldwin County election records indicate that the vote totals on the summary sheet could not have been accurate because they showed far more people voting in the governor's race than actually voted in the county.

    The summary sheet listed 51,178 total votes in the governor's race when there were only 45,032 total ballots cast in that county.

    If that mathematical impossibility is not enough, there are other factors that call into question the accuracy of the numbers on the summary. To believe the numbers, one must accept that about 6,300 people voted in the governor's race but didn't touch races including U.S. Senate and lieutenant governor. No other counties showed such a high discrepancy.

    In the final vote count, for instance, 25 voters in Baldwin County cast votes for governor but did not vote for a senator, records indicate. Statewide, 13,035 voters followed that pattern, a number that averages to 195 voters per county.

    The summary numbers also indicate that Siegelman got 5,000 more votes than Lucy Baxley, who won her race for lieutenant governor. Baxley's numbers in all other counties consistently bested or matched Siegelman's ballot numbers.

    ReplyDelete
  3. Funny how when you refute Roger's arguments with facts he ignores you.

    ReplyDelete
  4. That's because he is a fool. Ever noticed how he only plagarizes what he sees other people write?

    ReplyDelete
  5. I've read both the Gundlach paper and the two articles sent in above. A few points:

    * Gundlach presents a hypothesis on what caused the high number of Siegelman votes from Magnolia Springs. If you read his paper, you can see his hypothesis.

    * Neither the New York Times nor the Birmingham News present any explanation for the strange numbers other than a computer glitch. The News uses the term "fried numbers." Gundlach says computer's can't have glitches without human intervention.

    * Who's right? Well, you might try this: Read the two newspaper articles and Gundlach's paper and ask yourself, "Who seems to know the most about the voting system that was used in Baldwin County that day?" The answer is quite clear to me.

    * Gundlach presents strong evidence that the numbers were manipulated and offers suggestions that would enhance trust in our voting systems in the future. That's something all citizens should support, regardless of party affiliation. Will we ever know what actually happened in Baldwin County? Probably not. But Steve McConnell's piece for Gulf Coast Newspapers is the most up-to-date and comprehensive look at the issue that I'm aware of. I would urge anyone who cares about this issue to read it.

    ReplyDelete
  6. How do you get around the FACT that if the larger number for Siegelman were correct there would have been close to 6,ooo more votes than voters in Magnolia Springs?

    How do you get around the FACT that there is triple redundancy in the scanner system:
    1. the sign-in sheets with voter signature
    2. The poll worker sign-in sheet
    3. The number of ballots counted by the machines.

    All three match up ONLY with the results that were certified and declared Riley the winner.

    Gundlach proposed a theoretical statistical explanation that totally ignores the FACTS.

    ReplyDelete
  7. If you refuse to read Gundlach's paper, I'm afraid I can't help you. That's why I posted it: So people, of all political stripes, could actually read it.

    Gundlach's answers to your questions are in his paper.

    ReplyDelete
  8. Another one of Rogers conspiracy.

    ReplyDelete
  9. ...If you refuse to read Gundlach's paper...

    Yes, I think Gundlach, in the paper and elsewhere, handles all of the serious objections raised above (as so-called "facts", grrrh) quite nicely. And the New and Times work on the Baldwin County voting is pretty sloppy, at best.

    At this point, it might be almost impossible to sort out exactly what happened. But, if, say, we had a working DoJ, putting a bunch of people under oath and sorting it out shortly after it happened would have been worthwhile.

    Give what we know about the election, I come down on Gundlach's side. IMO, the election was stolen. It's, as mentioned, hard to prove but Gundlach makes a much stronger case than those who say it was just some random screw up. And, the context of Rove's team running both the DoJ and nearly every judge in AL just makes the case that the election was stolen stronger.

    Yes, Gundlach's suggestions for stablizing AL elections are good and, yes, both parties should endorse them.

    ReplyDelete
  10. Actually Gunclach's paper does not deal with the fact that in order for Siegelman to have the higher vote total, there would have to have been more votes than voters in Bladwin County.

    It is an abstract statistical analysis that does not look into the facts of the vote at all.

    ReplyDelete
  11. I know this is way after the fact, but I don't need to read Prof. Gundlach's paper to know where he went wrong. It's this statement:

    "Gundlach says computers can't have glitches without human intervention."

    Gundlach is a complete idiot. I don't think I've ever heard anything sillier from a Professor.

    The description of the creation of the summary sheet indicates a cartridge making some sort of electrical interface with some sort of readout equipment, probably an electronic interface with an LED readout. I can think of half a dozen places in a system like that that an error could occur, and I'm pretty sure we've all experienced them ourselves:

    - There could be dirt in the contacts between the cartridge and reader, resulting in a false reading.

    - There could be a fault in the LED, causing a false number or a scrambled number on the display.

    - There could be bad lighting, with the result that the person reading the interface saw something wrong.

    - There could be a dog-tired operator.

    - There could be a weak electrical supply, resulting in a completely erroneous display.

    That's just off the top of my head.

    The just insanely silly statement, "computers can't have glitches without human intervention," is the dead giveaway that Prof. Gundlach engaged in a process that was going to find evidence of fraud whether it occurred or not. Such analyses should be avoided like the plague. The anonymous commenters are correct: the reports from those who examined the actual process are more to be trusted than the statistical analysis of a clearly biased interpreter.

    ReplyDelete