by Greg Palast
If you liked the way Florida handled the presidential vote in November, you'll just love the election reform laws that have passed since then in 10 states, and have been proposed in 16 others.
These laws mandate a practice that was at the heart of the Florida debacle: computer-aided purging of centralized voter files. The laudable aim is torid registries of the names of the dead, as well as of felons and others legally barred from voting. But the likely result will be the elimination of a lot of legitimate voters and an increased potential for political mischief.
Traditionally, American voter registries are managed at the county level, and they've been managed pretty well. Certainly there are flaws, but in the overwhelming majority of cases these are clerical errors, not election fraud. Think about it: A citizen dies, another one leaves town. But the dead guy doesn't vote, and the one who moved simply registers and votes from his new home. No harm, no foul.
Trying to cure that relatively minor problem by creating a single, statewide database of voters is overkill. More importantly, it cannot be counted on to fix a very different problem that really is worth worrying about -- purposeful voter fraud. Rather, it creates the potential for new errors on a much greater scale and opens to the doorto political manipulations that are harder to detect than the old ballot-stuffing games, and nearly impossible to prosecute.
In Florida, a state-run purge removed thousands of legal voters -- more than half of them black -- in the months leading up to last fall's election. Most had no idea what had happened until they showed up at the polls. As the U.S. Civil Rights Commissionwrote in a report made public last week, it was this "widespread voter disenfranchisement" -- much more thanany hanging chads or butterfly ballots -- that was the "extraordinary feature" of the dubious Florida vote.
What happened there in 2000 can too easily be repeated in 2002 and thereafter, anywhere computer-aided purges are mandated.
Florida adopted its purge program in the aftermath of the 1997 Miami mayoral race, which was so marred by crooked voting that courts later reversed the outcome. The state contracted with Database Technologies (which soon was bought by ChoicePoint Inc. of Atlanta) to compile a central list of Floridians who were ineligible to vote -- notably, convicted felons -- and cross-match them with voter rolls.
In June 2000, on the basis of ChoicePoint's results, the Florida Division of Elections ordered county election officials to remove from their registries some 58,000 residents unless the counties had evidence that they were not in fact convicted felons.
One of those 58,000 was Linda Howell, who says she has never committed a felony. Her protest was immediately taken seriously, since she happens to be Madison County's election supervisor. The false accusation shook her faith in the purge process: "It really is a mess," she told me afterward.
Another was the Rev. Willie D. Whiting Jr. In fact, Whiting's rap sheet contains a single traffic ticket. His biggest "crime" was his resemblance on paper to Willie J. Whiting (no Jr.), a convicted felon born two days after the reverend.
Fortunately, Whiting lives in Leon County -- which, alone among Florida's 67 counties, independently researched every name on its scrub list. The county could only verify that 34 of the 694 cited -- 5 percent -- had criminal records. Because officials were skeptical of the list, Whiting was able to convince them he should be permitted to vote when he turned up on Election Day.
Researchers from Salon.com who investigated the lists in 13 Florida counties found that at least 15 percent of the names should not have been there. ChoicePoint spokesmen subsequently told me they don't dispute that figure, and they consider it a reasonable rate of error.
However, the company also defends its scrub list as "accurate" -- because its standard is that the list accurately records all names found in accordance with the specifications devised by the stateofficials who supervised the work.
And that's the problem.
The reason so many wrong names ended up on the scrub list is that Florida ordered ChoicePoint to input questionably broad matching criteriainto its sophisticated computer programs. According to various statements by ChoicePoint officials, the criteria were:
First four letters of the first name. Middle initial. Gender. At least 80 percent of the letters in the last name. Approximate date of birth. Last four digits of Social Security number when available -- which was not often, since fewer than 10 percent of Floridians had that number on their voter registration forms. Certain variations were also programmed in (Willie could match William; John Richard could match Richard John).
There are a lot of Whitings and Howells -- and Browns and Brownes and Harwoods and Haywoods -- so given these rough standards, massive misidentification was almost guaranteed.
The results were then skewed because of another piece of data passed on to local officials: the voter's race. Next to every alleged felon's name was the designation "BLA" or "WHI." This meant that if someone named, say, John Smith was identified as a felon, and 10 matches came up, only the ones of the same race were likely to be purged. This is logical, but consider the implications: Since about half of felony convictions involve African Americans -- while barely one in seven Floridians are black -- this methodology ensured that a disproportionate number of law-abiding black voters would be disenfranchised.
If Salon's 15 percent error figure is right -- and data like Leon County's indicates it ismuch higher -- almost 9,000 of the 58,000 names on the scrub list belonged to rightful voters. (Furthermore, 2,883 other names belonged to people convicted of felonies in states that restore voting privileges after a sentence is served. These people were also purged -- even though they should not have lost their civil rights merely by moving to Florida.)
Database experts say the state could have reduced mismatches to one in 500 by expanding the match criteria -- making name matches more specific and adding such elements as current and prior address and suffixes like "Jr."
But Florida officials counter that if they had done that, they would have erred equally in the opposite direction -- allowing ineligible people to vote.
You would think other states would run from Florida's methods. But in their current legislative sessions, Colorado, Indiana, South Dakota, Texas, Virginia, Georgia, Kansas, Montana and Washington have passed bills that -- while varying in specifics -- would follow the Sunshine State's lead in centralizing, computerizing and cleansing voter rolls. Sen. Christopher S. Bond (R-Mo.) has introduced a bill in which certain conditions in any state would trigger mandatory voter list purges.
To a largeextent, these bills are a response to "motor voter" legislation, which has added millions of citizens, particularly minorities, to voter registries. Since minority voters tend to be Democratic, it is not surprising that motor voter laws are popular among Democrats, and most of the bills attempting to purge the rolls are sponsored by Republicans.
But many factors go into the ill-advised rush to reform. Take the case of Georgia.
The day before the November election, the Atlanta Journal-Constitution and WSB-TV jointly reported that records indicated that deceased Georgians had voted 5,412 times over the last 20 years. They specifically cited one Alan J. Mandel, who apparently cast his ballot in three separate elections after his demise in 1997.
Subsequently, a very live Alan J. Mandell (note the two L's) told the secretary of state that local election workers had accidentally checked off the wrong name on the list. That may or may not explain what really happened -- but in the midst of the chad-mania that dominated the headlines last fall, details became less important than the newly energized drive for so-called reform. Under a law signed April 18, Georgia's secretary of state controls "list maintenance" and, for example, has taken over the power of deleting the names of dead voters.
This is too big a response to a little problem. In a state of 8 million people, 5,412 possibly bad votes -- not in one election, but in all the primaries, local elections and general elections that were held over 20 years -- is an extremely small number. If reformers really want to clean up voting lists, they should consider more funding for local election boards so they can do the job more directly, and better.
The centralization of state voter registries hands an all-too-tempting monopoly to whichever party controls the office of secretary of state. The highly technical (and, where contractors are involved, commercially confidential) nature of computer-aided purges makes bias in the cleansing of supposed felons, deceased voters and duplicate voters astonishingly easy to carry out and difficult to uncover.
Even uncovered, apparent bias is difficult to challenge. Florida officials defend their methods, saying they wanted to cast the widest net to catch the highest number of ineligible voters. Who is to say that, despite the devastating racial imbalance in the roster of those wrongly purged, Florida chose the wrong match criteria? One Florida Democratic leader admitted to me (on condition of anonymity) that, if he and his party had control of the voter files, they would have been tempted to pick some elimination criteria of their own.
After all, one man's overzealous purge is another man's inauguration.
Greg Palast, who has worked as a consultant to 19 states on government regulation, investigated the Florida vote for BBC television, the Observer of London, Salon.com and the Nation magazine.