In Colorado we are approaching a primary election on August 12th and bumbling county clerks and other election entities are scrambling to promote mail balloting. Ostensibly they are doing this for the convenience of the voters, to save money, and to increase turnout. At least those are the reasons we've been given in the past for why we should use this fraud-prone method of voting.
Now convenience is used. However, in El Paso County the only time we've had significant lines at the polls are when the clerk let a few precincts in the eastern part of the county run out of ballots. Some counties have been experimenting with vote centers using electronic poll books together with DREs and huge lines have formed when things went wrong with those experiments. So again the problem is with the county clerks, not precinct voting. However, voters are now so scared of what the county clerk has done to precinct voting that mail ballots may seem the better alternative to them. At least they won't have to stand in line until 1:30 AM the next morning to vote as many have had to do in recent Colorado failed election experiments.
So we are down to cost as a reason that has been given in the past for promoting mail ballots. But in a presidential election Colorado law requires county clerks to have all precincts open for voting in any case. Where is the cost savings?
Why then are county clerks repeatedly spending tens of thousands of dollars to send out promotional materials for mail ballots at public expense? Ostensibly, election officials should be neutral as to the choice of voting. But tens of thousands more mail ballots have now been sent because of intense and expensive promotion by our, and other Colorado county clerks.
Recently, under the guise of mailing a "Voter Identification Card" (incidentally these used to be a convenient wallet size and now measure 14 x 22 cm, or 5 1/2 x 8 1/2 inches) the El Paso County, Colorado, Clerk recently sent out another mail ballot promotion that had some very interesting characteristics.
In Balink-ID_theft.pdf I've reproduced a postcard voters were supposed to return to the clerk to get a mail ballot. I don't know about identity theft where you live but here only an idiot would put their name, address, driver's license number, birth date, and signature all together on a postcard. Voters have complained long and loud about just having their signature on the outside of a mail ballot. Now we get this nonsense and these are the people running the elections!
But the problems don't stop there. Note on the postcard (Balink-ID_theft.pdf) that a voter can have three, yes three, addresses. Now to register to vote you must provide a physical residential address, i.e, not a business, vacant lot, etc. As mail service may not be available to that address a voter may use a post office box, etc. So a separate mailing address is allowed. That contributes to the problems with fraud for mail ballots as many ballots may be sent to the same box, or the mailing address may be a forwarding service for someone in Beijing, or some other scam. But the address problem doesn't end there as a voter who wants to sign up for a "permanent" mail ballot is asked to list yet a third address. Presumably, once a "voter" has signed up for a "permanent" mail ballot they will be mailed a ballot to this third address for every election in the future unless "they" don't vote in two consecutive federal elections, their ballot is returned as undeliverable and "they" don't respond to the clerk's query. Note that with mail ballots there is no way to check whether "they" are corporeal now or in the future. And there is really no way to check if someone using a mail ballot is actually eligible to vote in the election.
Now some dirty Democrat in Boulder County who wants to rig El Paso County elections can use a mailing address for their "permanent" mail ballot that is valid and the clerk really has no way to check once the "they" are initially registered (hey, it happened to me and I only found out by accident). Of course no god-fearing Republican in El Paso County would ever do the same to rig elections in Boulder County but Republicans are going to need a lot of election fraud in 2008 if John McCain is going to win.
If you haven't figured out how to commit election fraud using mail ballots yet then I've got a great bridge (mostly) across Bear Creek you'll definitely be interested in. And I really shouldn't single out the El Paso County Clerk as he and his minions are simply going along with the rest of the lemmings.
Any computer geek will tell you that many of the problems with elections can be stopped or greatly limited with a properly designed and maintained voter registration database, which is true. The problem is we don't have such a database. In the past ten years Colorado has not successfully deployed even one large database and we are on our second try for an election database that will be used for the first time in 2008 elections.
How is that database doing? A letter to State Senator Dave Schultheis (Schultheis-SCORE_report.pdf) written by Al Kolwicz reviews results from Boulder County. Note that we have not been able to examine the entire database, or even been given an entity relationship diagram (ERD), so the review is extremely cursory. But even these results indicate a database disaster, e.g., in the limited data available 35 non-citizens are registered to vote, 33 don't have a precinct number and invalid precinct numbers are present in the database, invalid addresses are included, more ballots can be cast than there apparently are registered voters because of "confidential" voters (sign up for a mail ballot as a "confidential" voter if you want to work from Shanghai), and on and on.
Bear with me for a moment for technical details on a relational database management system (RDBMS) like Oracle, which Colorado is using for its voter registration database, and how the problems outlined in Kolwicz's letter are prevented by what is known as referential integrity (RI). RI is essential in a database if it is to be of any value. For example, two identical social security numbers (SSN) for two different people make that data worthless. Typically the data architect or modeler would put a unique index on the SSN entity in the RDBMS. Also, a SSN is numeric so any alpha characters, e.g., "-", would be stripped out before entering the data. Further, the range would be controlled by a "rule." SSN's are always positive and range from 001010001 to 999999999. Any number outside that range could not be entered into the database if properly designed and RI is intact. Other "rules" would control what data can be entered into other database entities, e.g., a number or odd character like "^" could not be entered into a last name field.
Other entities within the database may be affected when data are inserted, deleted, or updated. For example, if a new SSN is entered then a name must be entered as well, and perhaps other data. Or if a SSN is deleted from the database then the name and other associated information must be deleted as well. To insure all those things happen typically triggers are used with the RDBMS to take care of these relationships and insure RI.
RI may also be insured by stored procedures within the RDBMS, front-end programs, and external scripts (e.g., Korn shell on Unix machines and one prays they have this database on a Unix server). In a well-designed database there are always several layers of referential integrity to insure and maintain data quality.
Obviously, the design and development of a RDBMS is time consuming, expensive, and requires highly-skilled database architects, modelers, and developers, who were in very short supply when I was working in this field. And it is just as obvious from even the very truncated view Kolwicz has been able to obtain of the SCORE II voter registration database that the design and implementation was not done by people with the requisite skills to implement the essential referential integrity. And without proper RI the security of the database is impossible to maintain as well. And to make matters worse we can't even determine that all the requisite entities are defined in SCORE II.
As I was writing this newsletter KIm Zetter published an article in Wired Magazine documenting that in New York State 50% of the new Sequoia voting machines are flawed. Note that these are new machines built after a decade of experience behind the company. However, Sequoia has simply ignored all the critiques of electronic voting machines and simply produced more junk.
The mantra of voting machine vendors and election officials when questioned about the manifold disasters, catastrophes, and security breaches (typically referred to as "glitches") so evident since the widespread implementation of electronic voting is that the "election process" will protect the votes and voters against the weaknesses of the electronic voting machines and guarantee an honest and accurate election. Possibly someone besides election officials still believes that but even the voting machine vendors must be incredulous by now.
To add to the farce our elections have become, county clerks are using mail ballots counted on secret machines hidden in back rooms, and operated either by the voting machine vendors or government employees. History clearly demonstrates that is a steep path to election fraud.
I have been stating these problems for nearly a decade now so there aren't any secrets or revelations here. At this late date the only solution seems to be extensive audits of paper ballots where they are used. And for a meaningful audit a printed poll book is essential and the ballots must be counted by hand. Obviously an audit of DREs is meaningless and simply running paper ballots through an optical scanner a second time proves nothing.
Halderman, Shacham, Rescorla, and Wagner offer more extensive suggestions in You Go to Elections with the Voting System You Have: Stop-Gap Mitigations for Deployed Voting Systems. As Halderman et al. put it: "In light of the systemic vulnerabilities uncovered by recent reviews of deployed e-voting systems, the surest way to secure the voting process would be to scrap the existing systems and design new ones." As they point out, new systems will take years to design, develop, and test and many billions more of taxpayer dollars and in all probability the next generation won't be any more transparent or reliable than what we have now given New York's experience with Sequoia.
Why not just junk all of the electronic "vote counting" machines, DREs and optical scanners, and count the ballots by hand? There is no need, or even possibility with e-voting machines, that election results will be available Election Night. A longer wait is worth the accuracy.
It should go without saying that referential integrity must be implemented and maintained on the statewide voter registration database and security put in place that works. The database design and integrity must also be subject to public review by qualified citizens.
When something is broken and can't be fixed it is a common ploy of children and bureaucrats to attempt to hide it. With regard to electronic voting, mail ballots are just such a childish tactic, fertile ground on which election fraud will grow.
Dr. Corry is a Fellow of the Geological Society of America and an internationally-known research scientist whose biography has appeared in Who's Who in the World, Who's Who in America, Who's Who in Science and Engineering, among others, for more than ten years.
After service with 1 st Marines he became involved with the early space program in 1960, doing preflight testing and failure analysis on Atlas and Centaur missiles, including all the Project Mercury birds. In 1965 he switched to oceanography and did research at both Scripps Institution in San Diego and Woods Hole Oceanographic on Cape Cod. He has also taught geophysics at university and worked as a research manager for a Fortune 500 company.
Dr. Corry became interested in the problems with electronic voting machines in 2000 while a member of the board of the local chapter of the ACLU. In 2001 colleagues from Schreiver Air Force Base suggested he join the Institute of Electrical and Electronic Engineers (IEEE) Voting Equipment Standards committee and he has served on that since. By 2002 his interest in election problems had expanded to issues with mail ballots. Working with Al Kolwicz and others, he helped defeat Amendment 28 that would have mandated all Colorado elections be done by mail balloting that year.
He has climbed high mountains, been shipwrecked and marooned on an unexplored desert island, ridden horseback through Utah, Arizona, and Colorado, among other adventures during his career.