Our Shero, Karen Richards
You Hang In There, You Hear?
By Greg Moses
With the power of a new voter database coming online next year, will Texas be able to delete tens of thousands of voters with a few keystrokes of political will? Not if Karen Richards gets her way.
As director of voter registration in Texas, Richards has been asked to evaluate thousands of “business functions” that will be part of the new voter registration system, and her comments read like the barks of a guard dog every time the technology gets close to enabling an automated purge.
The frequency of her warnings against automated purges suggests that worries about the technology itself are not unfounded. Why would she have to repeat herself so often if the technology were not ready to swoop through voter lists? Considering the power of these databases nationwide, the Richards principles would make a fine template for federal law.
Time after time, Richards repeats her twin principles: (1) “only a user may explicitly cancel a voter in Texas” and (2) “events should not automatically change a voter’s status.”
Richards is meticulous in her logic. The eRegistry system made by election purveyor Hart InterCivic will match voter records against records of felons and deceased. But even if a felon record matches a voter, Richards argues that there can be no “certainty” of a match, because the possibility remains that identity theft may be involved. Says Richards, the computer may make all the matches it wants to “as long as flagging does not mean actually canceling the voter record.”
Or take the case of a registered voter who is contacted for jury duty and marks as an excuse for not showing up that she or he is not a citizen. eReigstry will see to it that the information gets tracked back to the voter database where the voter will be marked non-citizen and ineligible to vote. Say Richards once again: there should be no automatic deleting of voters for any reason.
If the database flags a record on the basis of some internal information flow, someone should look things over. But the Richards principles are clear and consistent. Do not have the computer delete any voter for any reason by automatic means. Or in language that we might call Richards rule number three: when it comes to correlating any information that would affect a voter’s status, the computer “should not attach anything more than an event to investigate.”
And Richards has commitments beyond the purge question. When a new voter is added says Richards, “a new voter card should automatically be created.” Sometimes comments from software vendor Hart InterCivic seem strangely nonresponsive. For example, when Richards insists that voter registrations should be created “automatically,” the Hart rep assures her that the user will have an “opportunity” to create such a card. But an “opportunity” is not “automatic.”
Or when Hart suggests that a “business function” will allow a voter to be registered and assigned a precinct without a complete or verifiable address, Richards says the function should be disabled. “Leaves room for error for voters not to receive certificates or appear on voter list,” writes Richards. Richards to Hart: disable that event.
With all these “business functions” to read and think about, we are beginning to understand the significance of March complaints from state staff that they felt overwhelmed by the flow of paperwork coming at them from Hart. These are not the kinds of evaluations that should go rushed.
Some of the technical questions about which ID numbers will be used are resolved in the comments between Hart and state election officials. Texas has a distinct preference for drivers license numbers as an indexing scheme, and computer tests showed that the TDL numbers were more efficient than SSN for searching voter records
Also, our hypothesis about the real value of the voter lists as campaign and party contact tools gets a shot from eRegistry “business functions” that will accept home phone numbers and email addresses. As Richards points out, the phone number collection should be disabled since Texas law prevents counties from taking the information. But the Hart response is telling: We can disable it until such time as Texas wants it back, as other states have found these contact data helpful. No doubt.
On the possibility of name prefixes we had to laugh out loud. Name prefixes can be enabled by the Hart system, but Richards sees no good reason for them. Perhaps, maybe a Mr. or Ms. could be used for printing letters, to which the Hart rep replies in writing, okay we’ll consider enabling Mr. and Mrs. The first time this happens, Hart gets off the hook, but the second time it happens, Richards writes, “why is the Hart answer talking about Mr/Mrs references?” LOL! Ms. Richards, you hang in there, you hear!
The many detailed world
of the Texas voter database project
Richards’ evaluation of eRegistry business functions have been carefully documented and preserved as part of User Acceptance Criteria or UAC that were completed during late March. In the meticulous tables of functions and comments found in a four-inch stack of UAC documents, the reader gets a sense of the detail demanded by Bob Futrell, the state’s project manager for the Texas Election Administration Management system or TEAM project.
For instance we know that the database to be delivered from IBM to Texas early next year will scan the 88,000 records of Hays County in .5 seconds to check for a duplicate drivers license number before committing a new voter to the database; that voter registration certificates for mass mailing will be created at a rate ot 83.8 per second; and that it will take ten minutes to create an official voting list for a county of 1.6 million voters.
These are just a few of the details that Futrell required IBM to find out about the existing Texas database so that when the lead contractor delivers the new system next year it will be perform at least as well as the existing Texas Voter Registration System (TVRS).
And these details in turn are only the top few sheets of paper that begin to outline all the “business functions” that the state will expect before it says yes to the computerized system now being developed by IBM and Hart InterCivic. In official project terms the UAC doc is marked as Deliverable 32 (D.32) and was turned in at the end of March.
The paperwork on these computer events still has to go through four more transformations as “business functions” listed in D.32 get refined into “how to requirements” or Detailed Requirement Specs (D.10) each of which will be assigned a unique ID or URID in the Requirements Traceability Matrix (D.11) before being committed to a User Acceptance Test Plan (D.19) that will finally be checked off as “passed” in what might be called the mother of all report cards, the Acceptance Test Results (D.21).
So if you have any question at all about what this new system will or won’t do, all you have to do is get yourself crosseyed over this stack of docs known as the UAC. In the UAC we find a collection of spreadsheets, the left columns of which have rows of business functions followed by columns of comments on each business function made by a few of the players in the database project, most notably the head of voter registration for Texas, Karen Richards.
* * *
Note: this is part of a series of articles by the Texas Civil Rights Review about a statewide voter database and election management system being developed by Texas to meet federal mandates under the Help American Vote Act of 2002 (HAVA).
As readers of this series already know, we cannot pass over the
opportunity to point out that the state has a pretty good voter registration system in place (TVRS or Texas Voter Registration System). The website for the Secretary of State says over 120 counties are online with the system, but a source at the office told us all but 90 (of 254 total) counties will be hooked up live to the new database when it comes online next year. Why the old system could not be upgraded to meet federal standards is a question we like to keep open.