January 10, 2006
Update on statewide voter database

Assistant Secretary of State Steve Excell called me this morning with an update on the statewide voter database and to explain why they're not releasing a copy until February --

His first explanation was that the database contains exempt information (Social Security and driver's license numbers) that would have to be redacted. Because it would be a special effort to redact this data, he claimed, the Public Disclosure Act doesn't require them to interrupt their other work to produce a copy of the database.

When I pointed out that it only takes a few moments to remove columns using, say, Microsoft Access, he gave the completely different reason that the 39 counties are all entering changes to the database and that they can't run an export while there are transactions. Which is probably true, except that exporting a table doesn't take all that long and it's not like the county election offices are entering transactions 24x7.

It's apparent that they have their reasons for not releasing any copies of the database until February, but that the stated reasons are not terribly compelling. It's left as an exercise to the reader to read between the lines.

I also asked Excell about the possibility of running the statewide database against the federal alien database. He said that Sam Reed has been trying to get access to the data for sometime, but that the feds have been giving him the run around and that the technology of the federal database is poor -- the federal database is antiquated and not set up to actually run a cross check against the state file and that prior experience has revealed so many errors that the data is largely worthless. Given everything else we know about the INS (now CIS) and Homeland Security agencies, I'm predisposed to believe this.

Posted by Stefan Sharkansky at January 10, 2006 12:05 PM | Email This
Comments
1. It's time to fire up the letters to the editor over Sam Reed and his failures. The counties are probably all trying to police up their garbage registrations before they transmit and get busted out in the press for their poor management.

Posted by: sgmmac on January 10, 2006 12:31 PM
2. That guy is a LIAR! If he wants the SQL to remove that stuff, it is something any secretary could do.


Posted by: pbj on January 10, 2006 01:06 PM
3. Sounds like they tried to dodge and you called them on it. The guy tried to mumble his way out of it but he didn't make much headway. Good stuff here...

Posted by: Interested Observer on January 10, 2006 01:19 PM
4. And cross referencing ANY data from ANY source can be done. Somemore easily than others, but there is always a way. Anyone whomakes up some mumbo jumbo excuse is insulting your intelligence. I do databases and this stuff every day. It isn't rocket science.

Posted by: pbj on January 10, 2006 01:27 PM
5. Well, for me it'd be rocket science. But that's why God created folks like pbj.;)

Posted by: Danny on January 10, 2006 01:36 PM
6. Stefan

Is the lapsed Jan-1-2005 release date required by law or is it a goal that is not solid.

Mr. Excell probably does not understand databases and should have had one of his computer experts participate in the phone call to answer your technical questions.

Unfortuneately, my "gut feel" is there's "a blow in the numbers" and unless Mr. Reed's delay in releasing the data is headlined daily, the database used for the Nov-2006 will be broken and of no value whatsoever.

Posted by: Green Lake Mark on January 10, 2006 02:46 PM
7. If they are using any real database engine, you can dump at any time without requiring a DB wide lock. If a database can't do a dump while the database active then it is impossible to do point in time backups -- i.e. you don't have regular and reliable backups. So either he is wrong or they are using a database that doesn't meet state requirements for DRBR. PostgreSQL (an open source database engine) has supported this for *YEARS*.

Posted by: Joby on January 10, 2006 05:11 PM
8. Stefan,

You are assuming that the state invested in some "current" technology. I will bet that they are on some IBM 360 legacy system and are trying to unearth a living cobol programmer.

Posted by: Fed Up on January 10, 2006 06:36 PM
9. It would be interesting to know whether the low bidder for the contract is old enough to vote yet...

Posted by: gaelwolf on January 10, 2006 07:16 PM
10. It would be interesting to know whether the low bidder for the contract is old enough to vote yet...

No one is too young to vote in King County...or too dead, or too gone...

Posted by: South County on January 10, 2006 07:25 PM
11. The question is; Is this better than what there was before ? The local radio news claimed that it was because it consolidated 39 separate databases into one. However, if there has not been adequate QA/QC on the data - then it is no better and the same old garbage in/garbage out.

Stefan is probably waiting to do the QA/QC on this voter database. Since they already bluffed/lied about when this would be done, it would naturally make us suspicious about the quality of this database. If it is poor quality of negligible improvement, then the Elections corruption has clearly spread through the King County pipeline to the Secy of State's office - and it is past time to remove Sam Reed from office - and he can take Excel and Handy with him. So the jury remains partly out -whether this voter database is worthy or if it is a piece of crap !

Posted by: KS on January 10, 2006 09:03 PM
12. Again, there is some level of automation that can be applied to cleaning up legacy data. I know, I have done it and even written articles on it for magazines. Clearly this fellow doesn't know anything about databases and is assuming most other people don't either. Unfortunately for him, Stefan has a CS degree just like me (except his is from a real university LOL!).

I could teach anyone here how to query the database, filter out columns, even remove faulty data in about one day.

Posted by: pbj on January 10, 2006 10:12 PM
13. The data from all 39 counties voter registration databases was loaded into the new Statewide Voter Registration Database in December.

The new Statewide VRDB was operational on January 01, 2006. All counties have been registering new voters since then and adding them to the VRDB.

This consolidation meets the deadline of the Help America Vote Act of 2002 that such a database be in place on Jan 01, 2006.

HAVA and state law also requires periodic list maintenance for dupes, dead persons and felons. The Statewide VRDB is connected to the State Patrol database, the Administrator of the Courts and we maintain a list of the Governor's pardons -- all necessary to identify potential felon voters.

The Statewide VRDB is also connected to the Dept of Health Vital Statistics database to get monthly updates of deceased voters. We also check again the Social Security Master Death Index to find our voters who die out of state (like in a Portland Hospital).

We are in the process of running the first dupe and death check now and will be done before the end of the month. It is not a simple computer dupe check -- state and local election officials actually go over the reports with human eyeballs to make sure mistakes are not made by the computer.

The felon check will run in March because that is the largest window between special election to do that check. We must send a written notice to every potential felons telling them that they have been identified as a potential felon and will be dropped from the voter rolls unless they can show us that their civil rights have been restored.

Stefan is not the only one requesting data. Our office has many requests. As we have done with our old database, we prepare a monthly extract of the non-cofidential data after each dupe/death check and will give everyone a copy of the same CD-ROM. There is a fee for the CD-ROM. The first extract will be available on Feb. 01, 2006.

Anyone who wants to buy a copy of the old database used for Initiative/Referendum signature checking can certainly do (as many did for the election contest).

Most candidates and political parties want the clean list with birth date -- a piece of previously-private data now public for the first time due to a law passed last session (2005).

With a small IT staff we are not set up to "stop the show" and do numerous extracts 24x7. So we provide new copies of the VRDB after maintenance cycle is completed after the previous release.

There is no skull-duggery, conspiracies, chicanery, or black helicopters.

There is a system in place to maintain the list and get everyone the data they need.

Call me or email me if you have questions...

Steve Excell
Assistant Secretary of State
Phone: (360) 902-4155
Email: SExcell@secstate.wa.gov

Posted by: Steve Excell on January 11, 2006 10:29 AM
14. "state and local election officials actually go over the reports with human eyeballs to make sure mistakes are not made by the computer."

Computers do not make mistakes, they only do what the human programmed them to do. Who will be double checking the human eyeballs?

Again, any data crosse checked between two information systems can and should be automated. Me Excell, if your IT staff is telling it you cannot be done, they are not being candid with you or they are sub-par. Only in cases when the checking is done against non-eletronic sources is that the only option.

The City of Seattle has a decent IT department. I suggest you consult with them to help instruct your IT staff on how to automate these processes.


Posted by: pbj on January 12, 2006 05:05 PM
15. Dear Mr. Excell;

As a taxpayer in the State of Washington you are asking me to believe that your department cannot perform the Database Management tasks that THOUSANDS of private companies, municipalities and schools can do. I beleive it is incumbent upon YOU to flag this deficency to Mr. Reed and insist on immediate funding for adequate systems OR relinquish the managment of the Databases to a private firm that can perform to expectations.

Posted by: Taxed to the Max on January 12, 2006 07:49 PM
16. Stefan,

You probably noticed this one already in the Seattle Times:
http://seattletimes.nwsource.com/html/localnews/2002736945_voter13m.html

They need to clean up the duplicate names and deceased "voters" before giving anyone else a copy of the database.

I wonder if they will keep records of the corrections they had to make.

Posted by: Micajah on January 13, 2006 09:40 AM
17. The reason human eyeballs, and not computers, must eliminate dupes is simple: There are many common names with the same birth date and identifying information. For example, "Baby Boomer" John Miller may have many matches in Seattle for the same birth date. It takes an election official to do further research to find who are the same, who are dupes, and who are different.

Posted by: Steve Excell on January 21, 2006 08:01 PM
Post a comment
Name:


Email Address:


URL:


Comments:


Remember info?