December 26, 2004
Garbage In, Garbage Out

One of the (many) reasons to set aside our tainted gubernatorial race is because King Ukraine County's voter information database is full of garbage. Leaving aside the thousands of questionable registrations that we'll continue to document, the database simply has so many errors in it that it would be folly to trust the county's data processing capabilities to give us the correct answer when the tolerance for error is as tight as it is in this race.

The following example is not by itself a show-stopping smoking gun, but it is illustrative of the mess to which we would entrust the determination of who gets to run the state for the next four years.

By state law, the voter information that every county must maintain and publish shall include a list of the most recent elections in which every voter has voted.

I wrote a script to process the June 2004 KC voter file to tabulate how many currently registered King County voters had participated in each of the last elections. Among the other figures, the report included the following:

Election
Voters
November 2004 3
November 2005 8
February 2006 43
November 2006 2
November 2007 181
January 2010 8
September 2011 1
November 2012 1
March 2015 2
June 2017 3
September 2017 1
September 2019 1
February 2029 1
September 2030 2
September 2055 1
Total Future Elections 258
Some of these elections, such as November 2007, include several people who are scheduled to vote twice!

[joke: The November 2004 is obviously wrong, but the other numbers might in fact be prescient. If these guys keep screwing up as badly as they have been doing, why should anybody bother to vote in the future?]

But in all seriousness, if King County's data processors can believe that 258 people have already voted in elections that are months and years into the future, why should we trust them to determine within 130 the number of people who voted last month? Posted by Stefan Sharkansky at December 26, 2004 11:50 AM | Email This

Comments
1. why can't you just get over yourself? Gregoire won on the state-mandated recounts and Rossi lost. Winner takes all.

Posted by: Manco_Dollars on December 26, 2004 12:55 PM
2. Yet another example of why no thinking person can trust the outcome of this election.

The reason King County cannot tell us what is going on with the votes and voter registration is that it does not involve the collection of fees, taxes, and revenue. Period. If it involves money, the county is all over it. No money, no activity.

Opinions like the one expressed above only draw more middle-of-the-road people and independents over to the side of questioning the results.

The GOP will prove incompetence, if not outright fraud, on so many levels that any court would be responsible to have the election nullified. At that point Gregoire will have claimed 'she didn't know' (like her sorority, the lawsuits, etc) that nobody but the left wingnuts will trust her.

Posted by: Larry on December 26, 2004 01:02 PM
3. Larry:

LOL & LMAOOOOOOOOOOOOOOOOOOOOO

at your sheer arrogance. WA is solidly Dem and will remain so, you Repugs better just move somewhere else already, hah hah

Posted by: Manco_Dollars on December 26, 2004 01:11 PM
4. We need to get the original data sources for each election -- especially November 2004. For polling voters, there is the sign-in list of voters, where the voter signs the pre-printed list, and also the sequential list, where the election worker writes down the name of each voter.

What original data exists to keep track of absentee voters? Do they keep all the signed ballot envelopes for a specified period of time? Do they sort these, say alpabetically, or by precinct, before doing the data entry.

The party shouldn't wait for data entry, which might be months from now. Nor should data entry be relied upon, since it is obviously inaccurate. The original data should be obtained through public disclosure immediately, and then added up by precinct and category.

So how many people voted (in past elections, that is) after the date that social security shows them as having died?

Posted by: Richard Pope on December 26, 2004 01:19 PM
5. Manco_$,

Keep talking, please! I'm serious, it helps our cause!

Posted by: Larry on December 26, 2004 01:27 PM
6. Without the source data, and the script you used to arrive at these numbers, they are meaningless.
Nothing more than your say-so, which no thinking person would put a lot of stock in.

Posted by: John on December 26, 2004 01:32 PM
7. John,

Who are you? If you don't like the analysis, why don't you go back to the Democratic Undergound?

Stefan is notifying us of possible serious issues in King County's voter registration database. I work in IT and if data like this were found we would move to immediately correct it. In many instances this might entail contacting our customers directly.

When can we expect King County's database to conform to state law and actually reflect reality? I, for one, would like to know about the original voter registration forms of these people, the ballots in Alaska, Precinct 1823, etc.

King County voter data is publicly-available information. If you want the source data, go find it! If not, just walk away, okay?

Posted by: Larry on December 26, 2004 01:39 PM
8. I am a King County voter.
I don't go to democratic underground, but do find your knee-jerkism amusing.

I work in IT as well, in databases in fact. You might think that Stefan is notifiying you of "serious" issues, but you WANT there to be "serious" issues, therefore you aren't sufficiently skeptical of the evidence.

While I can go and find King county(or any other county, but I guess only democratic counties are cause for concern), but without Stefan's script, which he used to arrive at these figures, it is meaningless in confirming these firgures. If you really work in IT, Larry, I would expect you to be aware that it the middle tier(e.g. the script)that is at fault when bad data is reached more often than it is the actual data.

Now Stefan can either present his code, or expect everyone to just take his word for it. Whichever way he chooses will speak volumes about his veracity.

Posted by: John on December 26, 2004 03:11 PM
9. "John", for someone who pretends to be highly concerned about transparency, you don't even give your real name and email address.

You're obviously nothing more than a troll. I could post the script, but what would be the point? The script is not the issue, the data is the issue. A CD with the data is available to the general public for $3.40. Anybody who questions my veracity can get themselves a copy of the data and check it for themselves.

Posted by: Stefan Sharkansky on December 26, 2004 03:22 PM
10. "...if King County's data processors can believe that 258 people have already voted in elections that are months and years into the future, why should we trust them to determine within 130 the number of people who voted last month?"

Uh, because they did a careful manual recount?

Posted by: Bruce on December 26, 2004 03:31 PM
11. CAREFUL MANUAL RECOUNT? LOL

Posted by: Side Splitter on December 26, 2004 03:50 PM
12. Quote from John:

"If you really work in IT, Larry, I would expect you to be aware that it the middle tier(e.g. the script)that is at fault when bad data is reached more often than it is the actual data."

This must be a quote from the well-respected IT Journal 'DUH!' Yes, John, I am aware of what you call the middle-tier - I write scripts for a living.

I would expect that if you work in IT, which you say you do (with databases in fact!) that you could WRITE YOUR OWN DAMN SCRIPT AND PROCESS THE DATA ON YOUR HOME PC!! Why do you need Stefan's script - do you want to re-process his results? Wouldn't you want to take the source data and write your own program to independently validate Stefan's data?

Sheesh, get a life.

Posted by: Larry on December 26, 2004 03:57 PM
13. Stefan:
This is my real name, I no longer use my real email when posting comments on right-wing blogs. I still get threatening emails from the last time I did.
The script is what connects the raw data to your figures. It IS relevant, if anyone were to try and fact-check your figures.

Larry:
Yes, I could write my own script. I could write a script to darw any end results you wish from the data. That would be exactly my point. Perhaps you should looking reading that journal you spoke of. As for my wanting and the source data, go back and read my original post (try doing it very s l o w l y, you might just understand it this time)

Posted by: John on December 26, 2004 04:23 PM
14. Obviously, the "source data" that John wants is not the King County voter registration file, since it would contain precisely the same errors that Stefan has set forth. Instead, by "source data", John wants the original voting records. What he wants are the polling place registers and absentee ballot envelopes from the future elections, so he can determine whether these 258 or so people really voted in these elections. Very funny, John.

Posted by: Richard Pope on December 26, 2004 06:44 PM
15. I suspect you would find a similar data entry error rate in any county in the state that maintains a computerized data base. Unless you can show King County's database is actually significantly worse than others, your statistical analysis is of little significance.

There is an old saying: "There are lies, damn lies, and statistics." Meaning that if you slant your analysis a certain way, you can make the statistics prove anything you want. I have noticed your analyses frequently take advantage of that fact. Whether by accident or design I can't say.

It is definitely true that our system works perfectly well until the margin between the candidates is smaller than the margin of error of the system for measuring votes. The question is how do we handle this. In our state, a manual recount is the final mechanism. Whether that is the most accurate means is beside the point; that's what state law says. What we don't do is have another election.

I find these calls for a re-vote fascinating. I didn't hear Republicans suggesting it when Rossi was up by 42, which appears to be well below the margin of error for a machine count. Now its a different story with Gregoire up by 130. I would have liked a re-vote in Florida in 2000. Oh well, we can't always get what we want.

Posted by: Wayne on December 27, 2004 12:30 AM
16. The question is how do we handle this. In our state, a manual recount is the final mechanism. Whether that is the most accurate means is beside the point; that's what state law says.

Sorry to burst your bubble, Wayne. A manual recount is not the "final" mechanism. State law also allows any registered voter to contest an election, which seems to be where this is headed.

I gather you aren't going to like this. Oh well, we can't always get what we want.

Posted by: Stefan Sharkansky on December 27, 2004 12:59 AM
17. Actually, I am unconcerned, as the contest will inevitably fail.

Posted by: Wayne on December 27, 2004 08:57 AM
18. If for nothing else, the flaws in the system need to be pointed out now, because you know and thinking Washingtonians know, if this result goes unfettered, with the Democrats in power (who like the system broken as it is - as it has always been in recent history) will be resistant against doing anything to remedy this for future elections.

Contesting this election is the right thing to do, not only to attempt to get the correct result (as the Ukraine did yesterday), but to at least get the revamping of this broken delapidated election system kick started for the entire state -especially King County.

Posted by: KS on December 27, 2004 09:37 AM
19. If it were simply a "recount", then much of the argument would be moot. However, this has turned into a recanvass. It is not the GOP's fault that KC kept finding extra ballots (what, 9 times??), due to incompetency. Regardless of the final results, it would appear that some heads should roll. The appearance of impropriety must be avoided or future elections will drag us through this over and over again. I compare this to the Superbowl, where the refs made a bad call, that lost the game for the opposing team. Tough luck, but we don't get a replay, after the whistle blows.

Posted by: Morgan on December 27, 2004 12:39 PM
20. Hi
I dont know anything about database. What I do know is if several hundred people are registered at 500 4th ave something is fishy. I also know that if something is getting done about this fraud it is in no small part duuuua to your hard work. The orbusmax faithfull thank you.
Dan Rheaume

Posted by: Dan Rheaume on December 27, 2004 03:59 PM
Post a comment
Name:


Email Address:


URL:


Comments:


Remember info?