AK: Letter to Division of Elections & State rebuffs raw vote demand

[Wilms]

Alaska: Letter to Division of Elections

By Jake Metcalfe, Chair, Alaska Democratic Party

January 24, 2006

"It is wrong that Diebold is trying to take possession of our votes and our public data by claiming that these are their proprietary information."

The following letter was sent by email to Whitney Brewster, chief of staff for Alaska Lt. Governor Loren Leman, the state's chief election official.

snip

The export of information in a common file format is not adequate and will not allow us to verify the state's posted elections results, which contain numerous errors and discrepancies. It has been documented that the GEMS software was programmed to contain a "double set of books" within it, and thus a print out from one file selected by Diebold will not enable a comprehensive understanding of what has caused the bizarre and inaccurate reports produced by the Diebold system for the state's 2004 election reports.

snip

A second issue addressed in your letter of Jan. 19 deals with our request for a copy of the electronic voter file as it existed immediately following the entry of data from the 2004 General Election. You maintain that you are not able to provide this electronic file because it no longer exists and cannot be recovered. We have acquired a copy of this file that was purchased from the State as it existed on Jan. 11, 2005. We have compared the numbers of voters who are coded as voting in the 2004 general election with the official election totals supplied by the Division (see email from Tom Godkin and attached spreadsheet dated 1/4/05). We have found a large discrepancy regarding the absentee votes. Godkin's spreadsheet shows that there were 60,296 absentee votes cast, yet the filed dated 1/11/05 shows that 63,371 voters cast absentee ballots. Of those 63,371 voters who were recorded as having cast an absentee ballot, 1,058 registered after the deadline to register and be eligible to vote in that election (Oct. 3, 2004). From the data we have it appears that 2,020 valid absentee votes were cast but not counted in the official totals. These discrepancies further support our need to examine the central data tabulator file.

A third issue addressed in your Jan. 19 letter is our request for the total number of votes cast for each candidate in each House District in the US Senate race in 2004. You state that "this information is available on the Division of Election's web site at www.elections.state.ak.us under 2004 General Election Results." This statement is false. It is impossible to tell from the publicly posted data how many votes the U.S. Senate candidates received in each House district.

snip

We hope that you will immediately provide the public records we have requested as required by law. To summarize, please provide the following public records as soon as possible:

1. A copy of the "central tabulator data file" taken from the Diebold-supplied computer used to run the "GEMS" (Global Election Management Software) application as that filed existed immediately following the official canvas.

This electronic data file will end in the extension "MDB" (Microsoft DataBase) or "GBF" (GEMS Backup File, a compressed version of MDB). This should be the version of the file containing all final vote tallies for the 2004 General Election. It will be located in the directory: C:\PROGRAM FILES\GEMS\LOCALDB

2. The total number of votes cast for each candidate in each House District in the U.S. Senate race in 2004.

3. The number of questioned, absentee and early ballots that were cast for each candidate in each House district in the 2004 General Election.

http://www.votetrustusa.org/index.php?option=com_content&task=view&id=830&Itemid=113





State rebuffs raw vote demand

STANDOFF: Democrats want 2004 base election data; machine firm is playing coy.

By LISA DEMER

Anchorage Daily News

Published: January 24, 2006

snip

Documents provided by the Democrats show that Brewster contacted Diebold and was told the public data can be released only after being transferred to a common format such as Microsoft Excel.

In a Jan. 6 e-mail, Diebold's lawyer, Charles R. Owen, told Brewster that "the structure of the database file ... is proprietary information."

Perhaps, but it's not secret. Anyone can examine Diebold's format on a Web site set up by activists who have been raising questions about the company, the Alaska Democrats said.

"Copies of these kinds of files have been sitting on the Internet for over two years, with Diebold's knowledge," said Jim March, an investigator with Black Box Voting, a private organization that calls itself a national consumer protection group for voters.

snip

http://www.adn.com/news/alaska/story/7386582p-7298824c.html


Plus the Dem Party Press Release

http://www.sitnews.us/0106news/012406/012406_diebold.html

[Fredda Weinberg]

1. Here we got again ...

Ah, there's the explanation for the presidential discrepancy

Early votes for statewide candidates were not recorded by House district but rather were tallied for each of the state's four election regions. Those regional totals then were reported for every House district, essentially inflating the vote total many times over.

http://www.adn.com/news/alaska/story/7386582p-7298824c.html

[Fredda Weinberg]

2. and Jim March's file archive ...

doesn't make Diebold's database schema public domain anymore than the copies they've sold.

[Wilms]

3. That was already refuted in the letter referrenced in the OP.

What's your deal. :shrug:

snip

A second issue addressed in your letter of Jan. 19 deals with our request for a copy of the electronic voter file as it existed immediately following the entry of data from the 2004 General Election. You maintain that you are not able to provide this electronic file because it no longer exists and cannot be recovered. We have acquired a copy of this file that was purchased from the State as it existed on Jan. 11, 2005. We have compared the numbers of voters who are coded as voting in the 2004 general election with the official election totals supplied by the Division (see email from Tom Godkin and attached spreadsheet dated 1/4/05). We have found a large discrepancy regarding the absentee votes. Godkin's spreadsheet shows that there were 60,296 absentee votes cast, yet the filed dated 1/11/05 shows that 63,371 voters cast absentee ballots. Of those 63,371 voters who were recorded as having cast an absentee ballot, 1,058 registered after the deadline to register and be eligible to vote in that election (Oct. 3, 2004). From the data we have it appears that 2,020 valid absentee votes were cast but not counted in the official totals. These discrepancies further support our need to examine the central data tabulator file.

A third issue addressed in your Jan. 19 letter is our request for the total number of votes cast for each candidate in each House District in the US Senate race in 2004. You state that "this information is available on the Division of Election's web site at www.elections.state.ak.us under 2004 General Election Results." This statement is false. It is impossible to tell from the publicly posted data how many votes the U.S. Senate candidates received in each House district.

You further state that the number of questioned, absentee and early ballots that were cast for each candidate in each House district in the 2004 General Election "are listed within the Statement of Votes Cast report located on the Division's web site." Contrary to what your statement implies, the number of questioned, absentee and early ballots that were cast for each candidate in each House district in the 2004 General Election cannot be determined from the division's posted election reports, and that is why we are seeking the database of public records discussed above.

If the Division is unable to break down the results of early voting by House District, as you state in the Jan. 19 letter, how are you able to break those votes down for the State House races? If early votes cannot be broken down by House District, then the totals reported for the State House races are questionable. Further, you have not addressed the odd, multiple-district groupings of absentee and questioned ballots that appear throughout the District-by-District reports - are these grouped absentee and questioned ballots also impossible to break down by House District?

snip

http://www.votetrustusa.org/index.php?option=com_content&task=view&id=830&Itemid=113

[Fredda Weinberg]

4. My deal is common sense

and your refutation doesn't address the explanation - which, like the problem they were having with the dual ballots, gives an innocent reason for the raw numbers coming out different from the final results.

I deal with data for a living - if you don't, you may not see how foolish the refusal to accept an extract is. As I pointed out previously, I'm not the only professional who noticed this.

[Wilms]

5. That was already refuted in the letter referrenced in the OP.

What's your deal. :shrug:

snip

A second issue addressed in your letter of Jan. 19 deals with our request for a copy of the electronic voter file as it existed immediately following the entry of data from the 2004 General Election. You maintain that you are not able to provide this electronic file because it no longer exists and cannot be recovered. We have acquired a copy of this file that was purchased from the State as it existed on Jan. 11, 2005. We have compared the numbers of voters who are coded as voting in the 2004 general election with the official election totals supplied by the Division (see email from Tom Godkin and attached spreadsheet dated 1/4/05). We have found a large discrepancy regarding the absentee votes. Godkin's spreadsheet shows that there were 60,296 absentee votes cast, yet the filed dated 1/11/05 shows that 63,371 voters cast absentee ballots. Of those 63,371 voters who were recorded as having cast an absentee ballot, 1,058 registered after the deadline to register and be eligible to vote in that election (Oct. 3, 2004). From the data we have it appears that 2,020 valid absentee votes were cast but not counted in the official totals. These discrepancies further support our need to examine the central data tabulator file.

A third issue addressed in your Jan. 19 letter is our request for the total number of votes cast for each candidate in each House District in the US Senate race in 2004. You state that "this information is available on the Division of Election's web site at www.elections.state.ak.us under 2004 General Election Results." This statement is false. It is impossible to tell from the publicly posted data how many votes the U.S. Senate candidates received in each House district.

You further state that the number of questioned, absentee and early ballots that were cast for each candidate in each House district in the 2004 General Election "are listed within the Statement of Votes Cast report located on the Division's web site." Contrary to what your statement implies, the number of questioned, absentee and early ballots that were cast for each candidate in each House district in the 2004 General Election cannot be determined from the division's posted election reports, and that is why we are seeking the database of public records discussed above.

If the Division is unable to break down the results of early voting by House District, as you state in the Jan. 19 letter, how are you able to break those votes down for the State House races? If early votes cannot be broken down by House District, then the totals reported for the State House races are questionable. Further, you have not addressed the odd, multiple-district groupings of absentee and questioned ballots that appear throughout the District-by-District reports - are these grouped absentee and questioned ballots also impossible to break down by House District?

snip

http://www.votetrustusa.org/index.php?option=com_content&task=view&id=830&Itemid=113

[Fredda Weinberg]

6. How does repeating a non-sequitor make it relevant?

The rebuttal doesn't address the explanation. If the early returns were double counted as were the dual ballots, that's a good reason for the final tallies to be different. As I noted in the other thread, the correction in the presidential race was consistent with the recounted ballots in the senatorial contest ... if this was a conspiracy to defraud, it would have to a lot more complicated than the Alaska Democratic Party is claiming.

[Wilms]

7. You seem intent on not seeing the problem. So I'm done. n/t

[snowbear]

10. Fredda...

nevermind.. (deleted)

[izzybeans]

11. If the issue is that there is a discrepency between raw data

and final tally then if you really deal with data you might suggest looking at the raw data file...which is what is being asked here. Naive assertions aside...this is all that has been asked. A statistician will want to see the raw data in order to track back the data transformations that seem in error...It's a simple task...but it most often involves reading the code to see what in fact has been done.

There are easy solutions to the proprietary problem and diebold is hedging their bets. Why doesn't Deibold the republicans and democrats meet in a big room with their own database programmers and figure out the discrepency (a database schema I highly doubt is unique and unknown in the public domain). Everything else is conjecture (yours and the dems) without seeing what happened between the raw data and the final tally. There is only one reason I can think of as to why that might be a problem and it has nothing to do with proprietary information. But then again that is mere conjecture because Diebold nor the state give any reason to trust them...you know reciprocity.

You could be correct, someone else might be, and then there are the unknown possibilities and there is no way to figure it out until a fairly mundane task is performed. Some people call it data forensics. I call it checking your work.

[Name removed]

12. Deleted message

Message removed by moderator. Click here to review the message board rules.

[OnTheOtherHand]

13. ???

This looks like a change of position to me. Yesterday you wrote,

All one would have to do is export the data to another db type like Access, Oracle, compare it with the numbers in the State's system, and you have a valid analysis. If the numbers do not agree, then you perform an RCA to see why.

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=410440&mesg_id=410548

And now you have concluded, I guess, that an RCA is necessary whether or not the numbers add up. Ergo, the other poster must be "a fraud, a diebold plant, or an idiot." Umm, yeah, whatever.

I dunno. I think the most likely outcome is that whatever may have gone wrong in Alaska, inspecting the .mdb will reveal no more than inspecting the extract. And so far I haven't seen a cogent contrary argument here based on knowledge of the GEMS design, or of Alaska election returns. But DU is not where I come for cutting-edge DB analysis (alright, I don't go anywhere for that -- too busy doing other things these days). For us to sit around and debate what the Alaska Dem Party may find out if they ever get access to the .mdb seems a lot like sports talk radio.

[riqster]

14. My bad

I should have explained myself further. The RCA would of necessity involve the entire application involved, and would occur after a thorough examination of the export file.

What changed my mind in re the necessity of getting into GEMS was the fact that there are multiple copies of the datafile in GEMS, and that the file one received may or may not be sufficient for validation.

As to my rudeness-a good data analyst would NEVER accept a plausible explanation as fact. The first four letters in 'analyst' are, after all.....

[OnTheOtherHand]

15. OK, that focus on seeing both copies seems plausible

As for the rest -- I think it depends on context. Fredda is certainly expressing more confidence in the outcome than I could muster, but I'm sympathetic because I've been through so many weird arguments about the exit polls, and even a few about Alaska. The Official Story actually seems entirely plausible, and more coherent than any alternative hypothesis I have seen so far -- and it's perfectly possible that the application is FUBAR as well. Or it is already known, depending on one's criteria.

[riqster]

16. Either way, there is no substitute for a complete analysis

And no reason to oppose one, unless one has something to hide.

[Fredda Weinberg]

17. You don't seem to understand proprietary design

Whether I deliver my product using Access, SQL Server or Oracle as a database, the design is my work-product. Look up "proprietary schema" if you need a million reasons to demonstrate you're wrong.

I use my real name - go look me up and try "fredda weinberg" "center for court innovation" if you need one reference. Try [e-mail] if you want more bona fides.

Maybe you should be more humble. I've reported your post to the mods and hope they enforce the board's rules.

[riqster]

18. Let me explain

Data is only part of a database. The schema is an important element, and so is the user interface, and all of the other applications/devices that touch the database. In this case, the data in question had passed through at least one application and thus, if the numbers do not add up, the entire system must be thoroughly analyzed to provide validation.

When one recieves an export file, one receives ONLY THE DATA. That is tantamount to getting the gas tank when one needs to examine an automobile. In this case, to extend the analogy, we have a dual-tank vehicle and would not even know whether we got the one that was being used during a given period of operation.

The idea that one can do a complete analysis with incomplete data is absurd, and anyone who flaunts their alleged technical superiority while exposing their own ignorance should expect a bit of pushback. If I were to go to my employer and say, "We don't need to do a full test, because the vendor says 'trust me'", I'd be unemployed in a big hurry-and rightly so. There is never, and has never been, and will never be, a substitute for rigor in the digital world. Anyone who says otherwsie is exposing ignorance, an agenda, or both.

Now; you are correct in saying that Diebold does have an intellectual property 'dog in the fight', and I accept your correction in that regard. I do not feel that it trumps the public good, however, and when I have written software, I was expected to provide full documentation and validation to the customer.

[Name removed]

19. Deleted message

Message removed by moderator. Click here to review the message board rules.

[riqster]

20. Bluster doesn't impress me

All I said was that there is no substitute for complete end-to-end testing, and that there is no good reason for not doing so. Playing dominance games is for children, and does not refute my assertation.

There is no point in reasoning with the unreasonable, so I'll let my posts inform those who have the ablility to listen. Adieu.

[phoebe]

9. see you're back and up to your old habits..

n/t

[kster]

8. Kick-n-Recommended...nt