Ballot Rotation vs. Vote Switching

[Bill Bored]

Several states require candidates' names to be rotated on the ballots so that no one consistently has the top position. Of course with e-voting systems such as Diebold's GEMS, this functionality is also supported.

But with GEMS, there's some additional functionality: Anyone with access to GEMS can change the order of the names on the ballot just by editing the Ballot Text and SWITCHING THEM! This can be done anytime before the election, so that the names the voters see on the screen are no longer linked to the corresponding names and candidate numbers in the GEMS election database. Whether the names have been changed on the ballot through ballot text editing or not, there is no way for the voter to know if the vote is going to the name selected on the screen or someone else in the database. Therefore, switching the names automatically switches the votes as well, undetected by the voter.

In a system designed to discourage fraud, one would think that once a candidate's name is programmed on a ballot, the software might have a built-in safeguard to make it impossible to edit the names, allowing only ballot rotations and keeping the linkages between the names and the database intact. But instead, anyone with access to GEMS can switch the names on the ballot any time before the election, or in the case of Optically Scanned ballots, any time before the ballots are actually printed, thereby removing any linkage between the names on the ballot and the votes counted in the GEMS database.

Again, this would go entirely unnoticed by the voters. They just look at the screen or the paper and see some names.

Now the question is, who would have such access to GEMS before the election? Since the software can be so easily exploited, what checks or pre-election auditing of the configuration is done to ensure that this does not happen? Is GEMS programmed in a bi- or multi-partisan manner by honest Boards of Election employees with observers, or is this work left to Diebold or some other contractor to perform?

The related issue is the ease with which GEMS access can be had by hacking its passwords, which are displayed in the back end MS Access database. It would be a simple matter for someone with no knowledge of computer programming whatsoever to copy a password, apply it to the election database, make the malicious changes before the election, change the password back to its original value to avoid detection, and sit back and watch the switched votes roll in.

While there has been considerable coverage of the potential for post-election switching of vote totals revealed by Bev Harris, there has been no reporting of the potential for pre-election corruption of the ballot by using the GEMS Ballot Text Editor that I am aware of.

So spread the word about this people! It would be helpful to know if other votin' machine vendors make this functionality available to their technicians and customers too.

Meanwhile, if there any precincts or counties where Bush-Kerry results seem to be reversed compared to polls, previous years, etc., we have another possible method by which this may have been engineered to occur -- before the election even started. It could also be applied to early voting which provided an opportunity to see which way the wind was blowing. If Kerry won the early vote, but lost on election day in the same jurisdiction for example, there could have been a ballot text switch on Nov. 1st or 2nd. You see, in GEMS, Early Voting and Election Day are considered to be two different elections!

Note that this has nothing to do with hacking, programming, malicious code or legitimate ballot order rotations required by law. It's just one of the features in GEMS. It can be applied to both Op Scan and Touch Screen ballots and it is undetectable by the voters.

At this point it's customary to ask for donations so I can continue my research and all that. Well, that ain't gonna happen. All I ask is for you to stay vigilant and continue investigating and make "them" PROVE this didn't happen on Nov. 2, 2004 or in any other election.

There are some lines in the Bible about how those who tempt us are as guilty as those who commit the sin. If someone would like to quote them, I think it would be appropriate here for believers and non-believers alike. I think this notion may be rightly applied to the designers of this particular software (GEMS) who are tempting others to exploit it. After all, we are only human.

[chomskysright]

1. great info: thanks!

[Iceburg]

2. There are several good yardsticks within the canvas results

that can be used to identify potential switches. I currently am only working on one of the punch card counties but I believe the same yardsticks would apply. The first set of extreme outliers can be derived using the 97.5th percentiles of Peroutka, Badnarik and the Unounted ballots (Ballots Cast minus Votes President = UV). I use the vote distribution of the other contests in which name rotation applies (all federal, state and county contests). And lastly for a little added pleasure I use the 2000 results.
To summarize, using a combination of percentiles, regression and comparative descriptives one can identify the likely switches.

[Bill Bored]

3. I would not be surprised.

But you are working on ballot rotation switches. I'm talking about straight Bush-Kerry switches. So the third party vote may not be relevant in my case because it can be kept completely intact.

I'm glad you posted here though Iceburg. While I've been working on this for a while now, seeing your other thread pop up motivated me today! Thanks!

BTW, is anyone looking at the Touch Screen counties in Ohio?
And do you know how the Op Scans were programmed there?

I was thinking more in terms of FL myself, but there were a few Op Scan and TS counties in OH, which may not be getting enough attention because of the anti-punch card "movement."

[Iceburg]

4. Even with straight Bush-Kerry switches the yardsticks

mentioned will still apply ... in fact if only the bush/kerry votes were switched the other contests (US Senator, CJSC, County Races etc) will provide even stronger proof if their respective ballot orders were untouched.
In Cuyahoga, I use any contest that has a democratic candidate (even if they are running uncontested) with an r^2 coefficent of correlation > .85 in my regressions. Fortunately for me, there are uncontested races to use as their relative vote awards remain contant, otherwise I need to identify and use only the precincts in a vote location with a homogenous ballot style for that contest.

Have you proven that the vote switches:
a) did not involve other candidates in the presidential contest
b) did not involve other candidates in other contests

You ask -- "BTW, is anyone looking at the Touch Screen counties in Ohio?"
No I don't.

"And do you know how the Op Scans were programmed there?"
I don't know how/who programmed them in Ohio, but the machines are typically programmed by a technican working for the supplier. It is unlikely that the county staff has the resources and the expertise to set-up the machines and tabulators by themselves -- in other words it is likely they outsourced their responsibility for delivering a free and fair election to the corporate lords.

[KaliTracy]

6. Warren County (among others) had a more votes for Connally (State

Supreme Court Black female Democrat) than Kerry (in other words, * got the vote, and Connally got the vote, not the Incumbent white republican male). Punch card machines mostly there, I believe.

[Iceburg]

7. Connaly received more votes than Kerry in 12 Ohio counties

						CJ/		USSen/
						Kerry		Kerry
County		Method		CJ	Kerry	Ratio	USSen	URatio
========================================================================
Mercer		Punch Card	6607	4924	1.34	4214	0.86
Auglaize	E-Voting: 	7312	5729	1.28	4660	0.81
Shelby		Punch Card	7830	6337	1.24	4960	0.78
Clermont	OPScan:CC	29464	25318	1.16	19745	0.78
Darke		Punch Card	8817	7663	1.15	5785	0.75
Van-Wert	Punch Card	4497	4026	1.12	3659	0.91
Putnam		Punch Card	4785	4348	1.10	3758	0.86
Butler		Punch Card	59532	54185	1.10	42129	0.78
Warren		Punch Card	27812	25399	1.10	18811	0.74
Brown		Punch Card	7407	7058	1.05	5413	0.77
Highland	Punch Card	6119	6012	1.02	4390	0.73
Miami		OPScan:CC	17206	17039	1.01	12130	0.71
Union		Punch Card	6489	6546	0.99	4464	0.68
Paulding	Punch Card	3509	3544	0.99	3073	0.87
Logan		Punch Card	6564	6644	0.99	4495	0.68
Clinton		Punch Card	5161	5244	0.98	3604	0.69
Crawford	Punch Card	7418	7626	0.97	5875	0.77
Adams		Punch Card	4010	4189	0.96	3156	0.75
Allen		OPScan:PC	14946	15822	0.94	12573	0.79
Williams	Punch Card	5880	6231	0.94	4331	0.70
Hancock		OPScan:CC	9349	9930	0.94	6920	0.70
Defiance	Punch Card	6238	6683	0.93	5071	0.76
Hardin		OPScan:CC	4399	4723	0.93	3622	0.77
Preble		Punch Card	6636	7165	0.93	5315	0.74
Morrow		Punch Card	5183	5651	0.92	4031	0.71
Greene		Punch Card	26223	29349	0.89	21407	0.73
Champaign	Punch Card	6031	6752	0.89	4612	0.68
Gallia		Punch Card	4579	5188	0.88	3529	0.68
Madison		Punch Card	5357	6080	0.88	3911	0.64
Holmes		Punch Card	2295	2622	0.88	1886	0.72
Ashland		OPScan:CC	7263	8345	0.87	6168	0.74
Meigs		Punch Card	3723	4334	0.86	3064	0.71
Henry		Punch Card	4283	5027	0.85	3240	0.64
Pickaway	E-Voting	7098	8388	0.85	6180	0.74
Delaware	Punch Card	22357	26491	0.84	17603	0.66
Fayette		Punch Card	3580	4244	0.84	2620	0.62
Columbiana	Punch Card	19298	22884	0.84	17486	0.76
Hamilton	Punch Card	159926	190956	0.84	145716	0.76
Muskingum	Punch Card	13367	16050	0.83	10956	0.68
Licking		Punch Card	24349	29350	0.83	20576	0.70
Lawrence	Punch Card	9829	11874	0.83	9158	0.77
Richland	Punch Card	19868	24056	0.83	19050	0.79
Geauga		OPScan:CC	16136	19571	0.82	13219	0.68
Marion		Punch Card	9452	11492	0.82	8582	0.75
Coshocton	OPScan:CC	5507	6763	0.81	4921	0.73
Jackson		Punch Card	4489	5519	0.81	3731	0.68
Wood		Punch Card	22905	28216	0.81	18925	0.67
Fulton		Punch Card	6558	8098	0.81	4971	0.61
Belmont		Punch Card	13935	17256	0.81	13266	0.77
Tuscarawas	Punch Card	14801	18460	0.80	13929	0.75
Morgan		Punch Card	2250	2810	0.80	1955	0.70
Vinton		Punch Card	2091	2612	0.80	1912	0.73
Wyandot		Punch Card	2841	3554	0.80	2418	0.68
Jefferson	Punch Card	14917	18674	0.80	13962	0.75
Fairfield	Punch Card	19417	24321	0.80	16745	0.69
Knox		E-Voting	7598	9613	0.79	7431	0.77
Ashtabula	Punch Card	18587	23545	0.79	17969	0.76
Noble		Punch Card	2063	2618	0.79	2037	0.78
Harrison	Punch Card	3011	3824	0.79	2867	0.75
Perry		Punch Card	5599	7205	0.78	5013	0.70
Washington	OPScan:CC	9401	12137	0.77	9896	0.82
Clark		Punch Card	25284	32824	0.77	23526	0.72
Carroll		Punch Card	4749	6190	0.77	4418	0.71
Medina		Punch Card	27335	35725	0.77	24336	0.68
Hocking		Punch Card	4625	6065	0.76	4279	0.71
Guernsey	Punch Card	5388	7072	0.76	5541	0.78
Pike		Punch Card	4456	5865	0.76	4647	0.79
Montgomery	Punch Card	104891	138262	0.76	99492	0.72
Seneca		Punch Card	8170	10770	0.76	7339	0.68
Portage		Punch Card	29778	39915	0.75	29461	0.74
Sandusky	OPScan:CC	10305	13909	0.74	9164	0.66
Scioto		Punch Card	12146	16438	0.74	12906	0.79
Wayne		Punch Card	14368	19455	0.74	13796	0.71
Ottawa		OPScan:CC	8052	10915	0.74	7123	0.65
Ross		E-Voting	10037	13701	0.73	9814	0.72
Lucas		OPScan:CC	92722	128874	0.72	87686	0.68
Lake		E-Voting	41088	57471	0.71	38554	0.67
Athens		Punch Card	12275	17369	0.71	12758	0.73
Stark		Punch Card	65106	92295	0.71	64932	0.70
Huron		Punch Card	7287	10354	0.70	6860	0.66
Lorain		Punch Card	53303	76512	0.70	55732	0.73
Monroe		Punch Card	2861	4164	0.69	3328	0.80
Cuyahoga	Punch Card	294973	433262	0.68	293815	0.68
Mahoning	E-Voting:TS	55343	81500	0.68	60291	0.74
Summit		Punch Card	101635	152897	0.66	113903	0.74
Trumbull	Punch Card	42750	65321	0.65	48913	0.75
Franklin	E-Voting	174323	275573	0.63	193051	0.70
Erie		OPScan:CC	11583	20652	0.56	15092	0.73

[Bill Bored]

8. I know, Warren, Butler and Clermont were the big 3 where this happened.

And of the 16 counties where it happened for either party (i.e., there were 4 in which Moyer got more votes than Bush), Kerry got 150,000 fewer votes than Bush, if memory serves me. You would think it wouldn't be too hard to hand count the big 3, or some of those other counties where the reverse happened (Moyer got more votes than Bush) to make it fair and balanced and resolve these issues. But this was not Blackbird's intention apparently.

[KaliTracy]

9. could that be done now? by laypeople? would $$ have to be paid

to do so?

[Bill Bored]

10. This came up here before. I think it could but money would

have to be paid. I'd think the effort would have to come from the people of Ohio, although the money could probably be raised nationally. I don't know what became of this effort or the one in FL. FL has Op Scan ballots in many counties that could be checked.

Say, THIS could be a good idea for Conyers' blog!

[Bill Bored]

12. Iceburg you asked:

Have you proven that the vote switches:
a) did not involve other candidates in the presidential contest

No. Only that with GEMS ballot text editing, they did not have to. Although it might be a good idea to switch Cobb and Badnarik if you switch Kerry and Bush, just to make it look ideologically consistent. I have not given this much thought with respect to the punch cards though. The assumption has been that votes were switched by physically sticking the cards in the wrong readers, using the caterpillar crawl. Perhaps other methods were employed.

b) did not involve other candidates in other contests

I have not considered that at all, but it could be done just as easily with other races in GEMS.

Management is always a problem when there is outsourcing. It's a lot harder than many people think. A vendor has no incentive to perform well except the possible loss of the contract. In the case of an election, this is trivial and given the cost of replacing hardware and software, it's highly unlikely anyway.

When people mentioned that elections were being "outsourced", I used to think they just meant the vote counting was mechanized. But if even the setup of the machines is done by unaccountable third parties, that's just insane.

[KaliTracy]

5. lucas county was opti-scan -- remember some problems there --

but not quite what they entailed.

[Wilms]

11. .

https://voteprotect.org/index.php?display=EIRMapCounty&tab=ALL&state=Ohio&county=Lucas&cat=02&start_time=&start_date=&end_time=&end_date=&search=

[KaliTracy]

13. wow. I knew they had a hard time certifying the vote, but there were

a lot of issues there with machines (that were reported) -- thanks for showing us this link.

My mother is a nurse, and is over 60 (meaning she's been in a lot of elections when she was working). She lives in Toldeo -- works in Sylvania (both Lucas county) -- the ballots to vote weren't delivered until very late in the day -- never happened before. People who were in the hospital ready to be released (but who weren't well enough to stand in line to vote) had to wait around -- or leave and not vote.

[Wilms]

14. I've only scanned a handful of those national reports. Ohio was busy.

[KaliTracy]

15. oh, Ohio had the "smoothest" election as could be expected, though! Isn't

the Ohio SOS always right?


If anyone hasn't already heard this, it really shows the arrogance of Blackwell


Secretary of State clashes with Congresspeople at hearing on Ohio's 2004 vote.
By Karen Kasler - March 21, 2005

A Congressional committee headed by Ohio Rep. Bob Ney (R-St. Clairsville) is investigating the 2004 presidential election. The committee held a hearing at the Statehouse Monday, and sparks flew when the state’s chief elections official took a seat before the panel. Ohio Public Radio’s Karen Kasler reports.


http://statenews.org/story_page.cfm?ID=8215&year=2005&month=3


****

As Blackwell Says, Ohio’s in 2004 was a National Model
by Steve Rosenfeld, Bob Fitrakis, and Harvey Wasserman
March 24, 2005

Ohio Secretary of State J. Kenneth Blackwell finally testified – something he had refused to do in the Moss v. Bush Ohio election challenge before the State Supreme Court and refused to do in Washington, D.C. His testimony proved so contentious that at one point Rep. Stephanie Tubbs-Jones, D-OH, told him to “haul butt” if he was unwilling to answer questions about irregularities in the 2004 election.

Blackwell vigorously defended his role in last fall’s presidential election at a congressional hearing on Monday, March 21, at the Ohio Statehouse, claiming critics have smeared his state as if it were a “third world country” rather than the national model of election administration that Blackwell said it was. In December, Republican state senators blocked a similar Democrat-sponsored forum from using the Statehouse, forcing testimony to be taken at the Democrat-controlled Columbus City Council chambers. Meanwhile, hundreds of disenfranchised voters testified under oath in Columbus, Cleveland, Cincinnati, Toledo and Warren concerning their voting day hardships.

“We had a good election in the state of Ohio. Not a perfect election – elections are human endeavors,” Blackwell, a Republican gubernatorial candidate and co-chair of the Bush-Cheney campaign, told the House Committee on Administration in Columbus. In opening remarks, he also noted Ohio coped with a million new voters and tremendous efforts by both political parties to exploit legal tactics to their advantage.

“While much has been written by the conspiracy theorists, I would like to point out that there has only been one complaint filed by the HAVA process,” Blackwell said, referring to the Help America Vote Act, which was enacted by Congress after Florida’s 2000 election debacle. “I am interested in clean, fair and transparent elections.”

more: http://www.freepress.org/departments/display/19/2005/1208

[Wilms]

17. That subject line was a coffee spewer!

Thanks for the laugh this morn.

[KaliTracy]

19. LOL.... thanks for the mental picture... i hope your monitor

cleaned up ok.

[L. Coyote]

18. Ohio Stats

2004 Ohio Voting by County: Summary	

	  State		E-Vote		Op-Scan		Punch

Pop	  11,435,798	1,799,299	1,382,208	8,254,291
# Voters  7,979,639	1,344,131	925,806		5,709,702
# Votes	  5,574,476	870,237		670,058		4,034,181
Percent   		15.61		12.02		72.37

This is the list of counties where Bush's increase in votes (2000 vs. 2004) exceeds Kerry's gain by over 2,000 votes. (Asterisk denotes electronic voting.)

		Bush Net Gain
Butler		12,353
Warren		11,948
Clermont*	10,174
Delaware	6,241
Miami*		5,547
Licking		4,758
Greene		4,605
Shelby		3,605
Fairfield	3,278
Darke		3,130
Mercer		2,825
Richland	2,631
Union		2,585
Auglaize*	2,502
Clinton		2,348
Hancock*	2,228
Lawrence	2,080

Total:          82,838

If only counties where the Dem percentage is less in 2004 are included, the counties with the highest Bush Net Vote gains are:

Butler		12,353
Warren		11,948
Clermont*	10,174
Miami*		5,547
Greene		4,605
Shelby		3,605
Darke		3,130
Mercer		2,825
Richland	2,631
Union		2,585
Auglaize*	2,502
Clinton		2,348
Hancock*	2,228
Lawrence	2,080
Marion		1,990
Van Wert	1,988
Clark		1,944

Total:         74,483

When the counties with the greatest percentage loss to Dems are listed, we have:

		Dem Percent     Bush Net		
Shelby		-5.59		3,605
Van Wert	-5.44		1,988
Mercer		-4.44		2,825
Darke		-4.22		3,130
Lawrence	-3.65		2,080
Clinton		-2.94		2,348
Miami*		-2.68		5,547
Jackson		-2.57		1,036
Auglaize*	-2.54		2,502
Paulding	-1.96		  704
Marion		-1.83		1,990
Hardin*		-1.83		  897
Highland	-1.68		1,459
Wyandot		-1.61		  686
Champaign	-1.61		1,415
Belmont		-1.43		1,374
Preble		-1.34		1,546
				
Total:			       35,132

The county stats do not seem to raise any big alarms. Precincts may be different. To see more go to the Ohio spreadsheet, 2000 worksheet, range A260:T400.

[Zan_of_Texas]

16. Here is more on the punch card reading.

Triad is a small, family-owned Ohio-based entity that counts the votes in 41 of the 88 counties in Ohio.

President is Brett A. Rapp
Vice President is Dwayne A. Rapp

Two journalists, Terri Taylor and Evan Davis, caught up with these two Rapps at their tiny headquarters after the election.

Some notes (not exact dialog) from that recorded interview, which I have heard:


1:51 We do not supply punch card ballots. We use standard Documation ballot readers that have been available since the early 60s. We have an interface box that connects to the personal computer - the header card tells the ballot reader which precinct is being tabulated. End card is at the end of the stack. Pink card is the CANCEL card. Green card tells it, end of stack. Ballot tabulator is Reset, ready to count.

3:06 Rotations. Stores ballot counts on the computer hard drive. Repeat for each precinct. At the central computer, that hard drive is storing all of the precint results for each precinct. There are several commands that the operator can give it, to give a cumulative report, or pct by pct reports. End of night - unprocessed pcts -- to make sure all tabulated. Then, run a final cumulative report. ...

4:47 We do supply the software, and the interface box that connects the ballot reader to the computer - that is part of our software.

So, on these punch card systems, the "interface box" supplied by Triad could be rigged. Or, it sounds like messing with the header cards or vote stacks could rig the vote -- what if you took 50 vote cards from Precinct A and put them in the stack for Precinct B, and vice versa? With the ballot rotation, that would mess up the vote totals. Many precincts in Ohio had 2, 3, 4 or 5 precincts voting in the same building, not separated by any physical barrier. Each would have had a different ballot rotation. Best I can tell, the punch cards for each precinct in the same building were NOT different colors either. It's almost as if the chaos (which was reported by those on the scene) was predetermined, eh?

[Iceburg]

21. Exactly ....

A oft cited case in the Florida 2000 litigation cases....
"In May, 1968, in Klamath County, Oregon, candidates' positions on the ballots were rotated in the precincts to avoid giving any candidate the unfair advantage of the top position everywhere, but the ballots got mixed up, and voters in more than a fourth of the precincts punched out rectangles for candidates they did not mean to vote for.

In 2004 Cuyahoga County 5 distinct ballot order styles were deployed in the contest for president. Approximately 90% of the 1436 precinct were multi-precinct vote locations. Since the state requires by law that the candidates names in the federal, state and county contests be rotated on the ballot, when we include the ballot styles from the 40 other candidate contests in Cuyahoga the total number of unique ballot styles deployed in this county was 1105.

This is an astonishing number of combinations when you consider that the ballot pages in the vote book must be in precise harmony with both the card readers and tabulators that eventually determine the results. What test procedures were implemented in Cuyahoga by Triad to ensure the machines, and their respective vote books and pages were set-up appropriately, and tabulators coded accurately

Clearly to test the validity of the vote machine setup would require considerable time. Further, to test the validity and accuracy of the tabulator would require not only considerable time but an extremely large test deck of ballot cards.

It takes little imagination and even less cynicism to visualize the possibilities of accidental or deliberate mis-tabulation of ballots, as the vote machines are set up, the punch cards transferred from precinct to the central counting location where they are handled and stacked by personnel with little understanding of the fragile process and the precarious application entrusted with accurately recording the voter's intent. Even if the execution was flawless to this point (no machines/vote books set-up in error, no cards mislaid) one wrong or misplaced precinct header card, just one single defective line of code and the voters intent is effectively annulled.

[Wilms]

20. .

:argh:

[Wilms]

22. .