Sarasota's Diebold scanners flunk test, fail to read blue ink on ballots

[WillYourVoteBCounted]

Sarasota's Diebold scanners flunk test, fail to read blue ink on ballots

Saturday, October 17, 2009 VOTING NEWS BLOG

WARNING: Voters in jurisdictions that use the newer Diebold/Premier optical scanners OSX with 1.0.2 firmware should use black ink to mark their ballots. The machines may not read markings made with blue ink. The OSC also has trouble reading markings by a number 2 pencil according to other reports. This is an intermittent problem.

In the logic and accuracy testing of Premier Optical Scanners, which will be used in Sarasota elections beginning October 19th, test ballots were run with blue marking of ovals. The result was an under vote in races where blue ink was used to fully fill out voting ovals. The machine tested was the Diebold/Premier AccuVote OSX Digital Scan Tabulator (Hardware Model A) with Firmware 1.0.2 .

Scanning glitch in vote machine test Saturday, October 17, 2009

A voting machine test this week didn't go off without a hitch.

A handful of individuals were on hand this week at the Sarasota County Supervisor of Elections office where Supervisor Kathy Dent and staff tested 16 ballot scanners that will be used in the Nov. 3 city council election.

In one test a ballot with both blue and black ink was unable to be scanned. Officials pulled that scanner from service and substituted another. It didn't work either.
...
"One big factor is this is an intermittent problem," Parish said. "Until the defect in the pickup of votes is discovered and fixed and then tested to stress levels, you will not have a solution. To do that the state needs to go after the problem and get it fixed."

According to the SOS website, the machine tested, is the AccuVote OSX Digital Scan Tabulator (Hardware Model A) with Firmware 1.0.2 (WCER7-500.3.1, BLR7-1.3.9)

The following counties are listed by the FL SOS website as having the AccuVote OSX have this model of Premier optical scan machine :

HILLSBOROUGH*, LEON, LEVY, MADISON, MONROE, OKALOOSA (some), OKEECHOBEE, OSCEOLA, PUTNAM, SARASOTA*, SEMINOLE, TAYLOR,VOLUSIA, AND WALTON. Only Hillsborough and Sarasota use the Premier/Diebold OSX exclusively for in person and absentee.

Another snafu in the making? Sarasota Florida is challenged once again with a flawed system under the leadership of Election Supervisor Kathy Dent. This is same Kathy Dent who presided over the Sarasota FL 13 2006 Jennings/Buchanan Race election debacle. Sarasota suffered a shocking 18,000 under votes causing doubt of the election outcome, and resulted in investigations, lawsuits and national embarrassment

The debacle might have been avoidable: Kathy Dent failed to heed the warning letter that the voting vendor sent August 15th. ES&S warned of a problem with the touch-screen smoothing filter that could cause voters to under vote. The letter urged Dent:

"train your poll workers and voters to expect this slightly delayed response time for their highlighted selections. We have included with this mailing a sample voting booth instruction sign for your review and use."

But Kathy Dent did not put the signs up in polling booths. During that same election, there was a initiative on the ballot to get replace the paperless touch-screens and replace them with paper ballot based systems.

The GAO concluded somehow that the touchscreens were not to blame. (In spite of the fact that the vendor felt it necessary to warn officials of potential undervotes on the systems). Factors including ballot design and the voters. of course, were left holding the blame.

Does your state use the Diebold OSX? If so, have extensive tests been run on it to check scanning ability for various types and colors of ink?

No new voting systems should be certified and or sold until testing is sufficent to protect the public from voting systems that aren't ready for prime time.


Voting News by Joyce McCloy.
Archives here at http://votingnews.blogspot.com/
Subscribe to receive the Voting News by email by visiting this link:
http://feedburner.google.com/fb/a/mailverify?uri=VotingNews&loc=en_US
email comments, questions us at ncvoter (at) gmail.com

=============================================

[MrModerate]

1. It's hard to tell whether the Diebold organization is criminally incompetent . . .

Or just plain criminal. Scanned paper ballot technology has been reliable for what, 60 years? And they've just now gotten around to screwing it up? Jebus wept.

[eppur_se_muova]

2. No one could have anticipated blue ink! nt

[WillYourVoteBCounted]

3. also Number 2 pencil

although the news article did not go into that, a Florida colleague sent me report
showing the machines also had problems with that.

And it isn't necessarily the ink, but the "spectrum" of the ink, i.e the machines can be
calibrated so that they don't read black gel ink for example.

Its complicated but it would take alot of testing to find all of the vulnerabilities.

How many states use the OSX now?

I thought Maryland was considering them, this is the NEW generation.

And we've seen that Diebold/Premier had other flaws in all systems for years.

[AndyA]

4. The fix is in! Republicans = black ink, Democrats = blue ink!

Florida...again...:wtf:

[nosmokes]

5. I find it odd that Diebold machines carry out millions of government approved

transactions at race tracks and casinos across the country and can in fact provide paper receipts with only a few minor glitches relative to the number of interactions until it comes to the simple act of voting, and then their success rate goes right into the shittoire. Astonishing.

[Festivito]

6. The circle fill-in seems to me a stupid implementation.

Rife with problems, open to blatant manipulation.

All one need do is print some ballots with the circles a little off so that some will count and some won't and send them to certain precincts. Then certain candidates don't win those areas and lose the election overall.

They're hard to fill for old and young.

What does happen if one extends beyond the border? Does it invalidate the vote or does it not?

The connect lines method is at least better, better than disturbingly bad.

[OnTheOtherHand]

7. I'll have to check this, but...

I think some studies indicate that in practice, the connect-lines method is even worse (ETA: i.e., seems to lead to a higher undervote rate). (Of course, that's assuming the ballots aren't deliberately misregistered. I don't know what method positively protects against that.)

When an op-scan ballot is done right, it is actually fairly error-tolerant. Even an "X" through the oval often will darken enough pixels to count as a vote; neatness doesn't count much unless one actually darkens an adjacent oval. Manual recounts of op-scan ballots have generally had discrepancy rates under 1%, which largely offset. (Of course even small discrepancies can be crucial, as we saw in Minnesota.) But sometimes they uncover howlers. And since recounts and audits aren't routine, there's no way to know how many howlers (intentional or otherwise) go uncorrected.

[OnTheOtherHand]

9. for instance...

Previous studies find higher rates of residual votes (and overvotes in particular) on ballots with the connect-the-arrow format (Bullock and Hood, 2002; Kimball and Kropf, 2005), which we expect to find in this study....

...Consistent with previous studies, we find that {in 2004} the connect-the-arrow ballot format produces significantly higher rates of residual votes for president and for ballot measures (although the effect on ballot measures is only marginally significant). More specifically, holding other factors constant, the residual vote rate for president was 0.5 percent higher in counties with the arrow method for completing the ballot.

Kimball & Kropf 2008, pp. 13 and 19 (pdf pp. 16 and 22)

Reportedly (see p. 3) Bullock and Hood, studying the 2000 Georgia general election, found twice as high a residual vote rate in connect-the-arrow counties as in fill-the-oval counties.

[Festivito]

10. Thank you. I should state that ballots need to be hand counted.

Eventually, all vote counts done by machine should be hand checked.

Yes, for the times it is checked the results will be good. The people who would to steal elections know better than to do it when likely to be caught.

Yes, for the times before it becomes common enough to wend a way into manipulation. Months, years, a decade, then it's back to stealing as usual.

Yes, machine counting is wonderful, but it not good enough without that final, overriding human recount.

The only better intermediate step I can think of is to PDF all ballots and place them on the internet, and let the machines read them from there. Let everyone who feels like it download each vote and from their own computers verify each vote, each precinct, each election.

Even those votes should be hand checked as desired by any American.

[WillYourVoteBCounted]

8. We used "circle fill-in" in North Carolina, our undervote was low, here's study

North Carolina went to mostly optical scan in early 2006.
We had a very high undervote for presidential elections since at least 2000.

Now, in 2008 we cut our undervote rate in half, with about 80 (of 100) counties mostly optical scan:

June 26, 2009 - A professor's study of North Carolina's 2008 Presidential election shows that optically scanned paper ballots were better at registering the intent of the voters than touch screen voting machines..more at the link
http://www.ncvoter.net/undervote.html

Prior to this we had 48 counties with optical scan ballots that used the "draw the line" method.