Forum Navigation
Topics
Log In
Log Out
:
Forum Search
New Today
New This Week
Advanced Search
Tree View
Forum Account
Edit Profile
Register
Forgot Password
Forum Tools
Help/Instructions
Policies
CLICK STATE TO SEE:
"WATCH LIST"
Marked with: 
"OPEN & HONEST"
Marked with: 
...
|
| 9-7-2005: Digital Image ballot scanne... |
|
| Author |
Message |
    
Bev Harris
Board Administrator
Username: Admin
Post Number: 1845
Registered: 12-2004
Best of Black Box? 
Votes: 1 (A keeper?) | | Posted on Wednesday, September 7, 2005 - 5:31 am: | 
|
This topic will turn out to be less boring than you think:
This week and next, Black Box Voting will explore different kinds ballot scanning systems, and what you can do with them. ("Ballot scanning system": You fill in an oval or draw a line on a paper ballot, place it in a scanner. Often called optical scan, but as you'll see, there are different types of scanners.)
You'll be surprised at differences in accuracy, security, and ability to exercise citizen oversight depending on the type of scanning technology in your jurisdiction.
Ballot scanning systems - Usually known as "optical scans," there are actually significant differences in scanners, one of which, the digital image scanner, offers more opportunities for citizen oversight. Attack points differ among the systems, as do security measures, but the biggest difference of all is in your improved ability to exercise citizen oversight with the digital image scanners.
A digital image scanner creates a picture of the image, which it stores. Because so many of us use digital cameras, most of us understand what a digital image is. There is not a generally-agreed-upon term for the opposite -- a scanner which cannot make a picture, but instead interprets information directly off the ballot.
The new digital image scanners essentially create photocopies of each and every ballot. Older optical scan styles capture an array that is too sparse to form an image. Instead of capturing a picture, the older versions interpret patterns as the ballot feeds through the machine.
The key difference between the digital image scanner and the older optical scan machines is that the digital image scanner stores the raw data itself, in the form of a picture of the ballots, whereas the traditional optical scan machines do not retain any raw data, and cannot capture an image with sufficient precision to produce a picture.
There is another advantage to digital image scanners, however.
Digital images of each ballot -- precise pictures of the ballot -- can be obtained by news reporters, candidates, and ordinary citizens through public records. You can request the images on CD and examine precise, high-resolution images of each ballot on your home computer.
We'll show you sample files, like those you can obtain for yourself in jurisdictions with digital image scanners. In the upcoming articles, we'll provide guidance on exactly which ballot image files to request. We think you'll be excited by the possibilities.
Touch-screen locations use ballot scanners
Even in locations with touch-screen voting, absentee votes are counted by ballot scanners. As you will see in the Black Box Voting articles during the next several days, the type of scanner used in your location will ultimately make a big difference in the integrity of your elections.
What is Central Count?
Central Count is the ballot scanning that takes place at jurisdiction headquarters for absentee votes and also for recounts. There is sometimes a difference in your Central Count ballot scanning equipment, as compared with the equipment used to scan ballots at the polling place.
Take a moment to call your local elections office to find out whether they use a digital image scanner (one that captures a picture of each ballot) or not. You may want to clarify whether they use different technology for Central Count than polling places.
Is your local jurisdiction involved in a pending purchasing decision?
If so, pay extra close attention to this week's and next week's consumer reports at Black Box Voting.
Can't the digital images be manipulated?
Some citizens are concerned that you can manipulate the stored picture of the ballot, which, in digital image scanners, is used to tally the vote. You will want to tune in for the follow-up reports this week and next to learn more about this risk, and the procedures that can mitigate risks.
In a nutshell, manipulating the stored ballot images requires either inside or remote access, and simple image editing skills, but leaves a clear forensic trail which can be analyzed by any citizen. The paper ballot should certainly not be destroyed in favor of retaining a digital image. Both must be retained for the system to have integrity.
Easy way to manipulate the traditional ballot scanning systems, especially the systems that use infra-red
Just switch pens.
Which manufacturer makes what?
Election Systems & Software (ES&S), Sequoia, and Diebold, make the traditional ballot scanners. Diebold has a ballot scanner, for instance, that captures an array of lines on the ballot, and interprets them as votes if they appear in the right places.
- The newer technology, certified digital imaging systems are made by Diebold and Hart Intercivic, and are in use right now in some locations.
If you have Hart Intercivic scanners, you have the newer technology, digital image scanners, for central count (absentee ballots), and if you have Hart Intercivic scanners at the polling place, they are also digital image.
If you have ES&S or Sequoia, most likely you have the older technology.
If you have Diebold, you may have the newer technology, digital image scanners for Central Count. Diebold uses the older technology scanners at the precinct.
Some Diebold jurisdictions use the non-digital image technology for both polling places and Central Count.
Diebold offers two kinds of enhancements for central count -- a tray that fits on the older scanners, and a new digital image scanner. Because Diebold ballot scanners come in different varieties, it may help to get the version number for both the precinct scanners and the central count scanners. Diebold's digital image scanner has a version number in the 2.0.xx range, whereas its other scanners have version numbers in the 1.94x and 1.96x range.
A tedious exercise? An extra-mundane story?
Hardly!
Tomorrow, we will show you what you can do with the different kinds of ballot scanning systems, and (depending on which kind you have) you'll discover a whole new kind of empowerment. |
John Washburn
Voting Rights Forum Participant
Username: Johnwashburn
Post Number: 81
Registered: 04-2005
Best of Black Box?
Votes: 2 (A keeper?) | | Posted on Wednesday, September 7, 2005 - 7:25 am: |
|
Easy way to manipulate the older infra-red and visible light systems
Just switch pens.
I realize this is "retail" election fraud. But for the infra-red optical scanners, it may be possible to pre-paint the ballots. What you need is an ink which is infra-red absorptive (nearly 90% absorptive), and optically clear; such as this pen. Pre-paint ballots with this ink for the candidate you prefer. (E.g. 200 for John Kerry).
The pens in the voting booth are optically opaque and about 50% absorptive in the infrared range. Since the scanners are designed to read the strongest mark (in case of erasure), the pre-painted mark will be tallied in preference to the elector created mark.
Since the rest of the ballots has marks which are 50% and nothing else (0%). The elector marks on these races will be tallied as per the programming.
The worst part of this scheme is it requires only one corrupt party with access (or B&E skills) to the blank ballots. Even if everyone else and the machines are scrupulously honest, the fraud is hard to detect.
In Liberty,
John Washburn
|
John Washburn
Voting Rights Forum Participant
Username: Johnwashburn
Post Number: 82
Registered: 04-2005
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Wednesday, September 7, 2005 - 7:34 am: |
|
Ah! The internet is wonderful thing. Here is an infrared camera from the same company that make the pens above. The camcorder is specifically designed to view the writing made by the infrared pens pen mentioned above. Both IR and optical light is recorded.
Click thorugh to the order page and there are a host of IR digital cameras and IR flashlights.
A way to check for pre-painted ballots?
In Liberty,
John Washburn
|
Catherine Ansbro
Frequent Voting Rights Forum Participant
Username: Catherine_a
Post Number: 832
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Wednesday, September 7, 2005 - 8:51 am: |
|
Is there any hope of getting permission to view ballots using this kind of camcorder? The person using the camcorder wouldn't have to actually handle the ballots. |
John Howard
Frequent Voting Rights Forum Participant
Username: Harmonyguy
Post Number: 110
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Wednesday, September 7, 2005 - 3:05 pm: |
|
My favorite topic !!!!!!
HG |
Pat A. Vesely
Moderator
Username: Pat_vesely
Post Number: 1933
Registered: 12-2004
Best of Black Box?
Votes: 1 (A keeper?) | | Posted on Wednesday, September 7, 2005 - 3:29 pm: |
|
I would assume that you could get a pair of glasses with an infrared bandpass coating that would give you some visual indication of tampering but I would have to know more about the frequency range of the detector that is used in the machine.
A more likely scenario for cheating is to have the technician who performs the sensor calibration slightly shift the sensitivity of the sensor that detects a given ballot position. If you make it harder to detect the marks for your opponent and easier for your candidate, you can skew the count by a small amount.
The best part of using this method is that even if you get caught, you can always claim that the calibration must have 'drifted' due to heat, humidity, or aging since you last re-calibrated it. Built in plausible deniability.
This brings up another point, I'm unaware of any form of calibration records that are kept that include documentation of critical settings and adjustments in the machine that are traceable back to the NIST.
Without exercising some control over how these adjustments are done and some standard to measure them against, there is no assurance that the machine will function in the intended manner.
This should be a requirement for any 'mission critical' system purchased under government contract using tax dollars. Those documents should also be made part of the public record and available to inspection by anyone. No oversight, no trust.
Just one more thing to keep in mind when asking our representatives to draft legislation.
PAV ;-) |
John Washburn
Voting Rights Forum Participant
Username: Johnwashburn
Post Number: 83
Registered: 04-2005
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Wednesday, September 7, 2005 - 4:39 pm: |
|
Is there any hope of getting permission to view ballots using this kind of camcorder? The person using the camcorder wouldn't have to actually handle the ballots.
The camera is both optic and infrared. Why tell anyone of the IR capability? Just ask to use your camcorder. WI statute allows video cameras as long as you are not trying to view the ballot of an elector. Recording the canvassing or recording the poll opening and blank ballots would seems to be fair game.
In Liberty,
John Washburn
|
Bev Harris
Board Administrator
Username: Admin
Post Number: 1853
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Thursday, September 8, 2005 - 11:38 pm: |
|
John (Howard or Washburn, you both possess amazing information) --
Do you know if the AccuPoll optical scan machine produces a digital image?
Are you familiar with any other certified vendors that have this technology on their ballot scanners?
What about the AutoMark -- does it produce a digital image? |
Bev Harris
Board Administrator
Username: Admin
Post Number: 1854
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Thursday, September 8, 2005 - 11:47 pm: |
|
Pat,
You pose very good questions about calibration. It isn't just touch-screens that are calibrated -- the optical scans are, too.
I believe they are calibrated by running a calibration sheet through that is provided by the manufacturer, though the process may differ from vendor to vendor. This presents at least two attack points:
1) A rogue employee at the vendor might be able to do something we don't want, when producing the calibration sheet.
2) The security surrounding the machines during calibration -- and who does the calibration -- is important. The setup for calibration is sometimes more casual and accessible than other phases of the elections process.
I also have wondered about the calibration logs kept. My husband and I were just talking about this at dinner. Where he works, not only must a calibration log be kept for items that require calibration, but it must be signed off on by a supervisor when the device is calibrated. |
John Washburn
Voting Rights Forum Participant
Username: Johnwashburn
Post Number: 90
Registered: 04-2005
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Friday, September 9, 2005 - 9:04 am: |
|
Do you know if the AccuPoll optical scan machine produces a digital image?
I have no information on any scanning done by an AccuPoll machine. I thought AccuPoll was an integrated (vote recording, ballot verification, and vote tallying) DRE. That seems to be the indication on the AccuPoll website. (BTW, I love the note that the AVS 1000 prevents undervoting. You must vote for someone, Mr. Washburn!)
Are you familiar with any other certified vendors that have this technology on their ballot scanners? No. The only precinct scanner I was able to examine with any detail (and not much at that) was an ES&S Opitech Eagle back in 1992. This is not certified equipment under HAVA. But, The Eagle used infrared scanning which I thought was decidedly odd for an optical scanner. The investigation of IR scanning is what led me to the idea of pre-painting ballots. Unfortunately, for IR-based optical scanning, IR security inks have only gotten better. For security inks better means less visible to the human eye (optic range) and more absorbptive (or more florescent) in the IR range. This make pre-painting ballots even more viable than it was in 1992.
Now, I would distrust any ballots printed by the vendor of the IR scanner. The ballots may come from the "factory" pre-painted.
What about the AutoMark -- does it produce a digital image? I have no information on the AutoMark.
In Liberty,
John Washburn
|
Bev Harris
Board Administrator
Username: Admin
Post Number: 1859
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Friday, September 9, 2005 - 9:10 am: |
|
All vendors must produce some sort of optical scan type machine, or else the jurisdiction that buys the system must buy one from a different vendor, complicating the tally, which merges absentee votes with polling place votes.
Any jurisdiction that uses absentee voting cannot depend solely on touch-screens, or even early voting, for that matter. (Early voting doesn't help if you happen to be in Australia at the moment). |
John Howard (hg)
Voting Rights Forum Participant
Username: John_howard
Post Number: 2
Registered: 08-2005
Best of Black Box?
Votes: 1 (A keeper?) | | Posted on Saturday, September 10, 2005 - 9:53 pm: |
|
So, just how does ES&S feel about its own Model 100 Precinct counter, currently in use in many jurisdictions?
These are THEIR words, not mine.....
"Election Systems & Software, Inc. developed its Model 100 Precinct Counter with a Contact Image Sensor (CIS) as its reading technology. The CIS integrates a light source, lens, sensor, and video signal amplifier into a single module. It is essentially the same type of CCD or CMOS sensor that is used in fax machines and page scanners, which are well known in the art and is capable of reading the entire width of an 8.5" wide response sheet. With this capability, the system was designed to employ readily available computer hardware and software to specify which segments of the sensor would be used to read the response sheet and to record, analyze and classify the generated data."
"The employment of the CIS created a number of challenges regarding the accuracy, consistency, and the selection of marking devices employed in the system. Response sheets printed out of specifications, overly sensitive read heads, and smudged sensors created problems that lead to false readings. These problems lead to the development of a Model 100 system setup routine in which blank response sheets were fed into the machine and the sensitivity of the sensors was adjusted to eliminate false readings. This solution proved to be less than ideal, however, because the sensitivity of the sensors was occasionally too low to give an accurate reading of the response sheet. Also, the accuracy of the machines was often subject to the failings of the operator. Furthermore, inaccuracies were detected due to improper ballot printing and skewed feeding of sheets through the system."
From:USPTO patent #6,854,644
Filed: September 16, 2002
Granted: February 15, 2005
Assignee: Election Systems & Software, Inc.
So, I have to wonder if they continued to sell the Model 100 precinct scanner after they discovered (by 2002) that it faced a number of challenges that lead to false readings? Having discovered these challenges, I also wonder if they'll be going back to their customers offering any kind of significant compensation for these systems. Perhaps there should even be compensation for the poor training of the operators since they now claim that "...accuracy of the machines was often subject to the failings of the operator."
HG |
John Howard (hg)
Voting Rights Forum Participant
Username: John_howard
Post Number: 3
Registered: 08-2005
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Saturday, September 10, 2005 - 9:56 pm: |
|
Bev - I've looked for data on AccPoll's optical scan system. Are you really sure it exists? I can't find a thing.
HG |
Bev Harris
Board Administrator
Username: Admin
Post Number: 1869
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Sunday, September 11, 2005 - 10:44 am: |
|
John -
AccuPoll
Then it appears that AccuPoll may not have a ballot scanner. If so, jurisdictions that buy AccuPoll will need to turn to another vendor for absentee ballot scanning.
ES&S
The ES&S information is interesting. Need to review it more carefully before writing more.
The issue of false readings
However, it is true that there are challenges that can lead to false readings for all of the vendors in all of the kinds of ballot scanners.
Some of these challenges are not due to the vendor's programming, but are inherent in the scanner hardware they select.
For Harri Hursti's invention, Hursti had to address corrective features because of problems inherent in some of the scanners selected by the vendors. He also noticed that one manufacturer that used a buggy scanner hardware failed to build in the appropriate corrective measures, and made several other design choices that favored speed over accuracy.
His invention will help vendors identify and improve accuracy problems, if they haven't already.
Ability for citizens to test the effect settings have on accuracy
This is another reason Hursti's invention is so important. It pits one technology against an entirely independent technology, and is highly likely to catch ballot scanner machines that are poorly maintained, or were improperly calibrated, along with a number of other quality control problems.
Whereas we must accept whatever settings are used by the elections officials, Hursti's software allows the user to tinker with results at different settings, using real elections ballot images, thus examining accuracy factors for ourselves.
Will any vendors or elections officials perceive Hursti's invention to be a hostile act?
It will be interesting to see who resists the use of Hursti's software, either by recommending to local officials that they obstruct the records request or by trying to take the image retention feature off the market.
If a vendor is openly cooperative with the concept of allowing the public to examine the image files, this indicates confidence in its product. We would consider such vendor behavior to be a strong indicator of integrity.
Observing elected official's behavior
It will also be interesting to note any legislator who tries to slip impediments to the public obtaining image files into legislation. We should also watch very carefully any secretary of state who advises election officials that allowing citizens to examine the image files should somehow be prohibited. These would be behaviors consistent an enthusiam for vote manipulation, regardless of the rationale used to try and justify such a move.
In fact, one legislator reacted to the news of Hursti's invention by instantly trying to find reasons not to allow it.
- "Too much handling of the ballots" (no ballots are handled when transferring image files to disk).
- "Only certain recounts are allowed" (Checking an image file is not a recount. We'll litigate that one if necessary).
- "Well the images won't match if checked, because of voter error in filling out the ballot, which produces false positives and false negatives."
That's like saying that balancing your checkbook erodes your confidence in the banking system because sometimes you have to go figure out why the balance doesn't match.
It will also be interesting to see if any vendor decides to pull the image capture technology off the market, now that we can check its accuracy.
Note that Hursti's tool is also a wonderful idea for election officials, one that the honest officials will like. There is nothing preventing them from checking up on their own voting system with this! We are meeting more and more elections officials who want to be able to do evaluations of their equipment that are independent of the vendor, and Hursti's invention gives them one such tool.
Image scanners vs. scanners that don't produce an image
Done properly, digital image scanners should be significantly more accurate than the other models, which interpret data from an array of lines which act as "indicators" of the vote (but are not actual images of the vote).
Keeping the most precise raw data possible is desirable
In general, if using any computerized vote counting system, you want the interpretation of the vote to be as close as possible to the raw data.
Image scanners collect and retain every pixel (or should). The older models which do not retain an image collect only a grid, or lines, then search for patterns that are indicative of a vote. That process is quite far from collecting the actual raw image of the vote itself. |
John Howard (hg)
Voting Rights Forum Participant
Username: John_howard
Post Number: 4
Registered: 08-2005
Best of Black Box?
Votes: 1 (A keeper?) | | Posted on Sunday, September 11, 2005 - 2:13 pm: |
|
May I take the liberty of adding to your statement?
"Image scanners should collect and retain pixels in sufficient resolution to be able to retain an image of the entire ballot paper. The older models which do not retain an image collect only a grid, or lines, then search for patterns that are indicative of a vote. That process is quite far from collecting the actual raw image of the voted paper-ballot itself."
The ES&S Model 100 precinct scanner collects a multi-line, multi-column picture of the vote target area, which is vastly better than that of the Diebold Accuvote OS precinct scanner (both visible AND IR models), which only collect a single-row, single-column bit of data from which to interpret a voter's intent.
While certainly no fan of ES&S, at least they do collect more than a single bit of data from which to attempt to interpret voter intent. Still, it needs to be understood that the precinct scanners from either of these vendors do NOT product true scanned images of the ballot paper.
For any election officials that take comfort in using the 'Ballot Images' menu selection in GEMS, while the output may appear like a scanned ballot, it is NOT a scanned image of a voted ballot. It is, in fact, a blank ballot template with the file
c:/GEMS/images/Filled Oval.wmf
superimposed over each of the places where the scanner decided that a voter had left a mark. This is why the so-called 'ballot images' appear to be so pristene.
HG |
Bev Harris
Board Administrator
Username: Admin
Post Number: 1874
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Sunday, September 11, 2005 - 2:33 pm: |
|
John,
Thanks for your very articulate clarification.
I believe ballot image files created by the Diebold High Speed Central Count 2.0.xx are retained in three places: The ballot scanning machine itself, which runs on a Windows operating system (unlike the Diebold precinct-based optical scan machines), GEMS, and the SQL database. The GEMS files in are a directory called "CC" or "Central Count." Within that are more directories, each identified by an automatically generated number from each batch. Inside these directories is an image file for each ballot in the batch. |
Ellen H. Brodsky
Voting Rights Forum Participant
Username: Ehbrod
Post Number: 20
Registered: 01-2005
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Thursday, September 29, 2005 - 9:52 am: |
|
In the description of the ES&S Model 650 Absentee Ballot Scanner it's description says it uses visible red light??? This is most confusing to me as it might block out a portion of the electromagnet frequency??? What is your opinion? Does this scanner have the ability to capture a digital ballot image? |
Catherine Ansbro
Frequent Voting Rights Forum Participant
Username: Catherine_a
Post Number: 939
Registered: 12-2004
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Thursday, September 29, 2005 - 1:45 pm: |
|
It sounds like it's using a light and sensor in the visible frequency range (as opposed to a light and sensor in the infrared which could be invisible depending on the particular part of the spectrum).
It sounds like they're presenting the fact that it's a visible light (the operator can see it and know it's "working") as a confidence-builder for election officials.
Any light or sensor will favor certain parts of the EM spectrum. It's a question of choosing the right one for the right job, and being aware of any particular risks or special requirements (special paper or ink) of a given piece of equipment. (E.g. the ability to rig an election by putting out pens with the wrong kind of ink for the kind of sensor in a particular scanner has been mentioned here.) |
Brant Lamb
Frequent Voting Rights Forum Participant
Username: Brantl
Post Number: 118
Registered: 01-2005
Best of Black Box? N/A
Votes: 0 (A keeper?) | | Posted on Wednesday, October 5, 2005 - 4:41 am: |
|
As in all the rest of this, if your ballot has an ID number on it and you get a legally binding copy(that is marked in visible ink, of course), and they have to post the votes, and you have the legal right to get your vote corrected on the evidence in your hand, you short-circuit all of this. |
|
|
