a crazy idea?

[garybeck]

let the scanner make a GIF image for each ballot.

zip up all the GIFs into one file per precinct.

post it on the internet.

let everyone hand count the ballots. at home in their underwear.

[saltpoint]

1. That part about the underwear is making us all hot, garybeck.

I mean, totally.

___

I like it. Any plan that clear and sensible has an enormous appeal to me. Once the GIF file is completed, could it be emailed to both parties' headquarters, the Sec. of State's office, and various media outlets?

It would be tough to tamper with a document that was going to all those places simultaneously.

It's Blackwell-proof.

[garybeck]

2. well if you think about it,

if one person tried to count them all by themselves, it would not be very efficient. there would have to be several people together counting them, to save time. makes the underwear part even more interesting. but I guess that doesn't have to be a requirement.

[saltpoint]

5. But I definitely like your idea on the GIFs. Internet traffic on election

night would be a headache, but democracy is worth a few headaches.

No kidding -- I think you're on to something.

[Wilms]

3. I've no way of knowing where those gifs are from, or have been.

But it's an interesting thought.

[garybeck]

6. if everyone has access to them

such a public document could not be tampered because everyone could have a copy of it. if someone altered it, it would be different than the ones everyone else has. that's why I say, release it to the public. let anyone see all the ballots.

it is true, that the GIFs could be tampered, prior to posting on the internet, with a Hursti-style hack. the machine could be loaded with fake GIF images and the results could be fraudulent.

so there would still have to be audits and open software. but I'm saying the release to the public is an add-on that makes Hand Counte Paper Ballots a reality.

if they make it public, many people will count them. if the public counts are different from the official results, the phone would be ringing off the hook and there would be an official recount.

[Wilms]

9. Yep. The 2nd and 3rd paragraph is what I was referring to. n/t

[Cos]

4. What does it solve?

I'm not against this idea, but it does seem like it would take a lot of effort, so I'm curious what problem you're proposing it as a solution to.

It doesn't actually substitute for an official handcount, of course, because it's harder to get verified results from a bunch of people counting at home than it is to just organize a group of volunteers to hand count at the polling place, with observers. It doesn't solve the computer security problem, because the only way we "know" those images posted on the Internet are the actual ballots is to trust the computers and the code running on them. It does make things more open, in some ways, but not necessarily more so than just posting the raw ballot data, even in plain text as a .csv file (which is actually what the city of Cambridge, MA, does for municipal elections - we use a hybrid of instant runoff voting and proportional representation, so the raw data is interesting to analyze).

So, define the problem more clearly, and then we can see if this can be part of a solution to it.

[garybeck]

7. you are correct

you still need audits and open software code.

I'm saying the public release of the ballots is an add on to the other security enhancements already being discussed. It will provide a way for people to hand count the paper ballots, if they want to.

[Steve A Play]

20. What about using just a 'stand alone' scanner

like the Canon DR 5020 which is already qualified by NASED as a ballot scanner for use in elections?

See this link, http://www.nased.org/ITA%20Information/NASEDQualifiedVotingSystems12.03to6.05.pdf

The Canon DR 5020 is used as part of the Avanti International 'Optical Vote Trakker' 1.5.0 (NASED # N-1-12-22-22-002)

Details on the scanner, http://www.usa.canon.com/cpr/pdf/Brochures/dr5020_spec_0304.pdf

The entire system would run on 'COTS' (Commercial Off The Shelf) software and generate both a CD-R of all ballots scanned and a .pdf document file with a watermark for the web. This is the same scanner used by BBV.org to scan documents like poll tapes for presentation on the web site.

http://www.bbvdocs.org/FL/volusia/Volusia-pct101-polltape.pdf

If you get a bank statement that has scanned images of your canceled checks, it's more than likely that this is the scanner that your bank uses. There is an optional attachment that can read 'magnetic' ink like that used on checks for automatically reading the account number. That device could be used to scan ballot serial numbers. Custom 'open source' machine vision software like Harri Hursti's could be used conduct a quick preliminary count from the scanned images that would be confirmed later by the hand count of the paper ballots.

Such a system would allow anyone to audit the hand count at any time while preserving the integrity of the original ballots.

[EFerrari]

8. How do I know whose file it is or if the contents are real?

I'm just a person with a computer.

[garybeck]

11. well

if it is released on a website, all you'd have to do is compare it to the one on the website and make sure it's the same. that can be done no problem.

the real problem is still in knowing that the GIF files actually represent the ballots. the scanners could be rigged, harri hursti style

that's why you still need open source, audits, etc. but adding free access to the scanned ballots is a good idea.

[EFerrari]

13. As a civilian, I have to say, it would go right over our heads. n/t

[GuvWurld]

10. I've heard a similar idea

The idea I heard was to have the paper ballots come across a conveyor with one or more high-speed digital cameras above making images of the ballots. Interested parties could get a CD of the ballot images. All CDs would have to be identical or different teams of counters would get different results. As many people or groups that wanted CDs could do their own count.

[CPMaz]

12. I like it, but would suggest one difference

have the voting machine print a copy of the completed ballot (serial numbered, no names) for the voter for them to take home with them. Later on, they would be able to visit a website with the gifs and make sure their vote was counted as cast.

Any variation gets a criminal investigation, with the results for all races/questions on the ballot on hold.

[garybeck]

14. I was just thinking something like that. check this out

there is a blank line on the ballot. the voter can write any letter or numbers they want on there. then when they view the ballots, they can search for theirs. only the voter knows their code. that would make it almost impossible to rig.

[Cos]

16. opens up another problem...

I have a concern about any method of verifying ballots based on letting the voter look up their own ballot later. It's similar to the concern I have about absentee ballots, as well, and why I favor more early voting in conjunction with restrictions on absentee voting (as long as early voting is accessible enough to make up the difference).

The problem is coerced voting, and paid voting. Let's say, for example, that your local organized crime syndicate pays you $10 to vote for their favored candidate for mayor, and threatens to break your store windows if you vote for the other candidate. You tell them sure, okay, you'll vote the way they want. How do they know you actually did so? If you vote normally at a polling place, they have to take your word for it. Your ballot is secret. So instead, they make you apply for an absentee ballot, fill it out, and show them your filled out ballot and envelope, which they either drop in the mail themselves or watch you drop in the mail. If we expanded early voting and restricted mail-in absentee voting, that'd be much more difficult.

But if we then allowed a voter to put a code on their ballot that they could look up later, anyone else could look up that same code. The people coercing you hand you a code, and you have to write it on your ballot. If they find a ballot with that code and the "right" vote, then you're safe. (Or, in a more benign example, then you get your money)

I want verified vote counting processes that preserve anonymous secret ballots as much as possible.

[galloglas]

17. No problem if...

1) All precincts use serially numbered ballots, packed 100 per envelope.

2) All precincts have no more than 1000 voters. Consequently, all ballots within that precinct will be number the same except for the last three digits.

3) Each voter signs poll book and draws a ballot, randomly, from the prepacked envelopes holding 100 ballots each, then goes to booth to fill it out.

4) Upon completion of ballot, it is taken to the ballot box and dropped through a scanner which creates a .jpg image, which is recorded to a dedicated hard drive at the precinct.
When 100 ballots are recorded, a CD is burned, thus securing the identity of the ballot, while the voter stays anonymous. When 100 ballots ballots are cast, the corresponding 100 .jpg images are sent directly to a dedicated hard drive array at the SoS office to be made visible to the public as soon as it is written to disk.

5) If the voter can, and wishes to, remember the last three digits of their ballot, they can confirm their vote has been counted as cast within hours.

6) With 1000 ballots per precinct, the number of ballots cast plus the number of ballots spoiled plus the number unused will always equal 1000. It retains the secrecy of the ballot while providing the security that all ballots are accounted for adding the total of all ballots casty, all spoiled, and all remaining unused.
Additionally, when each voter can identify their own ballot in the form of a visible and digitally copied .jpg of that ballots, it should be damned tough to do any cheating other than penny-ante retail fraud of "a drink of whiskey for your vote, sir!"

7) We checked this cost for Missouri and determined that we could store the next 100 years ballots, in digital form, for about $1200.
Seems to me that the use of digital images, rather than digital sums, with the requisite serial numbering would kill the fraud, since the digital images would have to reconcile with a precinct hand count of the ballots when the polls close. Any discrepancy would be evidence of fraud, no?

Any thoughts.

Phil

[Amaryllis]

19. Kip Humphrey came up with something very similar to this. There is an

old thread floating around somewhere on it...from 2005 sometime.

[Steve A Play]

15. Actually, I believe that several units already make .tiff or .jpeg images

of every ballot scanned. That's why Harri Hurst developed the VoteOScope software for BBV.org to count the votes using those existing images. http://www.bbvforums.org/forums/messages/1954/10268.html

The problem is that Diebold and Hart InterCivic whose equipment create the files containing the ballot images have claimed that the files are "proprietary information" and election officials have refused to release the files to the public. We're still working on that. So far they've offered to give us 'reformatted' files but we're refusing on the grounds that altered files can have the images changed.

[FogerRox]

18. leave my underwear out of vote counting, sir.

Being blinded by fury and lack of sleep-- sounds great.

[eomer]

21. One thing we need to add to this system is an indication

of how each ballot was officially counted that can be correlated with the individual ballot image.

As a vote for X, a vote for Y, an undervote, an overvote.

Without this feature you can't know how or whether any particular ballot was counted (which several posters want this system to afford them).

With this feature then you can both verify that the official sum is a true count of the official decision of each ballot and you can also understand and agree or disagree with the standard used to reach each individual ballot decision.

While I'm making suggestions, I believe that a comfortable cotton is best for an all-night counting marathon.