Monday, November 13, 2006

Recount Update

I just received an email from the Secretary of State's Office Communications Director Daniel Tapper showing the current status of the 2nd District Congressional recount. Presently there is a roughly 62-vote difference, down from original results of about 170 votes in favor of Courtney.

Now please be aware that I've got a serious head cold that is lessening my cognizance of simple facts, so forgive me for any errors I make here. But just for the fun of it, I compared the list of 2nd District Towns in the Current Recount Totals against the list of 25 towns statewide that are using the new Diebold Optical Scanning voting tabulators, to see if any of them had been involved in the recount.

Out of the 65 or so towns in the 2nd, only nine have the new machines. And out of those nine towns, only Mansfield, Tolland, Vernon and Westbrook have completed their recounts as of late today.

Some interesting facts emerge. All four towns had changes made in their results. Now, Secretary Bysiewicz's office hasn't released the reasons yet for these changes, and they could range from simple transcription errors, uncounted provisional and absentee ballots (I think; are absentee ballots counted yet?), and other factors.

They can also have been caused by tabulating errors in the machines. We'll have to see.

But Mansfield ended up changing +2 for Simmons and +3 for Courtney; Tolland +1 for Simmons and 0 for Courtney; Vernon -2 for Simmons and -2 for Courtney; and Westbrook 0 for Simmons and +1 for Courtney.

So that shows miscounts for a total of 11 votes in four small towns using the OS machines. In an election where just under 250,000 votes were cast, and a 62-vote difference remains.

It's way to early to know what caused these descrepancies, but I will be in touch with the Sec. of State's office tomorrow to follow up on this, and at least ask them where the changes originated (miscounts, transcibing errors, machine errors, etc.) I also think if the election stays this close, the Secretary may want to have a manual recount of ALL the paper ballots in the district, to compare the results of those nine towns using the OS machines to the reported figures.

Not only is this a necessity, but it's also a great opportunity to see how the Secretary's office deals with a close election using the OS machines. Better to have this happen now than to have a future presidential election turn into a "Florida 2000" nightmare!

We all know how well THAT turned out!

9:25PM - MY LEFT NUTMEG's comments thread has newer unofficial info from local recounts; one town (Lyme) added +39 for Courtney, so Simmons lost a lot of the 100-vote misread he got earlier. Boy, this is exciting! And it's getting to be NyQuil time for me soon!

(more info as it develops; discuss while I slug down some more DayQuil)

UPDATE: Just received a final update from Daniel Tapper
"This will likely be the final update for the night. 29 towns have now reported to us and 3 6 still have not. The margin is now Courtney by 66 votes. In the latest town to come in since my last update to you ( Coventry ), Courtney gained 3 votes and Simmons lost 1, which brings the margin to 121,221 to 121,155 in favor of Courtney."
And just for fun yet again, here's the awesome video I made of the new OS machines in action, at the Bridgeport City Hall annex in early September when Susan Byseiwicz visited for a demonstration:

21 comments:

CT Bob said...

The latest update is what has me worried. If only 29 towns have reported while 36 haven't, and in those 29 towns Simmons made up well over 1/2 the votes needed, Courtney's seat is FAR from assured!

Another "100-vote transcribing error" in any of those remaining 36 towns will eradicate Courtney's lead, and probably give the seat back to Simmons.

Oh well, we didn't think this would be EASY now, did we?

carterman said...

How are the recounts done in the towns with the new machines? Are they run back through the scanners or are they counted by hand? Any idea how this is done?

CT Bob said...

Carter, the SoS said that on 20% of all the OS machines where will be a manual random audit. Which means hand counting the ballots and comparing them to the results the machine reports.

I'd imagine that these small towns wouldn't take a lot to do a 100% manual recount of the paper ballots. Just ONE screwy machine could toss the election to Rob. Each machine can handle thousands of ballots during the election, so 1/2 of 1% would be about enough to screw the pooch.

I'll contact the SoS's office tomorrow and ask for details on how they're doing the recounts for the OS machine towns.

Anonymous said...

Courtney will end ahead. This is whole exercise should be a rallying cry to the citizens of CT for mandatory election reform.

CT Bob said...

Patience, I hope he does. What kind of reforms would you suggest?

Gabe said...

Hey Bob - This is a recanvass, not a recount. That means that the optical scan bubble sheets are just being run through a machine (my understanding is that it is not a machine that was in use on election day). The most likely cause of changes in those towns (besides differences in counting the absentee ballots - and the number is small enough for it to be that) is not a tabulating error in the machines, but a difference in sensitivity between the machine used on election night and the one used now (of course, there can also be a sensitivity difference between the same machine two different times).

The 25% hand recount of the paper ballots is state wide and is seperate from the recanvass that is going on in the 2nd. Whichever campaign is down at the end of the recanvass (Simmons, fingers crossed) can ask for a manual hand count of all the paper ballots in addition to a review of all contested paper ballots in front of a three-justice panel of the SC. This would be similiar to the Florida 2000 fiasco in that the justices would be attempting to divine the intent of the voter on contested ballots. That said, we do it with SC justices and, hopefully, without the screaming hordes of RNC staffers that showed up in Florida.

Anonymous said...

Marlborough is a wash- no changes there tonight.

CT Bob said...

Gabe, thanks for your input. I figured you'd be aware of the details given your position.

So they're using a separate machine to recount the paper ballots? I recall something about how each town would get 2 OS machines per polling place, plus at least one extra machine per town for emergency use. Is that the machine that's being used in each town's recount, or is it a "state-supplied" machine?

And I agree about the very low descrepancies; if the machines were crooked, they'd be turning out higher percentages, probably on the order of at least several percent per machine.

However, you never know if one of the scanning sensors got a spec of lint or something on it so it isn't 100% accurate. Since the machine allows partial votes (say you vote for Gov. but not Sen., that's OK as far as the machine is concerned) it conceivably can miss a few votes while appearing to function properly.

At least, it's possible; I don't know enough about the sensors to know if a bit of debris can be a problem.

Gabe said...

Every town has a spare - I believe that is what is being used for the recanvass.

Also, I could be wrong about this, but I thought the machines would spit out an undervote and thenpoll worker would have to put in an override in order to vote with an uncompleted ballot. Again, I could be totally off on that.

Finally, I don't think debris is as much of an issue as is the scanners just not having the exact same calibrations. For what it is worth.

Anonymous said...

you like howard dean?!?!?!?!? hes a total nutcase and his rival mehlman is the definition of tool (just throwing mehlman in there so you dont think im a republican).

CT Bob said...

Gabe, the machine will only spit out a ballot if there is NO vote at all anywhere, or there are two or more votes for the same office. You can have as little as ONE vote for ONE office, and it'll count the vote.

But just like the lever machines, you don't HAVE to vote for every office on the ballot.

Otherwise I would have been forced to vote for Jim Amann.

CT Bob said...

Upon futher reflection (now that Blogger has started working again) I won't say that I'm 100% sure about that, but I think I recall Susan saying "if at least ONE candidate is selected," the machine will accept the ballot, even if you don't vote for any others.

Gabe said...

Gotcha.

Bill Dauphin said...

I can confirm that the machines will accept "uncompleted" ballots: I recorded the results of one of the Vernon polling places on election day, and there were at least a few "blanks" reported in each race.

Anonymous said...

Thanks to CT for Joe!

CT Bob said...

Bill, thanks for the input.

Anonymous said...

b.a.,

Computer Security expert Bruce Schneier's commentary on the technical problems and solutions is a great place to start.

Three of his more recent entries are here

The Inherent Inaccuracy of Voting
http://www.schneier.com/blog/archives/2006/11/the_inherent_in.html

More on Electronic Voting Machines
http://www.schneier.com/blog/archives/2006/11/more_on_electro.html

How to Steal an Election
http://www.schneier.com/blog/archives/2006/11/how_to_steal_an.html

The best solution that I have seen , and the one I strongly advocate, which makes voting transparently verifiable and hard to mess with is this proposal by Ron Rivest of RSA security. He uses cryptographically mathematics principles (advanced number theory) to design the following.

Abstract:

We present a new paper-based voting method with attractive security properties. Not only can each voter verify that her vote is recorded as she intended, but she gets a “receipt” that she can take home that can be used later to verify that her vote is actually included in the final tally. Her receipt, however, does not allow her to prove to anyone else how she voted.

The new voting system is in some ways similar to recent cryptographic voting system proposals, but it achieves very nearly the same objectives without using any cryptography at all. Its principles are simple and easy to understand.

In this “ThreeBallot” voting system, each voter casts three paper ballots (with certain restrictions on how they may be filled out, so the tallying works). These paper ballots are of course “voter-verifiable.” All ballots cast are scanned and published on a web site, so anyone may correctly compute the election result.

A voter receives a copy of one of her ballots as her “receipt”, which she may take home. Only the voter knows which ballot she copied for her receipt. The voter is unable to use her receipt to prove how she voted or to sell her vote, as the receipt doesn’t reveal how she voted.

A voter can check that the web site contains a ballot matching her receipt. Deletion or modification of ballots is thus detectable; so the integrity of the election is verifiable.

The method can be implemented in a quite practical manner, although further refinements to improve usability would be nice.

The Three Ballot Voting System
http://theory.csail.mit.edu/~rivest/Rivest-TheThreeBallotVotingSystem.pdf

Clearly Rivest's proposal or some variation on this approach will make both verification and our confidence in that verification much higher. A mathematical contest for professionals, much like the one that generated the new AES algorithm that currently protects high government secrets and inter bank transfers would come up with a dozen or more ideas along the lines of Rivest's proposal and provide much needed public scrutiny to the whole design process.


As for the architecture and system design of voting machines, Berkley's David Wagner and Princeton's Ed Felten have come up with these very sensible recommendations

Building a Better Voting Machine

http://www.wired.com/news/politics/evote/1,71957-0.html


Once a better system is in place, I think close elections, as determined by each state's definition of close, should automatically cause a brief re-campaign and revote with the intention of getting a clearer majority for one side or the other. Whether this is would occur at (3%, 1% or 0.1%) it would do us all good to see close elections get more direct public attention during which the public can more participatorially determine the outcome rather than handing off the results to the bowels of the voting system as it does currently.

Anonymous said...

Bob,
You rock for being on this. Waiting to hear . . .

Kirby said...

It's posts like this that make be proud to be here at CTBob. Fabulous reporting, Bob!

CT Bob said...

Patience, I reviewed those web sites you listed, and do have some thoughts.

While the proposed voting system you listed appears very secure, it's hackable just like any other system. There is no way possible to create a voting system that is unhackable, as long as some people know the programming code, and the majority of the voters don't.

The articles you linked seems to agree that the paper-ballot OS machines are as close to secure as possible. For the simple reason that every single vote exists on a paper ballot that the voter him/herself filled out. It's the only real paper trail; the receipt producing machines can also be hacked/programmed/patched etc. to show results that don't reflect the way the person voted.

And without that original paper ballot, you're depending solely on something electronic to determine if and how your vote was tabulated.

And systemic things like polling places not opening on time, poll workers being barred from doing their jobs, etc. are part of a larger issus that I'm not addressing here. As far as the actual voting and tabulating of the votes go, I think we have a reasonable system.

Not that they're perfectly safe; I wrote extensively using info from Black Box Voting and Brad's Blog showing that the very model of Diebold OS Scanner could be easily hacked in minutes and a replacement memory card installed.

But even with that being said, the paper ballots are 100% verifiable evidence of how people voted. If a machine reports totals way out of whack with other machines or exit polls, the Sec. of State has the authority to demand a manual recount of all that machine's votes.

And any significant errors in that machine's tabulating would probably trigger a more massive recount.

At least the procedure is in place to ensure we'll get a reasonably secure election. But we'll never have a perfect system; hey, Democracy itself is an imperfect system, but we somehow get along with it OK.

CT Bob said...

Thanks, Kirby! :)