The Volokh Conspiracy

This finally, I hope, will bring up the undiscussed tension between providing a paper trail and the concept of a secret ballot. Even if this particular problem is solved, if I get a piece of paper indicating how I voted, I can sell my vote and prove how I voted or I can be coerced into a particular vote and the coercer can get confirmation of that vote. Moreover, to make a paper trail meaningful, there must be some method to track a paper vote record to a electronically recorded vote. Unless carefully managed, this requirement leads to the problem discussed in this article. I have heard no discussion of this conflict in the media or by the opponents of electronic voting. It is a non-trivial issue and may, in fact, have no solution that preserves privacy, security, and tracability.

deweber: My guess is that cryptographers can come up with something to save us from this problem. Here is one attempted solution. At the very least, we can make selling votes require a conspiracy among a significant number of users of a single voting machine.

I've been involved in some of these electronic voting projects. I attended the Caltech-MIT voting conferences. I completely agree that privacy is a more severe and likely problem that vote alteration.

Even without paper trail voting, it is usually possible to determine who voted for whom simply because of correlations in how the machines keep their records.

In my state of birth, voting machines were large mechanical contraptions: you pulled the lever, they counted your vote. Individual ballots were never preserved. In my opinion, this is the only acceptable way.

Punchcards and scancards are scandalous creations: both retain finger prints for one thing.

I was born and raised in Chicago. I personally was involved in "aiding" people to vote for the Democratic candidate. When you hand a voter a slip showing exactly for whom he voted my experience leads me to believe that there will be a precinct captain waiting outside the voting booth to accept that proof and pay the voter. We did it before, why should we not repeat our successes. What bothers me is many will decry this illegality (wink-wink) but say it doesn't really exist.

Same basic topic, voting, when will begin to use the technology already available hand print, finger print, purple finger etc to make certain no one votes more than once and dead men don't vote?

deweber, You should not keep your paper record trail. This shit isn't that hard people.

You vote if you want electric counting fine, you get your data card and you get a piece of paper. The card has you vote on file electronically, your piece of paper says who you voted for. You put you paper in a locked box, put your card with the cards.

You randomly check districts to make sure the paper count matches the machine count. If it does you certify the electric count state wide. If it doesn't you retest another sample, If it matches you write off the first as an anomaly if it doesn't you count all the paper ballets. Use that count and sue the electric voting machine company.

Nevins: check the penumbras.

The basic idea for how you do this is as follows:

1) The voting machine shows you an ID number. You are welcome to write it down or not if you want.

2) You select a candidate you wish to vote for.

3) If you wish, you may take a receipt with the ID number and the candidate on it. The receipt uses a digital signature to be tamper-proof.

4) If you wish, you may also take many other receipts with other people's ID numbers and votes on them.

5) When the final results are released, they include every single ID number that voted for each candidate. If anyone can present a signed receipt that wasn't listed, they know their vote wasn't counted.

6) Monitoring organizations can, if they want, gather outside polling stations and ask voters if they can scan their cards. Votes will also be made available electronically at each local polling area roughly as they come in.

I've left out a lot of details, but this should give you the idea of how you can convince yourself that your vote was counted, and with your cooperation, another organization can make sure your vote was properly counted, even though nobody can force you to vote a particular way or reveal how you voted.

I wonder if any court would accept a voting scheme that admittedly seeks to ascertain the winner "within the margin of error." I know the constitutional text is different, but haven't courts rejected the use of statistical sampling (as opposed to an actual count) in the Census?

The secret ballot was a 20th century development. In the mid to late 19th Century (at least in Iowa) the political parties printed the ballots with their slate of candidates listed, or a voter could get a ballot by cutting it out of the newspaper. The voter would put a ballot with his preferred slate in the box (generally the opposing candidates were not even listed). The party observers could see how an individual voted by the color or shape of the ballot the voter put in the box.

GREAT! The 20th century worked out so well, politics-wise...

I believe it's absolutely essential that we all have proof of who we voted for. In a secret ballot how would we collectively know the if vote was manipulated? Should we trust the government? I think David Schwartz hit it on the mark. We're already dealing with a vote buying scheme, its called entitlements.

Actually, the concept behind a paper trail is NOT that you take it with you (as you mention, that leads to selling votes) but a paper trail that you can verify (behind glass) after you've voted, and which can be used, if necessary, for a manual recount. A timestamp (or even the sequence of ballots) would THEORETICALLY allow someone to determine who cast which vote, but it's actually far more secure than mail-in ballots, which are sadly already in wide use.

The method the article describes is not as accurate as the article suggests. They assume the first person to sign the register is the first person to start and finish voting. That is not always true. Two voting machings may open at the same time and the person who signed in second may actually start and finish voting before the person who signed in first. The article and this problem in their methodology was discussed in Slashdot.org a couple of weeks ago.

and insert buy before "barrels of rum."

We could come pretty close to a secret and assuredly honest ballot for a single office by going even more low-tech than paper: Give each voter one marble. Inside the voting booth, there are several cans with a funnel-top that you drop the marble into, each can marked with the name and picture of a candidate. The cans of course are sealed for the election, with the top designed so it is not possible to remove a marble other than by breaking the seals and opening it.

And one other thing: you've got to vote naked so they can verify you aren't smuggling in extra marbles...

I think it was Bruce Schneier that wrote some time ago about the incompatibility between secrecy and verifiability. If you can give up secrecy, you can have quite simple systems where the accuracy of the count can be 100% verified. If you must have perfect secrecy, it's doubtful that there's any foolproof scheme that could be implemented ensures honesty in the count. (If you haven't figured it out yet, my previous "vote naked" post was a joke - besides being politically impossible to bring it about, it doesn't solve the problem of ballots being substituted after the election.)

So, when is the secret ballot actually necessary? It does frustrate vote-buying and voter-intimidation schemes by making it impossible to verify compliance, but under what conditions could such schemes be as big a problem as the forms of election fraud that open balloting would prevent?

Look at the other side: It seems like many American, perhaps even a majority, do believe that at least one Presidential election hung on voting fraud or errors within my lifetime, and that attempts to fraudulently affect elections are on-going: Republicans believe that the 1960 election was stolen in Chicago and that Democrats scheme to allow felons and aliens to vote illegally while excluding military absentee ballots, while Democrats believe that Gore lost in 2000 by uncounted punch card ballots in Miami Beach, and Kerry was sabotaged in Ohio by Diebold voting machines. Close to half or Oregon believes that their governor won by massive fraud. And pretty much everyone who's at all familiar with Chicago politics believes that elections there are rigged.

Almost all of these issues could be alleviated by open balloting, where the voters walk out with a receipt showing their votes. If there's a dispute about counting votes, pick a random sample, go find the voters and compare the records in the voting system with their receipts or how they say they voted. If you think dead people voted, follow up on those votes - and if you can confirm the voter wasn't qualified, there's a record that can be used to exclude these ballots from the count. The one thing that would be difficult to fix is Chicago, because the local authorities wouldn't allow a real investigation if it were in their power to stop it, but the FBI should be able to investigate if they messed with a federal investigation.