Two comments.
(1) The ability to verify does not, IIRC, guarantee you can prove (and thus be pressured). There are papers that study this topic, some using cryptography, some much simpler (but none very simple). IIRC, in one such scheme you cast three ballots, which sum (e.g. by exclusive-or) to your desired ballot; you get a receipt only for one of the three. Further details left as exercise. 
As a practical matter, the proposed solutions may all seem too complicated to deploy.
(2) Absentee ballots, at least as used in U.S.A., already lead to the coercion problem. (Is there evidence showing this to be a problem in U.S.A.?)
One idea is to keep track of when an entry in the database was accessed. If I check my number and see that someone else has accessed it then I know there is something up. Even if only a relatively small number of people checked, the law of large numbers makes it likely that you’ll detect fraud. Right now we know something is fishy in SC, but no-one can come out and say that their vote was recorded incorrectly.
Voter: “Someone call the FEC! My serial was accessed two days ago! I didn’t do that!”
IT Nerd: “Says here it was done from the same IP address you checked it from ten minutes ago.”
Voter: “Oops. I guess I did check it. I forgot. Sorry.”
Now play this conversation out a couple thousand times and you’ve got a lot of WOLF! cries.
-Joe
Oh, this is cute. They just had to overturn an election in Lee County (the South Carolina county where the sheriff just got busted for his involvement in the drug business) because they got 107% turnout.
See, this is why we’re better than Saddam Hussein! He only ever got 100% to vote for him!
See what happens when you have a Democrat president?
-Joe
The simplest record method is that you get a printed receipt that you can verify matches your intended vote, but without your name on it. Then you drop that receipt into a box before leaving the polling place. This would be just like the manual voting cards we used prior to the machines.
If there is any funky number business, then the receipts can be counted by hand. That’s a manual recount.
If you really want to have fun, print the receipts as scantron forms.
Then there is a recountable record that does not have the voter’s ID associated, but the voter checks before submitting, that is independent of the electronic record in the machines and any code finagling and whatnot.
I like this idea.
That sounds really good
Diebold no longer makes voting machines. That part of its business was spun off into a separate company called Premier Election Solutionsin 2007, and that in turn was bought another company called Election Systems & Softwarein 2009. So you don’t have to be scared of the Diebold boogeyman anymore.
The bigger part of Diebold’s business, by far, is ATM machines.
http://www.votersunite.org/info/ES&Sinthenews.pdf Yep ES & S are just so different. Half owner was a Diebold VP. Nope, no relationship at all. With the same kind of results.
Dan Rather Investigates Voting Machines -- Uncovers New Surprises About ES&S Touch-Screens | WIRED More on ES&S . they don’t test the machines.
Fine, just don’t talk about Diebold voting machines anymore. Talk about ES & S.
http://www.bradblog.com/?p=7906 Totally inadequate response. The voter machine companies are p[laying musical chairs and who knows who actually owns what? But our machines are now mostly in hands of Canadian and Venezuelan companies. Can that be justified? We are talking about election credibility. The machines should not be in private hands.
I don’t understand what you are describing here. If I get one of my three receipts (one of which is wrong), what use is that? The point is that if I can leave the polling place and verify that my vote was tallied correctly – by whatever method – then I could be forced to demonstrate that for someone else.
Also consider that ballots include many elections. It is not just the presidential or senatorial election that can be coerced. An anonymizing scheme needs to take all of the elections into account.
Absentee ballots are a small fraction of total ballots. Just because a small percentage of ballots are open to coercion doesn’t mean that we should open up all ballots.
Furthermore, voting absentee requires advanced planning and may have restrictions (there seem to be fewer restrictions in recent years).
There is a big difference between your boss saying, “You voted today? OK, show me your receipt and prove you didn’t vote for the black guy” vs. “I need you all to register for an absentee ballot in the next 6 weeks, fill it out, and then bring it into work so that I can mail it for you.”
Or some guy hanging outside the polling place, “Hey, you want to make five bucks? Bring out your receipt and show me you voted for Greene.” vs. a guy going door-to-door months before an election: “Want to make five bucks? Here is what I need you to do…”
That’s why I like the suggestion of a receipt that you verify for accuracy and then deposit in a lockbox before leaving the polling facility. That way you can verify that your vote was recorded correctly, and the receipts can be counted by hand to verify the machine tally.
Agreed. In this method, the computer voting system simplifies the process of filling out the form. Everything on the back-end is the same.
OK, this has to be heard to be believed.
Seriously, seriously, how different is this from the early Palin interviews?
It’s a difference of degree, but a weird one. At least in the Palin interviews, she’d respond to questions. This guy doesn’t even seem to know what an interview is…