Just like DRM [LWN.net]

Just like DRM

Posted Aug 9, 2007 9:19 UTC (Thu) by charris (subscriber, #13263) [Link]

IIRC, wholesale move to electronic voting started with the butterfly ballots in Florida in the 2000 election. A prime example of the need to be careful of what you wish for. That said, I voted using a Diebold machine in 2006 and it kept a record of the votes on a strip of paper that I could see, so there was a paper trail as well as the electronic memory. I suspect the big problem isn't the voting itself, it is the ease of jigging the machine to create false votes. How easy it is to jig the machines in practice, I don't know. During the last election cycle there were three recounts in the Washington governor's race and lost ballots kept turning up in King's County until the election results changed. So it isn't clear to me that paper ballots are really better than the machines.

Hanging chads

Posted Aug 9, 2007 10:31 UTC (Thu) by nowster (subscriber, #67) [Link]

Weren't the ballots in 2000 election machine-counted initially? They were on punched cards.

Hanging chads

Posted Aug 11, 2007 3:57 UTC (Sat) by giraffedata (subscriber, #1954) [Link]

Virtually all ballots in US elections have been counted by machine for at least 20 years. Some of those machines use more electronics than others. The exceptions have been rural areas where hand counting is cheaper, but those areas are now tending toward mail-in ballots counted by machine.

Hand counting has been a backup for when the machine fails, and various jurisdictions have various rules on what is adequate evidence of a machine failure. The legal controversy in the 2000 Florida presidential election had to do with interpreting Florida's vague rules in that regard (Florida isn't special -- almost all states had vague rules).

Since ultimately none of the counting issues affected that famous election (while a full hand recount was found to be not legally required, a bunch of journalists did one anyhow and found that no matter how you counted the ambiguous votes, George Bush won), and the only real unfairness was the misunderstood Palm Beach County butterfly ballot (if voters had voted for the candidate they meant to vote for, George Bush would have lost), I think the GUI makes a lot of sense. It's a more powerful medium than any paper ballot can be.

Hanging chads

Posted Aug 17, 2007 8:02 UTC (Fri) by forthy (guest, #1525) [Link]

while a full hand recount was found to be not legally required, a bunch of journalists did one anyhow and found that no matter how you counted the ambiguous votes, George Bush won

No, you remember incorrectly. They found that if you only recounted the four districts Al Gore wanted to have recounted, George Bush still won. But if you recounted all Florida, George Bush would have lost. Most of this just went under, since the publication was short after 9/11, and due to the "many scenario" theme in the article, the conclusion was not obvious (it also depended on the standard of "voters intent"). See Wikipedia.

Unfortunately, the Times article cited there is only available to subscribers. But it's certainly wrong that "no matter how you recounted, George Bush won". You can however say that it was too close to call, because due to the uncertainty with the recounting, you still didn't get a convincing result. If you can't get a winner, the rule "the winner takes it all" shouldn't apply, and if you can't do so, because your election process has a way too high error margin, even less so.

IMHO, if this recount had been published a month before 9/11, the conclusion would have been different, and George Bush would have had a snowball's chance in hell to stay president.

Hanging chads

Posted Aug 18, 2007 18:37 UTC (Sat) by giraffedata (subscriber, #1954) [Link]

They found that if you only recounted the four districts Al Gore wanted to have recounted, George Bush still won. But if you recounted all Florida, George Bush would have lost.

Thanks. I stand corrected. But looking at the free abstract of that article, I see it says Bush could have lost; i.e. it depends on your rules for counting ambiguous ballots. I found another source that says even in the 4-district recount, if you interpreted "overvotes" a certain way, Gore could win. (An overvote is where someone votes for two candidates; I don't know what the proposed methods for disambiguating those are).

IMHO, if this recount had been published a month before 9/11, the conclusion would have been different, and George Bush would have had a snowball's chance in hell to stay president.

So you think somehow Bush's presidency would end just because 50% of the people in Florida didn't vote for him? Many people, including the US Supreme Court, thought the actual count was not as important as the count defined by the rules of the game. And that it's no great travesty of democracy if there is a .05% error in the voting.