Go until the end, then stop.

[vakkotaur]

"Those who do not understand unix are condemned to reinvent it, poorly" -- Henry Spencer

Schmidt's Law: If you mess with something long enough, it will break.

The above lines are true for things other than unix. There have been a number of articles regarding the problems of voting machines and, indeed, almost any hidden electronic or even mechanical part to the voting process. Anything of the mechanism that is hidden is suspect. That which is hidden can be either quietly failing or have been set to commit fraud outright.

Various solutions have been suggested, from open source software to random checks to voter receipts. None quite manages to be satisfactory. Open source software takes time to be inspected. Random checks will not find every flaw - though they may scare manufacturers into being quite careful. Voter receipts are an invitation to disaster: "Show up with your receipt for my candidate and..."

What's needed, and even seen as needed, is a proper audit trail from vote cast to vote counted to election result. This must include the ability to re-count votes, and should be hard to spoof or at least make it easy to detect an attempt at illegally influencing election results. This audit trail has been called a paper trail, since that has worked in several places, even for things other than elections.

There are low-tech pen-and-paper ballots. These can slow to count, and to recount. They are bulky and take up space to store and are heavy things to ship. All that, however, makes them ideal. The bulk means it'd be hard for someone to 'stuff the ballot box' without detection. They've been around for some time. That means they're a proven system. They are their own paper trail, so there's no need to invent a new one. No voting machine parts or programming are needed that can be hidden.

Paper and ink ballots aren't perfect and there is still a possibility of them being manipulated. They do, however, tend to have a fairly large disadvantage for those who would tamper them: they require physical evidence be left behind or disposed of.

So they're a bit slow. Big deal. Presidential elections are held in early November, swearing in is done in late January. Each precinct does its own count. Counting all the votes is a big job, but it's distributed big job - which means its really many small jobs.

The problem of how to conduct a vote has been solved. There's no need to solve it again, poorly, and break a working system.

Comments

[michaelmink.livejournal.com]

R.A. Lewis was the long-time (and Pulitzer-winning) editorial cartoonist for the Milwaukee Journal from the early 30s to the late 60s. I have a collection of his cartoons, published pretty much at the end of his career.

One cartoon, published in 1960, shows some folks counting ballots in Milwaukee. Yep, Milwaukee used paper ballots as late as 1960. There are two weary counters, plus one person doing the old IIII scratch marks on a blackboard behind them. A clock behind them shows the time at 4.07 in the morning. Caption: "Great Age, Nixon 1, We Live In, Kennedy 1, Everything Mechanized." The book notes Milwaukee was still counting ballots the next day.

In 1950, the GOP candidate lost to Senator William Benton in a Senate race in Connecticut, by about 1,500 votes. The Senate was very closely divided at the time, so the seat was important. President Truman actually had federal marshals impound the ballot boxes in many rural jurisdictions that used paper ballots. The GOP candidate's name? Prescott Bush.

[vakkotaur.livejournal.com]

If it's between speed and convenience versus accuracy and recordkeeping, I'd figure accuracy to be important. The wolrd won't end if the counting takes a couple days. TV ratings might plummet, but that's not a voting problem.