Except that, from what I have read, a recount using optical scan sheets involves feeding them back through the scanners, and only of the recount is still within a small enough percentage are hand recounts done.

So the recount can be hacked as easily as the count.

Much of the problem here arisis from an elections "industry" and elections "professionals". They are exclusively responsible to the bi-partisan party of perpetual incumbents. That party is mostly interested in stopping "recounts" and "contests".

These lay bare little fraud but a lot of shabbiness that connotes a real contempt for political participation as distinct from political patronage.

::JRBehrman

It isn't a "bipartisan" problem here in DuPage County Illinois where Republicans control everything including the election commission. We have Diebold TSX's and OptiScan tabulators. In response to a FOIA the exec director claimed he couldn't turn over election materials because they'd been destroyed within about 6 weeks of the 2004 election even though he was served with the FOIA immediately after the election. When informed that he'd broken both Illinois and US law he said a different Illinois law supercedes those laws. He then changed his story and said the materials still exist but remained sealed and can only be examined in very expensive recount.

There's so much hinky going on out here
if anybody wanted a perfect test case to investigate (God knows the local media won't do it) DuPage would be a great place to start. Proving vote fraud will be hard to impossible anywhere because that's the real reason behind electronic voting: making stolen elections impossible to detect. But in Dupage they're so arrogant they've outright lied about the wonders of these machines and those lies can be easily refuted.

You assume that there is a zero error rate with paper X ballots and manual counting. This is not the case - the error rates in vote counting are similar to those in any high-volume mechanical process involving human vision and keying.

"To err is human, but it takes a computer to really foul things up...." A cute joke from the 1970's, but it's funny because it's allegorically true.

1) (durable) Paper ballots and unaided-human counting are subject to public review, audit and recounting. As with checks, paper ballots can be subjected to audit procedures to catch and rectify counting errors.

2) To steal an entire election recorded on paper ballots requires the "assistance" of a much larger number of election-thieves who must be genuine insiders, and every additional involved person is another chance to discover dastardly deeds. A well-placed voting-machine easter egg (sold to the highest bidder by a dishonest developer) could be used in multiple precincts by late-afternoon "voters" who need only the right information from the developer, and these "voters" need not be insiders.

3) All but the most fundamentalist hand-counted-paper-ballot people believe that optical scanning is an acceptable way to produce the equivalent of "immediate returns" as an unofficial count. By and large, it is even possible to use optical scan systems for official vote counting, subject to verification using well-designed statistical audit methods,

4) Unmentioned in most of the debates about the value, security, reliability of electronic election systems is the infrequency of their use: Usually twice per year, rarely more often. This is a recipe for experiencing reliability problems due both to operator unfamiliarity and startup problems.

5) Unmentioned in most of the debates about electronic election systems is the phenomenon of "strategic allocation", wherein some precincts receive plenty of machines while other precincts do not. This technique was used to some effect in Ohio in 2004, where the turnout was large and the ballot complicated. Waiting times in certain precincts rarely exceeded 20 minutes, while waiting times in other precincts rarely fell below 2 hours. (I watched this occur by direct observation, being in a precinct that had received inadequate equipment.)

6) Cost: Storage, maintenance, software updates and similar costs turn out to be much higher than original estimates provided by the vendors.

To sum up:

Paper ballots and hand counting impose far greater risk and cost upon those who would attempt election-manipulation than is possible with electronic systems, particularly touchscreen systems, where the risks of detection are far lower.

Paper ballots marked with pencils are countable in public, and subject to public audit without use of specialized equipment

Paper ballots are not subject to reliability issues with power, batteries, or screens

Paper ballots can be printed and managed under reasonable security precautions only after printing, while electronic election systems require essentially continuous security, and require much more environmental control in storage.

Paper ballots, marked with pencils are intrinsically parallelizable, and therefore are much more scalable for high turnout, high-complexity elections.

I support the use of touchscreen ballot-printers as an assistive technology to help those who would benefit from the accessibility that touchscreen systems can provide.

Simply put, the durable paper ballot is the most easily secured, scalable and auditable technology presently available.

We use optical scanners at my polling place. It creates a paper record. Even if the system is hacked, you could still re-count the paper ballots by hand count.

This would be a problem for incompetently designed statistical audit procedures. Anyone with a basic grounding in modern audit methodology can design and implement competent statistical auditing procedures.

Electronic recorded voting systems present a set of technical challenges that would either render them *impossible* to secure, or so expensive as to be infeasible.

(For a wonderful talk by a credible developer, see

Reflections on Trusting Trust by Ken Thompson.

Also see nearly anything written by Bruce Schneier with Counterpane Systems, about the subtlety of computer system security.

To quote a response to a criticism of a technical note on high-availability in Communications of the ACM: "Your alternate solution assumes that the channel that signals errors is itself error-free. If this is the case, there is no need for error correction and detection: simply use the signaling channel for all your communication. Unfortunately in practice it cannot be assumed that the signaling channel is error free".

There is no assurance that a recount is error free either.

I would trust a voting system designed by a team headed by Peter Neumann, Bruce Schneier, and Linus Torvalds. I would be OK with a voting system designed by less devious-but-trustworthy minds if it were Open Source hardware and software and included paper-backed audit facilities. But we ain't gonna get either: too much money and political power is at state for the big guys (Radical Right, Republican, and Democratic; federal, state, and local alike) to allow that.

sPh

Sorry, the *reprint* was in 1988... So no, I didn't read it. Though no American can fail to be aware of the depth of corruption in midcentury Chicago. Given the depth and breadth of corruption then, the presence of easily corrupted hardware/software systems would only have been worse. So I don't consider that an argument in favor of electronic voting systems. (It's a *very* good argument for transparency, though!!)

I should not comment on the unread book, but I'll respond generally by noting that Chicago in that period was characterized by deeply corrupt machine politics, which corruption was supported by multiple governments at multiple levels, every one of them so corrupt that they stand as the prime example of corruption in recent history. Lots of heavy duty organized crime, too...

My claim, among many others, is only that electronic voting systems provide evil-doers with easier, less expensive ways for their misdeeds than are required with paper ballots. My basic thesis is only that short of spending waay too much money to secure them, electronic voting systems make election fraud easier than paper ballots.

Do you think it makes any sense to consider "better vs worse"? I tried not to say "paper ballots are secure, while machines are not". I apologize for not being more clear. It's all relative, and in my opinion, paper comes out ahead because it's more transparent to review and audit than electronic systems.

Sorry about the link-closing error in my previous post.

sPh

I've discussed ballot processes with Canadian and British friends, who are mystified why the US has such a compulsion to use technology. One suggestion is that it is due to media pressure to get reportable results.

When I asked one Canadian friend if paper ballots would work only in smaller electoral districts (e.g., Parliamentary ridings), she observed that if the districts are too large, either it's time to split them, or to get more volunteer counters and judges. Either way, she saw that as a way to get more people involved in the mechanics of voting, which she saw as a very good thing.

--
Howard

*equal opportunity offense to both extremes*

"Those who cannot remember the past are condemned to repeat it" [George Santayana]

The reader that gave you a one was unduely harsh. What you suggest is not the best system, but it is good, and way ahead of all the proprietary OS voting systems out there.

Kevin Russell Cook