More E-voting Problems in California

thefultonhow writes "Wired News is running a story about Napa County, CA's problems with their new E-voting system. Not only did an optical scanning machine fail to record absentee ballots properly, necessitating a recount of 13,000 ballots, but now Registrar of Voters John Tuteur is saying that the machine used in precincts failed to count 6,692 votes. The incumbent Napa County Supervisor had originally lost his bid for reelection by only 50 votes (the recount of absentee ballots bumped that up to 107 votes), so with nearly 7,000 votes gone AWOL, this is a big deal." The first Wired link above shows that the discovery of the problem was apparently mostly chance: if none of the 10 (ten!) ballots picked for rescanning had exhibited the problem, they might not have figured it out. It also suggests a new strategy for rigging the vote: pass out pens of a certain type in districts unfavorable to your candidate, then calibrate the machine not to read that type of ink.

Comments

What does this have to do with 'e-voting'?

[goldspider]

This is a story about machines not being able to read a certain kind of ink.

The fact that the ballots are 'counted' by a machine doesn't make this an "e-voting" story.

This problem has been around for YEARS! Nothing to see here, folks. Take off your tinfoil hats and move along.

This is a good argument for punch-hole voting...

[rthille]

And if you want to bring it into the new millenium, then put a touch screen kiosk in there with a 'printer' which after you make your selections, it punches the holes for you and spits the ballot out. You then review it, put it in the privacy sleeve and walk it to the ballot box. Or you feed it back into the 'printer', where it's destroyed and you try again.

Why is this concept so hard?

Keep it simple

[twinpot]

Like a few others, I can't see where the win is with electronic voting.

Keep it simple: paper and marker pen. Used in many countries, simple to understand, no real hardware required, biggest equipment failure is a pen not working, no hanging chads.

In the New Zealand elections I've voted in it's really easy - a check mark next to the local candidate and another next to the party use. Simple. Results are known a few hours after polling closes. Easy to do recounts, even without any fancy technology, scales easily too.

If speed is the real issue, then vote using the paper and pen, then count them electronically. Count them twice using two different machines, and if the amounts are out by some pre-determined margin, then hand-count. That way you get quick results, while having no reliance on any complicated, error prone bit of technology. You can still recount manually if required.

Re:Keep it simple

[TrollBridge]

"Keep it simple: paper and marker pen."

RTFA - this IS about paper and pen ballots, and a machine's inability to properly record it. This has NOTHING to do with people voting on computers.

Re:Technology is not always the answer

[Hiro Antagonist]

No, E-Voting is a big problem looking for smaller problems to forcibly mate with. At least, E-Voting without a paper trail is.

Despite hanging chads or bad ink, giving a receipt to the voter, along with keeping a paper copy for the polls, is the only way to insure that voting is handled properly. Of course, the Diebold machines don't produce a paper trail, primarily because it's harder to stage a respectable coup that way (rimshot!).

On top of that, most, if not all, of the commercial voting systems are needlessly complex, and have insane operating procedures -- they suffer from a horrible case of Second-System Effect, and it shows in how inaccurate they are.

All you need is a simple storage device, a receipt printer, and a bin to collect hard-copies of said receipts should a recount be needed. Some simple encryption on top of that should keep the data reasonably secure, and a bit of random sampling out of the pile of receipts can be used to ensure that the electronic copies of the votes are, in fact, good.

A competent coder could write this system in a day, and then a team of coders could spend a month pouring over it to make sure that the code is good. Open-sourcing said code so that programmers in the general population could find additional problems would be even better, and the government could offer a small reward for those that track down and report bugs. Sort of like software bounty-hunting.

On the flip side, this wouldn't jive with Diebold, because the CEO has already promised the next election to His Favorite Candidate. And, before the Rabid Right flames me for being pissed at the CEO of Diebold, remember -- I'm mad because it's a conflict of interest, not because he's a Republican. I'd be screaming murder if the Democrats, who I don't like much either, tried to pull the same crap.

Re:Maybe some attention

[subjectstorm]

in the vein of easily auditable results -

one thing that's always bothered me about these machines is that, if they can't make a vote out (can't even see the ink maybe?), they tend to simply discard it. in this case, there were nearly 7K ballots that were not counted (if i understand the article).

why couldn't the database on the back end be configured to flag any ballots that seemed irregular for inspection? for instance, if the counting machine recorded ballot #41768 as being entirely blank, this could be flagged and brought to the attention of poll workers, who would then read the ballot and adjust the results accordingly (under intense supervision, with data noting who did the changes, and a saved copy of the original results).

this requires a paper-based ballot system. but totally electronic voting with no hard copies is a really bad idea in the first place

Re:Maybe some attention

[Ungrounded Lightning]

why couldn't the database on the back end be configured to flag any ballots that seemed irregular for inspection? for instance, if the counting machine recorded ballot #41768 as being entirely blank, this could be flagged [...]

Better yet - why is the mark read as "yes/no" rather than "yes/no/maybe" by the optics?

Back in 1966(!) I had a job that included operating an IBM optical mark reader. It did exactly that, grading each potential mark as black/grey/white. If there was a single mark on the paper that it considered grey, it could be programmed to kick the sheet out into a separate hopper for correction and reentry.

Of course for an election the stack in the separate hopper would be set aside for manual examination (to see if the "grey" marks were a light vote, an erased change, or a paper flaw) and manual tabulation.

Here we are 39 years later and the technology has gone BACKWARD on its way to incorporation into what is arguably the most important tabulation job in the country.