Slashdot Mirror
Diebold Fails Again in San Diego
ptudor writes "An article in today's San Diego Union Tribune reveals nearly 3000 absentee ballots in the San Diego primary one month ago were miscounted. 'The miscounts occurred because multiple scanners simultaneously fed the absentee ballot data into the computer tabulation system. The large number of ballots and candidates on them overwhelmed the system. Diebold spokesman David Bear said the company has provided a software fix to the county for the new problem.' The irregularities were found in a routine post-election review." You can also read more about the problems on election day.
Have you seen the "secret" video? Go here and take a look. I love how these things can't be trusted to add correctly.
Pen and paper: the only way to vote. Say no to machines.
Of course, there were only around 6000 votes in the first place..
"I can't give you a brain, so I'll give you a diploma" - The Great Oz (blatently stolen sig)
has ties to the republican party as one of it's largest donators. this whole thing stinks of day old feces.
I used to write mission critical software (as in, you-screw-up-and-your-user-can-die) for the US Army (Artillery Control). We had to pass internal unit test, integration test, system test, FQT, fielded IOT&E. At each point (past developer level integration), if an anomaly occurred, a trouble report was generated. All priority 1 and 2 reports HAD to be addressed and resolved. Priority 3 needed to be resolved or have a formal waiver.
1 - Failure to perform, user at risk
2 - Failure to perform, no workaround
3 - Failure to perform, workaround available
4 - Irritating/annoyance
5 - other
In the voting arena, I would say that problems with inaccurate counts would be priority 2 (since nobody dies directly). There should be NO WAY any fielded system should have those sorts of trouble.
The only reason we have the rights we have is that people just like us died to gain those rights. -- Cheerio Boy
There is something like 2000 pages on regulations and cirtification requirements your product most go through in order to be cirtified by the US Governement. The spec is unreal. I was invovled in this but can't say where. I will say Diebold was a competitor. Local governments don't have the same necessary must have requirements. The main issue is each state has separate laws for voting. You basically need to write software Helen Keller with a 20 IQ can use. That is tough.
"the Daily Show appearso to have a habit of making deceptive cuts"
People who use a Comedy Central as a new source are not qualified to comment on the news!
I love the Daily Show and I must admit that I use it as a news source. Therefore I am not qualified to comment on today's issues. Thank you.
Be happy. Nothing else matters.
I didn't see this in anyone else's reply, but if it's there and I missed it, pardon my redundance... The Computer Ate My Vote
The House Between - Original Sci-Fi Series
This article highlights problems also. In the follow-up it appears that Diebold still claims that their systems work, despite evidence to the contrary.
The real "Libtards" are the Libertarians!
Actually, IRV is not all it's cracked up to be. For more info (and other voting methods, like "approval voting") check out http://www.electionmethods.org/
The population density difference statistic is also misleading because probably 80-90% of Canada's landmass is extraordinarily sparsely populated compared to the U.S. (think boreal forest and tundra) If you take the strip within, say, 500km of the U.S. border, then the population density, while probably still much lower, won't be the almost 30x lower indicated.
The only significant difference in terms of the challenge is the vastly larger number of things that U.S. elections vote for -- judges, various officials, etc. In that sense, it is more democratic than in Canada, where many of those positions are appointed. And then there are all the propositions. So, per-person, there are many more things to tally, which is why I can see why an electronic or other mechanized solution might be desirable, even if problematic in other ways.
The receipt doesn't have to be given to anyone or even leave the machine. This has been discussed many, many, many, many, many times.
Run the printout under a plexiglass window and have the voter look at it and verify that the information is correct. Then run it through a second printer that gives it a confirmation or rejection code depending on how the voter responded to the "is this right?" querry. After that, it gets run into the takeup reel. The entire printing mechanism can be sealed in a tamper-proof box that can't be opened by anyone on the premesis, reducing the chance of tampering at the polling place by volunteers.
That takeup reel can even be OCR'd for 100% verification checks by a third party. None of this "spot checking" crap. Again, this reader can be built into the printing mechanism. If everything passes, toss the recipts in a cave somewhere for long-term storage. If they don't match then it's time to crack the seal and check by hand.
I live and vote in San Diego. I used the touchscreen devices; my wife used an absentee ballot. After using the Diebold boxes, I thought my wife had found a way to evade their problems. From today's article, it looks like I was wrong.
When I went to vote in the morning, at about 8:30 AM (well after the polls were scheduled to open), the machines were still non-functional (you've no doubt already heard the details) and the polling workers couldn't say when the help they requested would arrive. They suggested waiting or going to another polling location to submit a provisional ballot. (At this point, feel free to ponder why these were not tested by the vendor beforehand. Isn't that what YOU would have done?)
Nothing makes democracy feel real to you like being turned away from a voting booth.
When I returned in the evening, the missing cables were provided, instructions corrected and the devices functional. But not well.
In California, each voter receives a balllot information booklet before the election. With the old punch-card paper ballots, the booklet and the ballot were laid out in exactly the same way. You could transfer your decisions from booklet to ballot trivially. The touchscreen display, on the other hand, had the same visual look as the booklet, and the screen was laid out in pages, but page layouts did not correspond to the booklet. Candidates were in different locations on the touchscreen and the booklet. Matching up the two were a pain, and it took a very careful attention to detail to avoid error! Considering that the visual cues implied that that they should correspond, and that they did correspond in the old punchcard system, and I'd be very surprised if it didn't contribute to incorrect selections. (It was at least as bad, probably much worse, than the Florida butterfly ballots.)
Now, if you are replacing an existing system, isn't Rule #1 finding out how the existing system works, so that you know which functionality needs to be replicated?
The last page of the ballot is a vote summary. (Good idea.) It was multi-column on a virtual page that was one screen wide but much, much longer vertically than the physical screen. This is an atrocious user interface. (Imaging reading a PDF of a three-column, 8-1/2" x 11" page on a normal portrait monitor.) Prior to this summary page, the entire previous program was logical page = physical screen, with a horizontal prev page/next page paradigm. So, a bad user interface that's inconsistent with the rest of the application's UI.
Is that how you like to design YOUR software?
Finally, there's the fact that there's no paper record or physical trail of the votes. I can't begin to imagine how this passed Day One of requirements review!
All in all, it did not feel like the polished, professional effort that I want democracy and the control of our nation to depend on.
Actually, if you suspect you were counted wrong, you can pay some small fee to have your original answer sheet recounted by hand. If there's any discrepancy, they refund your seven bucks, and your score is updated.
Do you expect Diebold to do that?
Here in Vancouver, BC (Canada, again) our civic elections are reasonably complicated. It is a true multi-party system with independants allowed. We normally vote for 7 parks board trustees, 9 school board trustees, 12 city councillors (=~trustees), 1 mayor and a handfull of referendum questions.. Thing to note here is that for the 7, 9 and 12 seat positions, each voter gets to cast (up to) 7 9 and 12 votes out of all the candidates. Each of the parties (there are usually 3 or 4 parties running) usually fields a full set of candidates, and there are often independants, so it's not at all uncommon to be voting for 12 out of 50-60 (4*12+N) alderman candidates (as an example). It's not uncommon to also have between half a dozen (and up to 20) mayoral candidates. Then there are the referendums.
Voting is currently done on OCR... They are originally counted by computer, but if there are any questions, it's always possible to recount the paper ballots by hand (and it is done, from time to time). It's pretty easy to audit the computer results by picking a random polling station or two and comparing the computer reported count to the manual count. The system could easily handle a single-transferable vote system (like in Ireland) and have the machine counted results out before morning.
Much like in federal and provincial elections, candidates and/or parties can have scritineers at the ballot locations to ensure that everything goes as it should.
Because the system has a human-readable paper trail, I've never had any real quams about letting computers do the initial count. The technology is trivial (by today's standards) and well understood. None of this whiz-bang
bullshit.Sometimes boldness is in fashion. Sometimes only the brave will be bold.