A lot of what I do is deal with broken projects; systems that don’t work, software development that’s way behind or on schedule but shedding necessary features.
When things don’t work, there’s seldom a ‘smoking gun’; you have to dig a bit to try and find out what’s going on, toss up hypotheses and check them out, and one of the things I say a lot is “How would we know if that’s true?” The coders are claiming the database is performing badly; the DBA’s are claiming the application servers are misconfigured. OK, how would we know if any of those are true? Because once you can ask that question and get an answer, you’re on the path to defining tests which will let you make some firm statements.
So here’s an article about the election yesterday and e-voting:
WASHINGTON (Reuters) – Voters across the United States encountered scattered problems with new touch-screen systems on Tuesday as they voted in primary elections, but by and large the machines ran smoothly, state election officials said on Wednesday.
Voters in some polling stations in San Diego and Oakland, California, were turned away after officials had trouble starting up the machines, while others in Maryland and Georgia had to use paper ballots, officials said.
Aside from setup problems, the machines functioned smoothly, officials said.
“We had no technical issues at all, just the normal human stuff,” said Linda Lamone, Maryland state administrator of elections.
OK. How would we know if that’s true? We don’t, and more importantly, we can’t.
Folks, I’m coming to the conclusion that this is the most urgent issue on the table today. More important than fighting Islamist terrorism, more important than domestic security, more important than economic policy, and let me tell you why I feel that way.
It’s about legitimacy. Legitimacy matters, both domestically, and internationally. Here’s an example:
“Cherie Blair still believed that Bush had stolen the White House from Gore,” author Philip Stephens wrote in his book “Tony Blair: The Making of a World Leader. ”
Although Tony Blair was pragmatic about Bush’s victory, Mrs Blair was far less sanguine about the Supreme Court decision that gave him the keys to the White House.
She believed Al Gore had been “robbed” of the presidency and was hostile to the idea of her husband “cosying” up to the new President.
Even as they flew to Washington for their first meeting with the presidential couple, Mrs Blair was in no mood to curry favour, the book stated.
You want to imagine the impact when some disgruntled tech spins a story about deliberate, untraceable, errors coded into the machines that gather election results? When there has been a close, hotly contested election, and the presidency is at stake?
Kagan has written substantially on the importance of legitimacy in the international sphere; his positions are discussed in this column:
The researcher believes that “a great philosophical schism” has broken out between the continents because of sharp disagreements over the acceptability of the use of military force. The actions of the United States in Iraq have become a test case, in which what is at stake is “the future reputation of the United States, its reliability, and its legitimacy as a world leader”.
Kagan’s assessment of the present situation is right, as such. The United States has never been so powerful and so influential as it is now, but the wisdom and motives of its policies have probably never been so extensively and deeply questioned.
The country’s overwhelming strength makes it possible for it to wield its power even without legitimacy, but always at a high price. Acceptance is replaced by resentment, friendship by fear, and allies by vassals. Not even the most hard-line hawks in the United States can seriously want such a world, but the movement during the term of President George W. Bush has been in that very direction.
It’s still possible to push this back; but it’s going to take substantial action to do so.
I’ll suggest a simple standard: no e-voting system should be used unless it prints a human-readable ballot which can be stored in a ballot box, and unless the officials in charge of voting agree to test a significant sample of the paper ballots against electronic results, and to use paper ballots in the event of recounts.
And electronic scanners that count marked ballots should be subject to the same audit/test/recount requirements (note that as I understand it, the punchcard-reading scanners tend to be simple totalizators, which are harder to game – let me know if this is wrong).
So, A.L., how many of those big projects you work on rely on MS Access (like Diebold machines)? Hey, at least you don’t need a DBA!
Indispensable article on evoting
All experience with voting systems shows that most people do NOT check their ballots after they fill them out. Or else 99% of the trouble with punch cards wouldn’t happen.
Suppose you’ve got an electronic system, that prints out a reciept, and you want to rig it. You can set the machine to randomly alter some percentage of votes for your opponent into votes for you, and accurately record that on the reciept. IN MOST CASES THIS WILL NOT BE NOTICED.
In the few cases where somebody DOES notice that their reciept doesn’t match their intended vote, they’ll have no way to prove that it wasn’t an entry error on their part, and they’ll just recast the vote, and the second time it wouldn’t be altered.
This won’t reverse a landslide, but any casino will tell you “house odds” aren’t to be laughed at.
The paper record needs to be guaranteed to match what the voter filled out, to work, and the way to do that is to have the voter fill out the paper. Nobody has yet produced a system better than optical scan ballots.
What is the point of these machines anyway? Speed? Accuracy? Cost?
If we eliminate the speed of tabulating results as a requirement then it seems to me accuracy and (lower) cost are most likely assured. Seriously though, what’s the hurry? I get that too much time could open the door for “extracurricular activities”, but we’re only talking about a few days. And, besides, as has been suggested, a computer based system is hardly manipulation free (can you spell MYDOOM?).
And as A.L. says, legitimacy is the single most important result of an election.
My county in Maryland used the Diebold e-voting machines for the first time this week. Primary voting was easy and smooth, working just as advertised.
Of course, using the now-junked optically-scanned paper ballot system was also easy and smooth. In the event of a recount, the electronic record could be checked against the actual paper ballots, kept with the machine.
By the way, in Maryland it is illegal for poll workers to check the identity of the person coming to vote, to preclude intimidation. A Green Card holder I work with has had to emphatically decline getting motor-voter registered when getting her driver’s license renewed. So electronic tallying is not the sole threat to the integrity of the system.
Dear A. L.:
Speaking as someone who’s been an election judge for almost 20 years and who’s been a member of a design team that designed an electronic voting machine, I’ll give you my opinion of electronic voting: I’m agin it.
It’s almost impossible to decide where to start.
A.L.:
Social legitimacy, or the “right to rule,” is a modern term for an ancient concept that used to be called “sovereignty.” And although you didn’t touch on it, the social legitimacy of the Presidency is determined first by the assessments of voters, and clearly depends on whether the feel their effort to cast their ballot has been dealt with fairly and with integrity. The “legitimacy” that Mrs. Blair and others outside our system accord Bush begins with the legitimacy that US citizens accord to the election. The voting system is not the only component of social legitimacy, but it’s a necessary-and-not-sufficient condition.
My understanding of the electronic voting system is that adequate standards, which have been informally promulgated by NIST, can’t be implemented until the 2008 election. Now that, to me, suggests that the machines that can’t meet that “informal” NIST standard in 2004 should probably not be used until 2008, no matter how messy that makes things.
One of the big problems (and I say this as a Bush supporter) is that George Bush doesn’t see legitimacy as a critical issue. In fact, it’s almost a congenital weakness of Republicans that they are only concerned with a legalistic take on legitimacy. Bush didn’t not perceive the problems of governance created by halting the Florida vote count, and still doesn’t. He also tends not to see the issue of social legitimacy as critical within Iraq, as one can tell by the extremely low priority we place there on communicating with the population. And so one would expect a similar blindness toward the problem in the upcoming election.
The bottom line is that unless there are legal ramifications, there won’t be any change in process. A public outcry simply won’t cut it. And the first step would involve a move to make the NIST standards (which it accepted from an outside advisory group, I think) “formal.”
There are so many ways of rigging a machine to work one way one hour and differently the next that I would not even trust NIST to get it right.
And of course how do you know that there isn’t a “TEST MODE”?
Electronic counting of paper is a very good idea.
I’m sure it too could be spoofed but moving thousands of sheets of paper is much harder than moving thousands of bits.
No matter who wins the next election, the other side will be convinced the election was fixed. Electronic voting is so easy to hack, that the paranoid may not be wrong.
Please forgive a question from a foreigner (UK), but:
We use paper ballots. You get given your paper; you go into a booth where there is a pencil tied to a bit of string; you make your mark and stick the paper in the ballot box. After polls close, the boxes are tipped out onto long tables and the ballots counted by volunteers. The results are known by the next morning – in the last general election, as far as I remember, the last constituencies reported by midday, and that included a recount because the margin was 12 votes.
I feel like I am missing some crucial point here. Why do US voters need to use machines of any type (electronic or punch card)? I’m not trying to be rude, I am just genuinely puzzled. Surely the simpler the process, the more transparent, and the more confidence people will have in the result.
Ajay,
Short answer: you aren’t missing anything, you hit the nail on the head.
Longer answer: As you describe, counting paper ballots is cumbersome, labor intensive, and delays the time by which results can be known. These secondary traits have become primary design criteria for the US balloting system. In the aftermath of the 2000 Florida recount, they were set into law through a committment to electronic balloting.
It’s also true that paper methods are subject to their own (arguably lesser) varieties of fraud, but “legitimacy” and “transparency” considerations weren’t primeary in the switch to e-voting.
Doubly frustrating for those of us who cast our votes in 2000 using a optically-scanned paper ballots–a system that combines the virtue of the paper trail with that of electronic immediacy. But that is a mature and relatively cheap technology, seemingly “a bug and not a feature” in this context.
Ajay:
There are those who agree with you. I can venture a couple of guesses. First, we’re a lot larger with more different elective offices in different places than most Europeans can even imagine. For example, in my home precinct in Chicago our last ballot had perhaps fifty different offices some with four different people running for the office. In California 100 page ballots are not unheard of. These ballots would probably take days to tabulate manually. Second, speaking as an election judge, I already work a 16 hour day on election days. I would not take kindly to a multi-day extension of my work. Third, if there’s one thing I am certain of the longer the time between when the polls close and when the results are reported the greater the opportunity for fraud. In Chicago (punch card ballot, electronic tabulation) we can typically finish tabulating results in two hours. My precincts have never taken longer than four hours. So if you hear about precincts in Chicago that haven’t reported six hours after the polls have closed you can be fairly certain of monkey business.
Amac:
Just for the record studies done of punch card, mark sense, and fully electronic systems suggest that neither mark sense or fully electronic systems are typically any more accurate than punch card systems.
Correction: In California, a 1 page ballot. The description of the ballot can run to a not unheard of 100 pages. So the point about manually counting 100 pages for each vote taking days is incorrect.
Absentee voters get mark-sense, optical. Quick, verifiable, transparent. One does not need a “reason” for voting absentee.
Loren:
Thanks for the correction. I live in Illinois. What do I know?
The point remains–some ballots are very long.
It seems that there is a growing awareness that election integrity is now provable and practical through cryptographic means:
Recent New York Times article —
http://www.nytimes.com/2004/03/02/science/02VOTE.html
Most recent Congressional Research Service
report on election technology–
http://www.fairvote.org/articles/crs.pdf
Wisconsin has a great system:
1. When you show up, they write a number (sequential order, first come, first served) next to your name, confirming that you voted, and the same number on your ballot.
2. Your ballot has a series of broken arrows, something like this:
PREZNIT (choose ONE)
George W. Bush {– –(((
John F. Kerry {– –(((
pointing directly at each voting choice. To choose one, you use a black felt pen/marker to connect the two sides of the arrow pointing at your choice.
3. You insert the completed ballot into a large card reader. It registers your votes, and retains the paper record.
What’s not to like? Results tabulated quickly, paper trail if there’s a dispute or a recount needed. Am I missing something? Why isn’t a system like this more widespread?
I recall the first digital calculator I even owned. I just didn’t trust it to come up with the correct answer reliably, so I used to compute using the calculator and then recalculate everything using pencil and paper and a slide rule. Eventually I built up enough confidence in the thing that I could throw away my slide rule once and for all.
I don’t see why we can’t have faith in a properly maintained crytographic system with checks and balances and some failsafes. But we may need a paper trail for awhile until we get used to it. I haven’t given up paper credit card bills yet, though Chase keeps bugging me every month that they’ve extended the deadline for conversion to complete electronic accounts. Yeah sure. I hate companies that assume their customers are morons.
“..election integrity is now provable and practical through cryptographic means..”
Even if this is true (which I doubt, but I’ll read the links), there is a problem of transparency. You shouldn’t have to be a cryptography expert to have confidence that your ballot is fairly and accurately counted.
The comment one hears that “we trust bank and credit card computers” is not really a germane analogy. With these systems, we know the transactions we have made and can conduct an independent check..something completely impossible in an election system.
“With these systems, we know the transactions we have made and can conduct an independent check..something completely impossible in an election system.”
It’s not impossible. Read the story. Get informed.
“You shouldn’t have to be a cryptography expert to have confidence that your ballot is fairly and accurately counted.”
You could make the same argument for election systems dating back a hundred years: “You shouldn’t have to be a mechanical engineer to have confidence in a lever voting machine.” or “You shouldn’t have to be a computer scientist to have confidence in a tabulation system that counts punch card or optical scan ballots.”
Skepticism is good, but reactionary cynicism is hurting this country’s election reform.
I think the expectation for some minimal security precautions around the voting machines, of whatever sort, can hardly be characterized as “reactionary cynicism”. It’s a simple fact that in some voting precincts, especially those where one party is dominant and it’s hardly unknown for the other side to not have an observer on hand, what can politely be described as giving in to temptation happens with unforgivable frequency. This does matter because it affects the larger (district & state-wide) race. It’s often in such areas that security is at it’s laxest as well. Not coincidentally.
Yes, corrupt, machine-politics ballot-box stuffing has gone back for hundreds of years, but this was billed as a step forwards, not a step sideways or a step back. A step back is what it is increasingly looking like.
This is a step forward. It is a step forward for the blind voter that can cast a secret ballot without assistance for the first time in history. It is a step forward for the language minority voter that can easily access their ballot in their native language. It is a step forward for the minority voter where it’s been *shown* by Stanford Professor Tomz & Fellow Van Houweling that electonic and lever voting closes the racial gap in voided ballots by an order of magnitude.
Yes, confidence is vitally important as well, which is why I was happy to see this week’s NY Times Science *article* on making sure my vote counts.
This is a step forward. It is a step forward for the blind voter that can cast a secret ballot without assistance for the first time in history. It is a step forward for the language minority voter that can easily access their ballot in their native language. It is a step forward for the minority voter where it’s been *shown* by Stanford Professor Tomz & Fellow Van Houweling that electonic and lever voting closes the racial gap in voided ballots by an order of magnitude.
Yes, confidence is vitally important as well, which is why I was happy to see this week’s NY Times Science *article* on making sure my vote counts.