Slashdot Mirror
Cringley on E-voting
alfredo writes "I am shocked that this story from I Cringley hasn't been sent in and posted at Slashdot. I thought the slashdot crowd would be all over this. Robert X Cringley has a take on the voting scandal a bit different than what we have seen in the past, and promises more to come."
Simple way to make it secure... The electronic machine FILLS OUT A PIECE OF PAPER CORRECTLY AND COMPLETELY. The person INSPECTS this for correctness before making it his/her vote. -- E-voting keeps the democrat from crying "hanging chads, dimpled chads... RECOUNT, RECOUNT, RECOUNT!"
The American Civil Liberties Union said in California that certain counties in the recent recall election were disenfranchised by not having touch screen voting
No, The American Civil Liberties Union said in California that certain counties in the recent election were disenfranchised by using punch card voting. The fact that one of the alternatives to punch card voting is touchscreen voting does not mean that the ACLU was demanding the counties use touchscreen voting, just that the counties not use the punch card voting systems which had lost their legal certification anymore!
Cringely's intentions are excellent but he plays into the biggest, most disasterous, most helpful to the voting companies fallacy in the entire mess:
Recording method of votes and tabulation method of votes are entirely separate, orthogonal concepts.
The first has to do with, do you make a mark on a piece of paper, pull a lever, or touch a button on a screen? The second has to do with, are the votes recorded on paper and dropped in a box to be counted somewhere, or are they put on a hard drive to be just added together somewhere?
The first is what electronic voting salesmen are mostly selling the systems based on. The second is what electronic voting's enemies are mostly complaining about, as it alone is what makes almost all of the potential cheating possible. There's *no reason the two have to go together*! You could have a touch-screen voting machine which prints out a scantron sheet, which then is dropped in a box and counted like a hand-filled-out scantron sheet would have been.
A lot of the support for "electronic voting" has come from the fact its proponents have attempted as much as possible to prevent the false choice of "Punch cards VS electronic voting!" and hoping pieces of paper won't come to people's minds. But much of the remainder of the support on this issue have come from people using the advantages of touch-screen voting to sell "electronic voting", acting as if the touch-screens are inseperable from the idea of storing votes for tabulation on fragile, black-box electronic media, and banking on public confusion about All Things Computer to assume people won't notice this.
Irritable, left-wing and possibly humorous bumper stickers and t-shirts
Another concern that I have is the desire of government to jump from the trailing edge of voting technology to the bleeding edge of voting technology. The Florida election results clearly showed the problems with punch card voting. However, many of these problems were due to poor ballot design, poor maintenance of voting equipment, or poor training or poll workers and voters. (A large number of hanging chad problems were caused by the simple failure to clean out the chads from previous elections.) Boston, Massachusetts switched from lever driven mechanical voting machines to paper ballots and optical scanners. There were problems with the transition, but most of the problems were procedural in nature and not technical in nature. The combination of paper ballots and optical scanning has a very good track record. The paper ballots provide a nice audit trail that can be used to verify the results of the optical scanning and computer tabulation.
I live in Somverille, Massachusetts where paper ballots and optical scanners have been used for years. The systems is backed up by experienced poll workers. I've never heard of any problem, let alone a serious problem, with this system as it is implemented in my city.
Congress should have proposed moving to the best voting technology available that has a proven track record. This would avoid the issue of bleeding edge technology that has an unproven track record. The biggest problem with computer based systems that have closed source code and no paper trail is the inability to properly inspect and test these systems to make sure that they are as good or better than the technology that they seek to replace.
What does this mean? If you want a program that does X, Y and Z, and you get one that does X and Y, it could still be useful and worth the money you spend.
/. a few days ago, only even more so, becuase many of the slot machines' systems of ensuring fairness are made impossible by the voting systems' requirement of anonymity. You can argue that this is an implementation problem, and that the problem is just that the current implementors are just putting the minimal amount of effort into trust, and that's just not enough. But I would say it is fundamental because the amount of effort required to make the system trustable is so great that it is unlikely anyone will ever be bothered to reach it. People will always inherently want to cut corners..
The problem here is, what if Z is the most important requirement for the project?
There are a number of different criteria that are desirous in a voting system. However, a few of them are absolutely necessary. The ones that are necessary are that it must not introduce statistically significant amounts of error, it must be anonymous, and it must be auditable and trustable. If you lack any of these qualities, you wind up with a system which is worse than nothing at all, becuase the system is not just flawed: it is potentially damaging.
Punch cards become an unworkable option because they violate the first of these. The potential margin of uncatchable error is large enough that it was larger than the margin of victory in the deciding area of the last presidential election.
The electronic voting systems currently being pushed have almost all of the desirable voting-system qualities, lack the last of the necessities: they are inauditable and untrustable. This is not just an implementation problem. It is a fundamental problem-- becuase any auditing methods for the system must themselves be electronic, and thus as susceptable to being cheated as the system itself. It is perhaps possible to create a trustable electronic voting system. However, it requires an absolutely obsessive-compulsive attention to detail, something along the lines of methods used on the Las Vegas slot machines mentioned on
You have to remember, it isn't enough for a voting system just to produce a correct answer. It has to to the greatest extent possible eliminate doubt. If you have a system which is not trustable, but by coincidence just happens to be accurate, it's still going to be a problem because the elected candidates enemies will be able to go around for that candidate's entire political lifespan claiming that they stole the election-- and really, who can definitively say that they're wrong?
Irritable, left-wing and possibly humorous bumper stickers and t-shirts
First, the reason there's no paper trail despite all of Diebold's other machines having a paper trail is that the Diebold voting machines aren't made by the same people. Diebold bought another company that was already making voting machines, and they haven't had anything like enough time to "merge" the two companies' engineering groups. You see this all the time in IT, some company (Cisco, for example) buys another company, and starts selling their product (the PIX, for example) with their name on it (so now it's the Cisco PIX), but it takes years to actually do more than piddle on it to make it smell like the parent company. Looking at the Cisco example, the PIX is still an odd-man-out product in the Cisco product line.
Second, it's not hard to produce an audit trail *and* assure the votes cast will be anonymous. You just have to make two decisions:
1. The auditable ballot is the real ballot.
2. The vote is complete when the auditable ballot is complete and saved, not when the "user-friendly" ballot is complete.
There's two basic ways of doing this.
One way is to make the touchscreen machines a more convenient way to generate your traditional ballots. That is, the touch screen produces a human-and-machine-readable form (OCR, punch card, whatever). You're taking advantage of the fact that the machine's card punch always punches clean through, that its printer always colors inside the lines, but no more than that.
The other is to let the user see the auditable ballot, but keep it inside the machine. Once it's printed, the user punches "VOTE" or "CANCEL" below the window, and the ballot is delivered (visibly) to the ballot box or the shredder.
Intermediate between these, have a printable ballot that's got a random machine-readable tag on it that the user can deliver into one of two slots, the ballot box or the shredder. After the machine has read the tag it verifies that the voter didn't just shred a blank piece of paper... but the tag is not stored after the ballot has been accepted and it's generated anew using an external entropy source (such as the timing of the voter's screen-taps or keystrokes) for each ballot, so there's no trail leading to the voter.
Any of these would work. The first one could be retrofitted to existing optical or punch card systems, which would allow for precincts to complete their votes even if their electronic machines are down.
I am looking forward to Cringley's next column where he proposes to answer the question of why auditing capabilities were not inlcuded in the touch screen voting machines.
I can answer THAT for you right now. He's going to (correctly) assert that the reason there is no paper-trail requirement is that the political establishment DOES NOT WANT ONE. The original vote tally is a one-time process, but the recount process can drag on forever, and THAT is what "they" want to avoid forever more.
To ensure perfect aim, shoot first and call whatever you hit the target
There are lies, damned lies, and then there are statistics.
Article:
The bad news is that in 2000, only 28 percent of software projects could be classed as complete successes (meaning they were executed on time and on budget) while 23 percent failed outright (meaning that they were abandoned).
According to my math, that means that 49% of projects took longer and cost more than they were supposed to. Note later in the article, this 49% is considered wasted:
Article:
Two hundred and seventy-five billion is a lot of money to spend on software development, especially if 72 percent of that money will be either wasted completely or used to develop something that doesn't work intended.
But something's wrong. Let's come up with a product and let's call it OS X or Mandrake or Windows XP. All of the above were not completed on time. In fact, I'd say I'd rather have a polished late product than release something on time for the sake of doing so. (Name good software that was released on time someone?) So I guess all the money spent on all of them was wasted.
Someone hit this guy with a clue stick.
-- Political fascism requires a Fuhrer.
Rigging the voting machines is a really hard way to rig an election, you need a lot of people to be in on the fix.
Why would that be? It only takes one well-placed person can write the malicious code and hide it in the software. Indeed, you may not need to be well-placed at all.
But even assuming what you say is true, so what? Look at the stakes. Look at all the past examples of election tampering, many of which involve large groups of people.
It isn't paranoia to be concerned about these machines, for this one simple reason: any other flaw in our democracy can be addressed by our democracy, but not this. Once we lose the vote to these machines, we lose the capacity to remove the machines from the process. It's a one-way street, and once we're on it, the only recourse will be violence, a la 1776.
So we should take great care to make sure we don't take that road.
Is this truly the only Earth I can live on?
What is probably even more crucial is a discussion about voting being accessible and easy. What's amazed me in the post 2000 election discussion is how fast we've stopped talking about all of the voters who were disenfranchised by having huge difficulties getting to a working election center.
The underlying reason why all of use really want to see internet voting is because it would be easier for us to vote. We can pay all of our bills online. We can file our taxes online. Why can't we vote?
The reason is because it is a really difficult security problem to solve. I'm just amazed there isn't more discussion about how to solve that problem than the discussion talking about a poor implementation of the short-term, band-aid solution.
Specifically, I thought http://www.eucybervote.org/xootic2000.pdf has described a really good start to how to really solve the security problem.
What is the size, in thousands, of the voting population in the UK?
o What is the size, in millions, of the voting population in the US?
Doesn't matter, since both countries will break everything done into very small jurisdictions.
The US has 280+million
Ohio has 11.3 million
Franklin County has 1.2 million
Columbus has 750k
my pollworking ward has 10k
my pollworking precint (four pollworkers per precinct) has 850 registered voters and of those 850 registered, about 120 will vote in an off year election, about 300 will vote in an even year, 500 will vote in presidential election
Here in Franklin County we use machines, but with four pollworkers, I imagine we could count paper ballots up fairly quickly, even if 500 people vote. (After all, that's why there's four of us.)
The main problem in the USA isn't how we gather votes, although there are problems in some states (Florida). There is a more fundamental problem in that we aren't using the right voting mechanism. In the US, we use plurality voting -- a.k.a "first across the line" -- to determine who wins an election. This means that a candidate for whom only 30% of the people voted can win an election simply because there was no other single candidate with more votes.
This has a number of problems, but they can all be summed up by saying that plurality is one of the least fair, if not the least fair, way of determining the winner of a democratic election that you can get. Consider:
- Say 40% of the people vote for candidate A
- 35% of the people want candidate B
- 25% want candidate C
In the US, candidate A will win. However, what if all of the voters for C would rather have B than A? Then 60% of the population would rather have B than A, and the minority candidate has won.This situation encourages strategic voting; that is to say, voters for C have to decide whether they want to vote honestly, for C, or whether they should vote for B just to make that they don't get their least favorite candidate, A.
This is why we only have two parties in the US, and why -- despite the large number of Greens and Libertarians, neither party has a chance of winning. We don't even know what percentage of the US population is Green or Libertarian (or anything else, for that matter) because they aren't voting honestly. They're voting for the lesser of two evils. This system practically guarantees alienation of the largest number of people -- the majority ends up with a candidate they don't want, unless they lie when voting and vote for the candidate that they dislike the least who also has the best chance of winning.
There are voting mechanisms which allow people to vote their true opinion without being alienated. The most popular are Condorcet -- complex, but the most fair; Approval Voting -- not as fair as Condorcet, but much simpler, and can be implemented with existing voting technology; and Instant Runoff -- less fair than approval, no more simple -- but better than plurality.
Many democratic countries do not use plurality voting, although plurality is the most common. For example, Australia, Northern Ireland, and the Irish Republic (among others) use single transferable vote[1]. In fact, 68 countries (~2b ppl) use plurality, 31 countries (~400m ppl) use single transferable vote, and two countries (~18m ppl) use IRV (instant runoff) -- this is according to International IDEA Handbook.
There is a huge amount of information about Condorcet and Approval Voting available on the web. The Citizens for Approval Voting page is a good start, if you're at all interested in improving voting in the US. If you're interested in the mechanics and mathematics of the systems, start with Condorcet -- most sites that talk about Condorcet are less about how to get it implemented politically, and are more about how it works, fairness tests, and how it compares to other systems. The Wikipedia entry for "voting system" is particularly useful.