A group of computer scientists and voting rights activists predicted that new voting technology in use throughout Maryland, the District and parts of Virginia would lead to widespread problems and possibly thousands of lost votes. So far, most of the glitches that have occurred seem to be minor and the result of simple human error.
Read the story:Washington Area Avoids E-Voting Misery (washingtonpost.com, Nov. 2)
washingtonpost.com tech policy editor Robert MacMillan was online Wednesday, Nov. 3, at 11 a.m. ET to discuss what happened, what didn't and what might still come to light over the next days and weeks.
A transcript follows.
Editor's Note: Washingtonpost.com moderators retain editorial control over Live Online discussions and choose the most relevant questions for guests and hosts; guests and hosts can decline to answer questions.
Robert MacMillan: Good morning and welcome to washingtonpost.com. I'm Robert MacMillan, tech policy editor and a staff writer for the Web site.
I'm here to take your questions about electronic voting, that is, the touch-screen machines and other modern voting technology that more than 50 million Americans used to choose their candidates yesterday. Feel free to drop a line.
Silver Spring, Md.:
How can we be certain that the Diebold electronic voting machines were not hacked into or previously programmed to assure Bush a win? It's pretty well known that the owner of the company said he'd do whatever needed to be done to assure a Bush victory.
Robert MacMillan: Thanks.
You know that old saying about death and taxes? Same thing goes for the Diebold AccuVote machines you presumably used yesterday. Despite the assurances of state officials, it's entirely possible that the machines were programmed or hacked into.
On the other hand, it's extremely unlikely. Elections chief Linda Lamone has responded to the reports that criticized the machines and taken measures to improve their security.
But even if that's a total sham, you'd have to imagine some interesting Manchurian Candidate scenarios to wonder why a Democrat like Lamone who has spent much of the past few months taking heavy flak from Republicans would want to score a victory for Bush, especially in territory that's pretty darned Democrat-leaning.
Robert MacMillan: Settlement Ends Campaign To Oust Md. Election Official (Post, Nov. 3)
Will anything be able to quiet the worries about a Diebold fix? On Monday every Republican insider's face showed their belief that the Democrats would win, except for Rove's. On Tuesday the exit polls said the Dems were winning. Then the numbers come in and say otherwise. With black box voting and overtly partisan manufacturers, not to doubt would be insane.
We need a precinct-by-precinct plot of exit poll 'error' and voting method asap. Has it been done yet?
Robert MacMillan: Please do continue to be skeptical. You might be wrong in your suspicions, but if you're complacent, you're guaranteed to lose. I suspect that your read on the insiders' faces is based at least in part on your interpretation, but maybe you're right.
As for a precinct-by-precinct plot of error and voting method, I'm not entirely sure what you mean. We do know pretty much how each precinct votes, technology-wise. I might be missing your meaning here.
Who counts the e-votes? Are they volunteers?
Robert MacMillan: Same people that count all the other votes. There are volunteers and elections board workers, though I'm sure the spread varies throughout the country. For the final final FINAL tallies, that's all certified/verified by elections boards for the most part, and that doesn't happen until days after everything's been settled.
To say that the glitches are due to human error is to say nothing. The incorrect floating point unit on one of the Intel pentium processors was due to human error. All programming bugs are due to human error. It is the nature of the glitch that is important. Since the e-voting machines were found to be subject to glitches that had not been eliminated during testing, we know for certain there are glitches in the machines. No more no less and it should be reported that way.
One must also remember that testing of digital systems does not prove the absence of glitches. Only through a combination of rigorous testing and careful analysis of the design and implementation can one gain adequate confidence that a system does what it is supposed to. There is adequate evidence as reported by Johns Hopkins that e-voting systems have not undergone an independent rigorous design and implementation review necessary to achieve a high level of confidence. Election officials are technically incompentent to understand the issues. The two major professional organizations for computing (IEEE and ACM) recommend a hard (as in paper) audit trail.
The worst problem I can imagine are bugs found that miscounted the ballots by 3 percent for Bush over Kerry. Ensuring that possibility may never happen again should be a high priority for the nation.
Robert MacMillan: You're right about human errors. The question is at what point is the error the fault of the designer of a technology or the fault of a user. There are valid arguments to either side in e-voting.
On the idea of elections officials being too incompetent to understand the issues: We all use technology to make crucial decisions without necessarily understanding how it works. Who ought to run the election system?
Takoma Park, Md.:
Have you interviewed anyone at the Independent Testing Authority? What exactly did they test, and how did they test the Accuvote-TS machines? Who did you talk to?
Robert MacMillan: There's the fodder for another story. Some media outlets have handled it in more detail than I have. I've only skimmed it some, but it's begging for a closer look. For all the skepticism I keep tossing at folks who are plainly peeved at the paperless voting system, they have a point here. It would be nice to have some kind of independent auditor serving only the public interest to be able to say to us that these testing authorities are really on the money.
Short answer to your question: It would be great if they called me back. I'll keep trying. One of them was quoted in a recent article in the Rocky Mountain News, but that's one of the few times I've seen them quoted.
Takoma Park, Md.:
The way I see it, it really doesn't matter to me if the Diebold machines seem to work flawlessly. We have no idea what is recorded after we "push" the digital button on the screen. How would we check that the recorded vote reflects voter intent? Perhaps a recount -- but as things are, the only hard copy we can print out to use in a recount will just vomit out the same thing that was recorded on the machine, and not necessarily what the voter thought she was voting.
Robert MacMillan: A question to which I've not yet received a satisfactory answer.
Why is it that exit polling used to be a fairly accurate way of determining the winner, but in the last two elections, it has been woefully inaccurate and favoring almost exclusively Republican candidates?
Robert MacMillan: I think Bob Kaiser is also doing a Live Online chat now and would be far more qualified than I to answer that question.
ES&S, Diebold, and Sequoia all received large sums of money to install their touch-screen voting machines in many states across the country.
These companies have admitted ties to right-wing Republican politics.
Why no investigation into voter fraud?
Robert MacMillan: By whom? Media? Congress? Speaking for myself and post.com, we're always looking into where the money is and who's connected. Diebold's ties are out there. Whether the company is committing voter fraud remains unknown. As for the other companies, the same thing applies. We'll keep checking it out.
I'm curious about something - would any of the Democrats, Greens, et al., here be as incensed if they suspected that it were people with close ties to John Kerry and John Edwards who were running these companies?
Takoma Park, Md.:
I used an e-voting machine yesterday and it worked fine. In fact, I liked it better than the other voting machines I've used in the past. But I'm not at all sure that it actually recorded my vote. Why can't we have printouts again?
Robert MacMillan: When did you have printouts before?
This is a fascinating issue because the assumption that e-voting is dangerous is premised on the fact that paper balloting is safe and effective. The same people who will monkey with e-voting will monkey with paper ballots.
After standing in line for two hours in Arlington, I think the answer isn't less technology, but more. Why, in the most technologically advanced democracy in the world, do we stand for two hours in the sun waiting to vote on five machines? It's absurd. We rely on technology everyday to do sophisticated, confidential tasks.
Robert MacMillan: Good point. I'd rather face a malfunctioning e-voting machine than a hired goon with a gun sitting on the ballot box encouraging me to vote for a particular candidate. There's a fine history of that sort of thing in this country, and not always so long ago.
That said, people who say that touch-screens and modern "dre" voting systems have solved all our problems are more than likely incorrect. We have to keep working on this, plainly.
Why is there no investigation into corporate touch screen voting machine makers and their ties to Jeb and George Bush?
Robert MacMillan: Who should launch an investigation? I'm still curious to know whether this ire would apply if Democrats were perceived as the puppetmasters behind the e-voting machine manufacturers.
Washington, D.C. :
I read that most of these e-voting machines were designed with closed-source systems on top of proprietary Microsoft XP operating systems. At what point do you think we will as a country insist that at least the voting technology programs be made available to open inspection? Seems like a delicate thing to leave to the whims of a few companies.
Robert MacMillan: I have a feeling that no one of any influence will insist on it until someone in the media can explain why this technological setup is a powderkeg situation. Well, sounds like that's my job.
Takoma Park, Md.:
Regarding the idea of "user error" ... a badly designed interface can easily lead to "user error." The manufacturer could have made horrible decisions about what it considers intuitive. If the voter can't figure out how to toggle a vote, if a judge can't figure out that he has to program a card a certain way ...
Apparently Diebold was still hiring technicians days before the election.
Robert MacMillan: Those technicians weren't altering the hardware, so that's kind of a non sequitur. If they were still fixing buggy software, that suggests there's something wrong. Good to know.
Meanwhile, badly designed interfaces can lead to user error. User stupidity and haste, not to put too fine a point on it, can do the same. I'm not all that smart, nor am I all that technologically sophisticated, but I can operate a voting machine (don't ask me about putting on a necktie without a mirror).
That said, the machine makers need to continue to refine the user-friendliness of the machines. And users need to continue to slow down and breathe during the voting process. And if there's truly a problem that the machine caused -- and not the user -- then that user should raise holy hell about it.
So electronic voting machines seem to have dodged a bullet, at least for now. Is there any indication that hackers tried to tamper with the results of any major races? I read somewhere that at least some of the machines relied on very insecure 802.11b wireless networks to transmit voting tallies to the central servers in a number of states.
Robert MacMillan: I haven't heard about any, but there are thousands upon thousands of precincts that we can't examine with a small staff. Looks like we'll have to rely on tips and reports from local media to see what might come to light.
My idea is this: The exit polls showed that Kerry was winning. The results showed the opposite. The exit poll predictions were in error.
For each precinct they exit-polled, calculate the outcome predicted by the exit poll -- the outcome reported by the machines. Call that the precinct error.
Now sort the precinct error according to the type of machine used in that precinct. If Diebold machines were adding Bush votes, their precinct errors should be higher.
Robert MacMillan: I'll take your word for it. And thanks for explaining... and now you can see that I really am not all that smart. ;)
Takoma Park, Md.:
Of course I can't say a paper ballot is safe and effective! I'm not operating on that assumption. The new voting machines are supposed to address issues of accuracy, security, accessibility, etc. So what if a paper ballot is susceptable to abuse? We shouldn't use that as an excuse to be complacent about a new technology that we are capable of improving.
Robert MacMillan: You are absolutely right. I think I can say, without sounding like a radical of any stripe, that complacency is the worst of all possible options.
I voted early and had a "problem" with a touch-screen machine that still haunts me. I am sure I depressed the second choice on thepresidential ticket 'Kerry/Edwards'; however, the Bush/Cheney was lit/selected/checked by the machine. Of course I went back and corrected the "error", but I'm sure many of the users in the precinct where I voted may not have noticed or known how to go back and correct their choice. I am registered as one not affiliated with D or R's, and I can't help wondering if there was some "funny business" going on with the machines.
Robert MacMillan: It's possible. I'll never say it's not. I'm glad you bothered to check your results - the machines are required to present a summary page. I have little sympathy for voters who are in too much of a hurry to check their results.
Robert MacMillan: I take it you all saw the news reports that Kerry's gonna concede.
Mt. Lebanon, Pa.:
Is there any competition for e-voting machine and software manufacturers or is it just more federal money going to the right- wing darlings like Diebold?
Thanks much. Political Atheist and Professional Electrical Engineer
Robert MacMillan: Dear Atheist: Yes, there is competition in the voting machine industry.
Does this preliminary finding that e-voting was largely glitch-free mean we can expect even more touch-screen machines in upcoming elections?
Robert MacMillan: I think there will be more computerized voting systems in more of the United States by the time various states and localities hold their next elections. I do not think that this is a direct result of articles like mine saying that reports of glitches seemed sporadic.
It was bound to happen once Congress made 4 billion of our tax dollars available for it.
Has there ever been a good reason NOT to have a paper trail? I've heard several politicians, election board officials, and machine manufacturers issue assurances that they're not necessary, but no one's given a reason not just to go ahead and do it.
Robert MacMillan: Some people say a paper receipt that a voter can take out of the polling place can be used to provide evidence of a vote delivered to someone else's instructions, ie, voting for reward. That's illegal. The solution is to force the voter to cough up the receipt before leaving the polling place.
Disabled groups, particularly the blind, don't like the paper trail because they want a voting system that will let them choose their candidates without anyone's help. A piece of paper that requires sight invalidates that proposition. I wonder if it could be printed in Braille? There's a thought.
Your introduction does not specify the nature of the glitches. However, I have read that some machines froze while people were voting. This is typically due to poor electrical design and power variations in the power the machines use. At least one glitch I read about was due to the machine not being plugged in. Another glitch I read about was that Mikulski's name did not appear on a ballot when it should have. This is a programming error. In theory the designers should be aware that the machines will be used by many people non-expert in computing and have eliminated by design almost all possible user glitches. I will give them the power being plugged in but little else. Otherwise, I consider the glitch a design flaw.
Robert MacMillan: I've heard the same reports.
Beckley, W. Va.:
Wonder why we demand a receipt at the ATM, after we have eaten at a restaurant, a receipt, purchased goods at the local K-mart, a receipt. After we've had our car repaired we want to see the bill plus sometimes, the old parts. Yet when we vote, no paper trail, no sense whatsoever, voting is too important to be done WITHOUT a paper trail. Just go to an IRS hearing about your taxes without paper trails.
Robert MacMillan: Good points, one and all.
Why is there no paper trail for these machines? Seems like it would be pretty easy to hack into and change tallies enough to sway an election, and with no paper to back it up, what's the recourse?
Robert MacMillan: Do you think that if there were a paper trail of some kind, hackers might be able to manipulate it so the paper says one thing but the inner workings produce the result the hacker desires? There's an X-Files scenario.
It's not a D.C. area question, but a couple things give me pause about yesterday's results. What exit polling I have seen (Florida and Ohio) appears to be contrary to reported results. Also, some rough calculations with vote tallies suggests strongly that fairly huge numbers of people who voted for Gore in 2000 switched to Bush yesterday. I have yet to meet one and likewise your Mr. Kaiser.
Who knows? But when substantially electronically generated results conflict with independently derived empirical data as well as common sense, I think it is only prudent to be concerned. Unfortunately, there is no reliable means of assuring or verifying the integrity of any electronic system. Paper ballots may be cumbersome and inefficient but those faults also limits the extent to which they can by systematically manipulated, a limit that does not exist with electronic systems.
I am a tech head from way back, and I love electronic voting machines to use, but I think they should be banned. Period.
Robert MacMillan: It would be nice if the people who run the e-voting companies were to solicit brainstorming scenarios from concerned parties. But of course they'd probably be nailed by the Justice Department for collusion. What a world...
How come e-voting critics missed the mark so much? It seems there were a lot of scare tactics lobbed about hacker attacks and other ills. What happened?
Robert MacMillan: Well.... all sorts of things could have happened. We'll never know. I think that conspiracy theorists missed their mark, despite the headline (and this is my assumption about their political beliefs) currently on our homepage.
I do believe, however, that there are plenty of e-voting critics whose concerns are valid. It would be wrong to say, "ok, we've reached the pinnacle of voting technology. To innovate further would be useless." It would be nothing but useFUL.
I wonder if electronic voting poses a more practical problem: It takes longer, because there are fewer machines in each polling station (due to their high cost) and voters must wait in longer lines. Other methods such as optical scanners or punch cards should allow more voters to vote at once since their use is limited only by the amount of space at the poll. And of course, there's no problem of a broken machine making the lines even longer. How much has this aspect of e-voting been studied?
Robert MacMillan: I think it will be studied more now. It's clear that there were not enough machines of any kind in the polling places that I and my reporters visited yesterday. Reading dozens of newspapers yesterday from around the nation provided similar accounts.
I think the best answer to that is to buy more machines, whatever they are.
After using the electronic voting machine for a second time (first was the primary in the spring), I must say that the machines in Maryland are simple to use, a step in the right direction to the future of voting.
There seems to be a larger than normal amount of absentee ballots this year. With that in mind, I believe the future should include casting your ballot via the Internet. There are a lot of hurdles security-wise, shouldn't that be the goal? It seems to me that people need to embrace the idea and work out all the problems rather than declare it unsafe and not consider it.
Robert MacMillan: It's hard to say whether Internet voting will be a reality or not. There are too many security questions right now to make it a completely reliable method of voting, it seems, but I'm no techie so I can't say for sure.
Robert MacMillan: OK everyone. That's it for me. Thanks very much for coming online, and thanks VERY much for reading my articles and for visiting washingtonpost.com. Please feel free to send in tips. Please keep 'em brief, as I'm the reigning ADD champ in this office. Thanks again.