UC Berkeley NewsView of Campanile and Golden Gate Bridge
NewsCenter
Today's news & events
News by email
For the news media
Calendar of events
Top stories
Untitled Document
Web feature

Transcript of press conference on court decision delaying recall election

- The following is a transcript of a Sept. 15 press conference by Henry Brady, UC Berkeley political science professor and an authority on polling systems, following the 9th Circuit Court of Appeals decision earlier that day postponing California's Oct. 7 special election.

Henry Brady: When we looked at data from around the country and we found out that without a doubt, again and again and again, punch cards perform badly compared to almost all other systems. In fact, in terms of all modern systems, basically punch cards always perform the worse. So that started to get me concerned about punch cards and that got me more deeply looking into other kinds of data which I’ll talk about in a moment.

Why did I get involved in this particular case? Well, the answer is, as I say, punch cards perform badly. What do I mean by that? Punch cards throw away votes. That’s the easiest way to say it. They throw away votes. People come to the polls. They mean to vote. They intend to vote. They try to vote. They make a mark on the system trying to vote, but the system does not record their vote.

Here [pointing to Votomatic machine] is an ancient punch-card system. The reason it’s here, by the way, is that the punch-card system was actually developed by a University of California professor, Joe Harris, who was here at the Institute of Governmental
Studies, and this was found in a closet, actually, a few years ago. It’s one of the initial models of the punch-card system.

As you all know, it involved a stylus – this little thing here - that you push through a punch-card - here’s the punch-card, just to remind everybody what it looks like - with little chads in it that you are supposed to punch through, and you do that by poking holes. But you can’t see what happens when you do that. You can’t see through those holes, so people don’t quite know what happened when they voted. It’s hard to check your work with a punch-card system.

Punch-card systems are unlike anything else we use in everyday life. Paper ballots are like marking a form. Electronic systems are like ATM systems which all of us know how to use today. Optical scan systems are like standardized tests where you fill in a bubble. Those are things all of us have done, we’ve done repeatedly, and we know how to do them.

We don’t do, on a regular basis, anything like punch-card voting. There’s no other system that uses this principle. As a result people don’t understand how punch cards work. As a result, it’s very hard for them to do what they want to do, which is to vote. Furthermore, the system seems to have a lot of problems even when people are doing the right thing. There’s a chance that you’re not going to push out that chad since you can’t see whether or not you did it. You can’t see whether that machine is working or not. So, as a result, these machines are prone to a lot of errors. And that’s what we find in the data.

These [pointing to charts] give you some idea. In the state of California, and almost all the data I submitted in the Declaration to the Court was based upon California data. So I’m not taking Florida results or results from other states and trying to understand what’s going on in California, I’m taking data from California and talking about what goes on in this state – so this is about this state.

This shows you, for each system, what the average rate of residual votes is. Residual votes are what we call undervotes and overvotes. Undervotes is where there appears to be no mark, in this case, upon a punch-card if we’re talking about punch-card voting. Overvotes are where there seems to be two chads removed or more for a particular race, thus invalidating your vote. Both of those things happen much more frequently with punch-card. The net result is residual votes are extremely high for punch cards. For all other systems, here’s central count, optical, Data Vote (which is a form of punch-cards different that these…this is the Votomatic Pollstar kinds of voting systems). Data Vote uses a system where you can actually see the name of the candidate and you actually cut out a piece of the card with the name of the candidate next to the piece you cut out. So Data Vote works better. Why? I think, because the names of the candidates are next to where you make the mark.

The trouble with punch-card is that it radically separates the place where you make the mark – namely, this punch card – from the place where candidates names are listed, which is this, the ballot. That’s the basic problem, I think. And there’s also maintenance problems and other problems with punch cards my colleague Roy Saltman talked about in his declarations to the Court and that are discussed in the decisions.

So other systems perform much better. This disparity [pointing to chart]… this is about 2.3%, this is about .89%…that’s something like one to 1-1/2 % of votes that are just thrown away.

Now what does that mean? If you went to an ATM, and every time you went $1.50 was taken away from you, you’d be pretty upset. That’s the problem here. We have a system that basically takes 1-1/2 votes away from people who are unlucky enough to be voting in punch-card counties. There’s an additional problem with punch-cards, and this may not be so clear, but here I’ve plotted [pointing to 2nd graph] residual votes versus percent minority and census tracks, and unfortunately this isn’t very dark but this is Fresno county in 1996. As you go higher and higher in minorities you get more and more residual votes, up to something like 7%. In 2000 they changed to a new voting system, an optical scan system, so instead of having a high level of residual votes, lots of thrown-away votes, and instead of having lots of throw-away votes especially in minority neighborhoods, suddenly the situation comes down and even in minority neighborhoods, we have low rates of residual votes. We’re not throwing away votes at the same rate anymore.

This is how we know that replacing punch cards with other systems will improve the performance of voting systems and lead to us not throwing away votes, even when people mean to vote. I’ve done this study again and again with different counties in different states, a lot here in California, and I find again and again the same pattern. Get rid of punch cards, residual vote rate goes down, we throw away fewer votes. That’s good for democracy. That’s equal protection for votes, which was at the center of this law case.

Indeed, what the court did was this. They said, “Look, we’ve got two competing interests here. One is timeliness of elections. We all want to have the election as soon as possible to get it over with. On the other hand, we want to make sure we’re not throwing away people’s votes.” And they said that throwing away people’s votes was more of a problem than timeliness of election.

They said that for the following reason. This is not like a normal election where if you don’t have the election, an office is unfilled, the current governor is finished and somebody else has to come in, or the office is vacant and decisions can’t be made. In this case, if we don’t have the recall on October 7, there will still be a governor there to make decisions and government can still run.

So this is a much different situation than the ordinary election and the court felt, in these circumstances, it was worth postponing the election. A tough thing, they said, not an easy thing to do. It was worth postponing the election in order to make sure that we count every vote. And that’s what this was about – counting all the votes – and that’s what the decision was about.

I’d be happy to take questions and to talk more about other explanations for what’s going on in these data, what I think the court case is about, and so forth.

Q: Can you explain one more time about the correlation between punch-card systems and minority votes? Is it that in some areas that have a larger minority population they are using the punch-card system more? How does that work?

A: There’s two things going on here. First of all, those six counties, which comprise about 44% of the California electorate – almost half the electorate – those six counties (they include Los Angeles and San Diego and Santa Clara, for example) they’re much more likely to have minorities live in them. So there’s a higher fraction of minorities. So in the first instance, minorities are more likely to have to vote by punch cards than non-minorities.

But the second thing that’s also true is, it turns out in those areas where there’s lots of minorities, we find higher rates of residual votes. Now why is that so? It’s probably mostly because those people have low levels of education and these are hard systems to use. They’re hard systems to use for anybody, by the way. The important thing to know about this line is that even with people who are non-minority, even with people with high education, punch cards throw away more of their votes than other systems. So everybody has trouble with punch cards. But people with less education have even more trouble with punch cards.

That’s not a surprise. We’ve designed a lousy system for voting and it’s hard for people who can’t sit there and figure out how the system works.

Q: Are you pleased with the outcome?

A: Yeah, I’m pleased. I think that we shouldn’t be using punch cards in an election. I think that this recall election is about democracy…that what the recall’s about, it’s about democracy. We have here a concern with counting people’s votes. That’s about democracy. So this is about democracy triumphing.

We will have a recall election. Be sure about that. But we will have a fairer recall election because we’re going to count all the votes if we wait until March.

Q: Are you confident that if this goes to the U.S. Supreme Court that they will consider it?

A: I don’t think I’d want to predict what the Supreme Court’s going to do. I will say the following, that this case is full of references to Bush vs. Gore, which was, of course, the decision about the November 2000 election that many say decided who was going to be the president. So it refers to that case. That case was about different standards in different counties for counting votes and it was about, all in all, several thousand votes, depending on which standard you took. If you took the tougher standard, you were going to get fewer votes recounted and if you took the easier standard you were going to get more. But the difference between those two cases was several thousand votes.

These systems here are about 40,000 votes, by my best estimate. So 40,000 votes won’t get counted if we use punch cards. That’s 5-10 times greater disparity than what we found in Florida that led the Supreme Court to say, “Equal protection for votes matters. We’re not getting it in Florida because of the standards that are being used, the different standards across the counties – remember, Broward was using one and Palm Beach County another. And they said, “For that reason, we’ve got to establish fair standards.” And that actually led to the result that we got because there wasn’t time to actually do a recount under those standards.

Here we’ve got 40,000 votes, 5-10 times more votes. It seems to me that the folks who wrote Bush vs. Gore should have exactly the same concern, except maybe 5-10 times more.

Q: Do you have an opinion on some of the apprehension that has been expressed by computer scientists about the potential for cheating on electronic voting systems?

A: Every kind of system is vulnerable to some forms of cheating. One of the reasons we went to punch-cards in the first place, and lever machines especially in the first part of the last century, was because people were worried about stuffing the ballot boxes which was easy to do. So every system has its vulnerabilities. And those are all things that we should be concerned about. But I will also say that every system has solutions to those problems, which I think can overcome the difficulties.

Q: Have you done any research into the cheating problem?

A: I haven’t done direct research although let me be clear about that, I don’t think the computer scientists who have talked about the possibilities of doing this have done direct research. What they’ve said is they can figure out ways to change that computer code to fool that machine. I don’t think they’ve done any research whatsoever to find out if anybody has succeeded in doing that. And part of my concern with those folks is they know a lot about computers but they don’t know much about voting.

Q: I think their theory was based…

A: Their theory…

Q: on that you can change a lot of different data in a computer compared to the difficulty of changing mechanical data. Computers let you do everything faster.

A: You’re getting a little bit deeper into this than I want to get, but let me just say the following. Of course everyone in the room knows that all votes now, except paper ballots, are counted by computers. So if your worry is that computers can be hacked and problems can occur, you’ve got to worry about punch cards, you’ve got to worry about optical scan, you’ve got to worry about direct-record electronic. The only system that doesn’t have that is vote-by-hand, where they’re counted by human beings.

So this is not just a problem that electronic systems have. I think that it’s a bit of an exaggeration to say that they are vulnerable, they’re not really that vulnerable compared to other systems.

And let me just say another thing. Electronic systems do a heck of a lot better job at counting the votes. They do a fabulous job at counting the votes. They also allow multiple languages to be used and they also allow for you, at the end of your voting, to have a summary of what you did and to ask you if that’s what you mean to do. As a result, that can cut down on any problems you have in voting.

Now optical-scan have some nice features too, I’m not trying to make a case for electronic versus optical scan. I’m just saying that the newer, non-punch-card systems have a lot of nice features that we should probably take advantage of, and not just decide not to use because of some computer scientist’s speculations about what could go wrong.

Q: [inaudible]…what were you doing before? Estonia?

A: I’ve done a lot of different things and I am easily bored so I seem to move from thing to thing, but I hope completing the things that I already started.

The reason that I got involved was because in November 2000 I saw what had happened in Palm Beach County. I did some research the day after the election and posted a memo to the web about that and said, “Oh my gosh, it looks like something happened in Palm Beach County, specifically that people who meant to vote for Al Gore ended up voting for Pat Buchanan. And my paper was one of the first on the web that not only showed that there was an anomaly in terms of more Buchanan vote than you expected but that there was real evidence that those were Gore voters who had made that mistake because of the butterfly ballot. That got me very interested in voting systems.

To go back a step, I’ve had a long-standing interest in people’s participation in politics in America, why people vote or don’t vote, why they get involved in community work and all sorts of things like that. I’ve been concerned that there’s tremendous biases, that folks at the upper end of the income spectrum are about five times more likely to participate in politics than people in the bottom fifth of the income distribution. It seemed to me that that might have an impact on politics as played that if rich people have a lot more voice in America, perhaps they are heard more than poor people. And so I started looking at this data and I was shocked to find – maybe I shouldn’t have been shocked – that in fact once again it looks like poor people – people with lower education and minorities – are having their votes not counted. So again, another instance of their voice not being heard, but in this case in a particularly cruel fashion. These are people who came to the polls. They wanted to vote. They tried to vote. And the machines didn’t do it.

Q: Is March soon enough for all the counties to be converted to more modern voting systems?

A: Yeah, one of the things that has been brought up is an earlier agreement between the state and some of the plaintiffs in this case that in fact punch cards would be replaced by March 1st, 2004. The reason that agreement was reached was because at the time that date was reached, when the agreement was made the only outstanding state election was November 2002. The feeling was that it was impossible to implement new systems for November 2002, because it takes a year and a half to two years to put in new systems. You’ve got to put out the contract on them, you’ve got to buy them, you’ve got to train new people, and all those things. So the date of March 2004 was chosen. At the time, nobody thought a recall would occur. One of the issues here, in the court case, is whether or not the plaintiffs could revisit this issue given that things have changed. And the court decided that yes, the plaintiffs could revisit the issue, partially because an additional plaintiff had been added. And that was the NAACP and the feeling was that the NAACP had a distinct and different interest in the original plaintiffs. Specifically, they were concerned about minority voters, and that since this case was so much about minority voters that that gave this case the character of being a new concern and therefore the whole issue should be revisited. In fact that’s the reason they came to the decision they did: “We think it’s a problem so let’s postpone.”

Q: There’s two propositions on this…

A: Ah, that’s another story, an interesting one. They’re moved as well. It’s interesting that the court said, in their decision, that it’s really easy to make the decision about the initiatives. They should be moved, no problem. We don’t have a hard time with that one and here’s why, they said. First, they were initially scheduled for March 2004. The only reason they got brought back to October is simply because this new election came up. Second, neither of those initiatives, if passed, would be implemented before 2005 so there’s nothing that’s pressing about getting a vote on those initiatives. Third, they said, one of those initiatives is racially charged and because it’s racially charged, it’s especially important to make sure the voting is as fair as possible and that minority votes are counted. For that reason, they said it’s easy, slam dunk, to decide to move them.

Then they said, “But there’s a harder issue here…should we move the recall?” That’s a tougher one, because we have competing interests. Timeliness is important and the constitution seems to suggest that maybe you should push forward as fast as possible, but on the other hand you’ve got this competing interest in equal protection for votes. And they said it’s a hard one to weigh but equal protection for votes, in the end, is the one that comes down and rules the day.

That’s what we’ve got here, competing interests. And one of the things I should say is that I’m saddened to see the recall forces, who have pushed forward with the recall because they care about democracy and they care about people’s voice being heard, not getting on side with this and saying, “We care about democracy, we care about people’s voice being heard, we care about these votes being counted as well.” Instead they’ve written a series of amicus (friends of the court) briefs in which they’ve argued that no, this doesn’t happen. They’ve argued that these are minorities who abstain. These are people who’ve come to the polls and really abstained. Well, minorities really abstain at a very high rate…they come to vote in a presidential election or a gubernatorial election, they get there and they decide to abstain. There’s no evidence that’s true. We’ve shown again and again that amazingly, when we change systems, those people no longer decide to abstain. So what’s the best inference? The best inference is not abstention. It’s machines that throw away votes.

Q: Is there a reason why, if they knew there was going to be this issue about polling and polling places and voting machines, that the Secretary of State couldn’t have called this as a mail-only mail ballot for the election?

A: Wow, that’s an interesting notion. I don’t know. Perhaps that’s possible. That’s not something that we’ve done in the past in California. Oregon does it, some say with great success. I think it’s a complicated issue. It does have some complexities to it in this state. I don’t know when registration closes in Oregon but you do have the problem that registration closes 15 days before the election (in California), and so you’d have to quickly get the ballots out to people, and then if you mail them out the day after you got everyone registered, would you really necessarily have time for people to vote. It may have been impracticable.

You should ask Kevin Shelley that, actually, it’s a good question!

Q: Who can appeal this? Would it be Shelley or [recall initiator Ted] Costa?

A: No, Costa is amicus to the court; he’s not allowed to appeal this. It’s the state that can appeal this. I’m almost certain that they will, that Kevin Shelley, secretary of state, will. I may be wrong but that would be my prediction.

OK, thanks everybody for coming, it’s a pleasure.