My Post on Daily Kos and MyDD
This is the full text of my post as it appears there. Some of the comments on the daily kos post are worth reading, and I also have a poll up there, so do visit the link, read the comments, and vote in the poll.
The key protagonists:
Rush Holt, a Democratic US Representative from New Jersey, formerly Assistant Director of the Princeton Plasma Physics Lab. One of very few scientists in Congress, he often tackles related to science and technology. He was one of the first members of Congress to bring up the problems with electronic voting after HAVA, and in 2003, filed a bil requiring voter verified paper trails which got over 140 co-sponsors from both parties, but was blocked in committee (the committee in question was chaired by Ohio Republican Bob Ney, often referred to as “the Representative from Diebold”).
Bev Harris, founder of Black Box Voting. A citizen activist who was horrified by what she learned about elections, fraud, and computer voting after the 2000 election, she has worked tirelessly to investigate and bring problems to light. Black Box Voting does the hard work of visiting polling places and elections offices, demonstrating security flaws in computer voting machines, filing requests with courts, and finding out what is and isn’t happening. Bev goes around the country giving amazing presentations about what she’s found.
Last year, Rep. Holt filed H.R. 550, a bill which would:
- Require paper ballots or voter verified paper trails, and ban the use of computer voting machines that don’t provide them.
- Set a minimum audit requirement. At least 2% of precincts, randomly selected, must be counted by hand, to compare with the computer count and catch possible problems.
- Open up the process in several other ways, such as require that source code used by voting machines be public, and require the Election Assistance Commission to open up its bidding process.
Black Box Voting opposes Holt’s Bill
This Monday, Bev Harris wrote an op-ed piece opposing H.R. 550, representing Black Box Voting’s official opposition to the bill. She wrote,
Like an antibiotic that’s too weak, we believe that H.B. 550 will create a more resistant strain of election infection. Like a placebo, people may think the election system is getting well when in fact, the medicine is only a sugar pill that makes everyone think it’s better. For a minute.
Black Box Voting and other critics of H.R. 550 from a similar point of view, make several specific criticisms of the bill, including:
- A voter verified paper audit trail (VVPAT) added to touchscreen voting machines (DREs) does little good because it often fails, voters often don’t look at the paper to verify their vote was recorded correctly, and the paper tape is hard and time-consuming to count.
- The 2% audit requirement is too low.
Harris’ essay goes into broader territory, explaining how political issue campaigns have both an “inside game” and an “outside game”. She describes how these two differ, how they can work differently on the same issue, and how they can affect each other. She seems to be implying that in this case, Holt is playing the “inside game” but undermining the “outside game” on election reform – though she doesn’t say that directly. However, she clearly feels that Holt’s bill is a strategic error, for two main reasons:
- It is a “placebo” that will not solve the problems, but make people think the problems have been solved, undermining efforts at real reform.
- Audits and computer security are the wrong path. The only real solution is to ban secrecy and make the entire vote counting process open, which includes requiring hand marked paper ballots. Holt’s bill does nothing about this, so it’s going in the wrong direction.
Rush Holt Responds
In An open letter to voters clarifying H.R. 550 two days later, Rep. Holt praised Harris and Black Box Voting, but pointed out several misrepresentations of H.R.550 that he saw in Harris’ article. I urge you to read his letter in full, but in summary, these were his corrections:
- The 2% audit requirement in his bill is explicitly a minimum requirement, and does nothing to prevent states from setting higher standards. Currently there’s no minimum federal requirement, so states can (and generally do) eschew audits altogether.
- His bill doesn’t simply call for DREs with a VVPAT as the solution. It explicitly mentions hand marked paper ballots and Holt favors them. What his bill does do is ban the use of DREs without a VVPAT, and requires that the VVPAT be considered the ballot of record (meaning that if a count of the paper votes differs from the machine count, the paper is considered the official count).
- His bill would not lead to delays in improving voting standards. H.R. 550, filed in 2005, calls for its new standards to be in place for the 2006 elections.
Rush Holt urged election reform advocates dissatisfied with the provisions of H.R. 550 not to oppose it, but to try to amend it with improvements. He urges all of us to continue lobbying for his bill.
An Exchange with Nancy Tobi of DFNH
Democracy for New Hampshire is a grassroots organization formed by supporters of Howard Dean in New Hampshire, after his campaign for president ended. It is loosely allied with Democracy for America groups all over the country that came out of the Dean Campaign. Nancy Tobi of DFNH has been very active in election reform work, and indeed, when I saw Bev Harris speak in Cambridge, MA last year, DFNH was one of the main sponsors bringing her here.
Monday, the day Bev Harris posted her op-ed, was also the day TrueMajority emailed all of its members asking them to call their US Representatives in support of H.R.550. That day, Nancy posted on the DFNH web site that TrueMajority’s Holt Bill action alert should be reconsidered. She wrote,
The Holt Bill is well intended, but unfortunately, it is not just about paper ballots; it includes several dangerous provisions that are not good for our democracy at all. Consequently, there are many election activists, including most of us on the DFNH Fair Elections Committee, who do NOT endorse the Holt bill as written.
She included links to several posts she had already written about the flaws in this bill.
Holt responded, again clarifying and making corrections. I urge you to read her post and his comment, which is the first comment below her post.
I Support H.R. 550
As I wrote in a comment on DU, I think Harris, Lehto, and Tobi’s line of argument boils down to “if it’s not good enough, then it’s damaging” and I really don’t buy it. I also think that in some cases, they misunderstand the bill or interpret it unfairly. Holt’s bill does nothing bad.
Critics point out many ways in which it will fail to really fix things, and I agree with many of their criticisms. But it provides several important improvements, such as:
- Going from no audits to some au
dits. - Going from no paper ballots to some form of paper ballots. Even roll-tape VVPAT, with all its flaws, allows officials, courts, or activists, to order a meaningful recount, and remember, the Holt bill doesn’t require DREs with a VVPAT, it merely disallows DREs without a VVPAT.
- Going from closed source code to requiring that the code be open to the public.
I want all of these things.
I also want other things, such as banning DREs (except as optional ballot-marking machines for the disabled); full hand counts or a requirement that every set of ballots be counted separately by two different devices from different makers; a requirement that any person be allowed to observe at the polling station and the counting process without needing credentials; quick posting of precinct results on the door and on the Internet before transmitting anything to central tabulation; raw ballot data provided online in a timely manner… I could go on.
The Holt bill doesn’t do much of what I want. But what it does do, I do want. The things it does are improvements. We’ll need to keep working. Holt knows, I think, that we can’t get everything we want through Congress now, but he feels that his bill pushes the envelope and strikes a right balance between what we want and what we can get. I think we need to support it.
I don’t buy the “placebo” argument. I think it’s nihilistic. I file it with “things need to get worse before they can get better”. It’s the view that if you take a small step forward, you deflate the movement’s energy or the public’s appetite for taking more and bigger steps. Things don’t work that way. Victory begets victory. Steps forward are opportunities to build organization and coalition. They make the next steps smaller than they’d need to be now, which makes them easier. And they take nothing away from the energy of an informed, active grassroots movement that wants to do more.
I urge all election reform advocates who oppose computer voting and want to move towards a more open, verifiable, trustworthy election system, to support H.R.550. And I also call on all of you to understand that H.R.550 is a small step, and to keep on working for more.
cos says
Oops, Bob’s cut & paste didn’t quite work! The actual post has paragraphs, links, bulleted lists, whitespace, blockquotes, the works. It’s much more readable that way đŸ™‚
<
p>
I’ll try to fix it later.
cos says
Yay, that was easy, I just copied the HTML over and it worked.
stomv says
because that’s exactly what voting is. You’ve got a whole bunch of data sources (people) and you’re attempting to measure a datum from each source — which person they want to vote for in a particular election race.
<
p>
So, when you’re measuring, there are different kinds of errors:
1. Random, uncorrelated errors. If you had to measure 10,000 twigs with a ruler, you’d make some mistakes. An extra inch here, an inch short there. The good news is that, on average, these errors are less and less likely to impact what you care about — the average length of a twig.
<
p>
Reducing errors like this make your calculation (average length of a twig) more accurate.
<
p>
2. Correlated errors.
Let’s say you’re measuring stick is a 12″ ruler. If the twig is less than 12″, you know it’s length because the ruler has tick marks. But, if the ruler is more than 12″, you just eyeball the extra length. Furthermore, let’s say you typically guess “short”, that is, you are undermeasuring the twigs longer than 12 inches.
<
p>
This is a correlated bias. You’re only mis-measuring the longer twigs.
<
p>
3. Multiple negatively correlated errors.
Let’s say that, in addition to the situation outlined in (2), the ruler’s tick marks are also wrong. You eyeball a 14.5″ twig as 14″, but it turns out that a 9.5″ twig is measured on your ruler as 10″. So now we’ve got two errors, shifting the errors in opposite directions. In this case, it might be that the errors tend to self-correct; two wrongs making a right. If you get a better ruler with correct tick marks, you’d only fix one of the errors, and actually end up with a less accurate measurement overall (average length of a twig) because the second wrong is no longer available to make the first wrong a right.
<
p>
Why is this important? Well, it’s important to consider what kinds of errors are happening in elections. Are they of type (1), (2), or (3)?
<
p>
Some are type 1. A system doesn’t correctly read the voter’s perhaps incomplete but intended mark. Maybe a mechanical voting machine double counts. These kinds of errors can be eliminated and the system gains accuracy without changing any bias it may have. That’s good.
<
p>
Some are type 2. Maybe people who can only speak and read Chinese tend to vote Republican, but the ballots aren’t in Chinese and therefore these voters aren’t able to correctly implicate their intentions. Now, we’ve got a bias that’s hurting the Republican candidates. Maybe it turns out that people who vote Republican tend to be more comfortable with touch screens (they use the ATM more đŸ™‚ ). Then, perhaps the Democratic candidates would be more likely to lose votes to error. Maybe Diebold has a systemic error in their code that (un!)intentionally tallies more votes for Candidate A than Candidate B, regardless of voter intent.
<
p>
These are the most important types of errors to fix, but also often the hardest to discover, prove, or fix. These are the worst kind of errors because we’re not measuring what the people really want, and we’re consistently wrong in the same “direction”.
<
p>
Some are type 3. The trouble with these are, if you only fix one of the two biases, you’ve moved the scenario from a problematic but more inoculous (3) to a very problematic and harmful (2).
<
p>
So the question is, what kinds of problems will H.R. 550 address? Problems of type 1, 2, or 3? The auditing will help discover problems of all three types, which is good. The open source code is especially helpful for problems 2 and 3, which is especially good.
<
p>
At the end of the day, it seems like HR 550 will reduce both generic errors and some of the real problematic ones, which is most important.