Three Ways to Strengthen H.R. 550

In the spirit of Wilms, I propose three measures to strengthen HR.. 550.

I attended the most recent citizen lobbying effort for H.R. 550 held last week. I believe that what 550 does right makes it our best hope: it prohibits undisclosed software, ITA conflicts of interest, wireless communication devices, or other connection of voting machines to the Internet. I also agree with those who believe that 2% audit provision needs work.

To begin, I think we can establish that it is safe to call for amendment without endangering the bill. Other ER posters have, rightly, trumpeted, Holt's recent open letter to critics of 550. Here is what he wrote about amendments:



I have been concerned that if election reform activists call for changes, the bill's prospects could diminish. Holt's statement is reasurance that calling for committee markup and then amendments is not going to hurt anything. He's talking specifically about audit size, but I don't see how the same principle wouldn't apply to other audit amendments.

Three measures that I believe would strengthen the audit:

First: There must be multipartisan participation. Democrats, Republicans, and small parties should be allowed to send accredited observers to witness every aspect of the audit, from the random selection of precincts to the hand count itself. They should be allowed to bring video cameras with them. Small parties would presumably not be able to send observers to every precinct, but they should be free, under the auspices of the EAC, to observe any audited precinct in the country. The EAC would be the final judge of any reported irregularities in the audit.

As we know, the EAC would have to contract out the audit, and H.R. 550's language requires public bidding. As Warren Stewart notes, this would allow voting integrity groups to bid for the audit contract. That's a commendable measure, but what happens if citizen groups don't win the contract? The only transparency measure that I am aware of in the bill is that it requires public reporting of the audit results. Not very reassuring. The integrity and transparency of the audit must be assured, no matter who wins the contract.

Second: The randomization process should be codified. I have spoken with an experienced software tester about random selection, and he believes that the best way to ensure randomness is to use exiting lottery-ball technology. Random number generators are not ultimately trustworthy, and at the very least create the perception of "black box" selection of audited precincts.

Third: The selection of precincts should take place on election day itself, with audit personnel and observers ready in each county in the country for the selection of precincts. Audit workers should present themselves at the selected precincts immediately after intitial reporting of the results. This is a tight window of time, but custody of the ballots should change after initial reporting, beacuse of the well-founded concerns about memory card alteration by local election officials with access. If the auditors showed up before initial results were reported, then unscrupulous officials would simply be prevented from manipulating those precincts.

The first step to making H.R. 550 work is, of course, getting the committee process moving. So contact your Congressperson, as well as the majority and minority offices of the House Administration Committee, and call for markup:

U.S. House of Representatives Switchboard: 202-225-3121

Committee on House Administration (Majority Office): (202) 225-8281

Committee on House Administration (Minority Office): (202) 225-2061

As you might infer, I have no strong opinion about the sample size of the audit. It does seem to me, without my having any statistical background, that if the precinct selection is observably random, and ballots are in the custody of auditors shortly after the close of the polls, the 2% audit will be meaningful. But whatever you think of sample size, I hope we can agree that these changes would strengthen the bill considerably.

If we can get 550 passed, we can use it as a hammer, much the way you have suggested, to break down the wall separating us from fair elections.

If one can't support 550, then one should solely request HCPB, and be done with it.

I am especially concerned with who wins the contract for the auditing. :scared:

I would also like to see language that OUTLAWS the use of vendors for Ballot Definition Settings (Programming). This is apparently done by many small jurisdictions that lack the ability to do it for themselves. Not that this would serve as a panacea, but hiring a vendor to program an election just isn't sitting well with me.

Along that line, these same jurisdictions contract such services spend a heck of a lot of money doing that (in addition to purchasing, maintaining, storing, transporting, replacing, training workers for, and let's add insuring), these e-systems. Yet these very jurisdictions are probably the ones that could most easily, effectively, and economically employ HCPB.

How many went down the e-road on the assumption that HAVA mandates it?

Does HR 550 specifically say, 2%, statewide, PLUS one precinct per county? IIRC its a maybe....

precinct per county. See Section 5 of the bill. Sec. 5(a)(1) specifies at least 2%, and Sec. 5(b)(1) specifies at least one precinct per county.

SO a big county audits an additional 100 people..... out of say 600k voters.

And in a smaller county, audits an additional 100 people.. out of 100k voters.

I have a source that claims there were 6,310 precincts in NJ, which sounds about right. That would be something like 570 average votes per precinct in 2004. It may vary a lot from county to county -- I don't know.

OK, so say 6310, then a 2% audit is 126 precincts. Looks like NJ has 21 counties, so it doesn't run into the problem Iowa does, where there are lots of very small counties. Looks like Salem is smallest and would be expected to end up with about 1 precinct out of 126 anyway.

(I don't think a likely regulatory implementation of Holt is random 2% plus one per county, because I can't think of any rationale. If the reg went that way, it would more likely be as described by Bill Bored: random 2%, then add one to any counties that came up empty. But nothing in Holt bars a more stringent audit.)

This stuff sort of depends on the scenarios you are worried about. If a hack is implemented statewide, then a 2% statewide audit is likely to catch it (if it can catch anything -- there are lots of details to worry about). Perhaps more likely is a hack that would yield, I dunno, 10% of votes on 10% of machines in a large county? Actually, looking at NJ's demographics, you don't seem to have any really big counties, i.e., the biggest counties seem to be around 10% of the population. So that sort of hack yields about 0.1% of the total statewide vote per county.

Personally, I think that if people start looking at the actual math, the worries start seeming more tractable, which isn't to claim that a 2% audit is bulletproof. There's a lot to consider.

Pun intended.