On September 15, 2009, primary day in New York State, paper ballots and scanners were used for the first time by counties taking part in the state’s voting machine pilot program. The pilot allowed Boards of Elections to gain important experience preparing and deploying the new systems in a real election. Of the 47 participating counties, 18 chose to use the scanners county wide. The remaining counties chose a limited number of polling places to roll out the new systems, deploying them in anywhere from 1 to 248 polling places. 15 counties chose not to participate in the pilot at all, including the state’s largest Boards of Elections in Westchester, Nassau, Suffolk, and New York City. The decision by these counties not to participate in the pilot is proving to be a questionable one given how much experience is being gained by this limited roll out.

Having read through several reports on the pilot and spoken with several county election commissioners, observers, and poll workers, it seems that the primary day phase with its extremely small number of races, generally ran quite well. And true to the goals of a pilot, it also turned up some issues and problems that will need to be addressed in the November general election, and in 2010 when the entire state makes the switchover. Over the next three posts I’ll comment on three areas where valuable lessons can be learned from the first phase of the pilot – machines and ballots, privacy concerns, and training.

The Machines and the Ballots

As far as I can tell, the new ballot marking devices and scanners worked without problems although not without a few caveats. In terms of the vote count itself, the machine results matched those of the state’s required 3% audit. I talked at length with observers and participants in Putnam County, where a Court granted a motion for a 100% hand recount of the paper ballots in the Republican County Sheriff’s race. Independent observers report that the recount was well managed, with two teams each consisting of 2 Democrats and 2 Republicans inspecting every ballot and agreeing on the voter intent. Only 5 or 6 votes out of a total of 4,517 cast on the scanners were changed as a result of the hand recount, leaving the outcome unchanged. All of these were the result of ballots marked in a way that the scanner did not recognize, but which clearly indicated the voter’s intent to the recount teams.

And that’s the first caveat. It’s vitally important that the State establish what a valid vote is, particularly for write-in votes which can be difficult to interpret. The State Board of Elections has made a start at this with regulation 6210.13, “Standards for Determining Valid Votes“, but it does not go far enough. Critically needed in the regulations are illustrations, but at the present time the official version only has a blank section 6210.16, “Ballot Examples for Counting Paper Ballots”. The Board needs to get that illustration section completed and make sure the standards for interpreting ballots are disseminated and understood around the state.

New York’s scanners are required to accept a reasonably large range of marks – check marks, crosses, partially or overly filled in bubbles should all correctly register. If the mark is partially recognized by the scanner, it will return the ballot to the voter noting that it is ambiguous and giving them a chance to correct it, but there will always be some voters who mark ballots in ways the scanner can’t register. Presumably, as voters become more accustomed to using paper ballots they will mark them more consistently, but there clearly is a need for thorough circulation of training materials to the public about using the new systems, such as New York makes available here.

Another ballot design issue that popped up in more than a few places relates to the party icons New York State requires next to candidate names. The icons are very small, and some have a bit of white space in the center. This led some voters to incorrectly fill in the party icon rather than the blank oval. Some rethinking of New York’s ballot design is clearly in order, but a real overhaul, which should eliminate the state’s user-unfriendly full face ballot, is going to require changes to state election law.

Long Start Up Times

A recurring report on the Dominion ImageCast scanner/Ballot Marking Device was that it took a very long time to boot up, typically around 45 minutes but sometimes more. Many poll workers, not expecting the long start up sequence, attended to other set up procedures first and turned on the machine shortly before the scheduled start of the election. This led to late starting times in many poll sites. The long start up sequence of the ImageCast was a known issue and a fix has been proposed by Dominion which must be approved by the State Board of Elections. In the meantime it’s important that poll workers be made aware that the ImageCast has to be switched on early in the setup process, and not close before the start of the election.

Part 1 Summary

All in all, this first very limited use of a new voting system worked pretty well as far as the machines and the ballots go. As noted above, there is definitely a lot of room for improvement in terms of ballot design, and the general election in November will tell us more about machine performance when turnout is heavier.

But a real concern that came up again and again from all over the state had nothing to do with the voting equipment, but rather with voter privacy while marking, carrying, and scanning ballots. More on that important issue in the next post.