01 Jan 2014
January 1, 2014

#VAAG Recount Lessons

0 Comment
1956 Recount

I had the opportunity to spend 2 1/2 days before Christmas with the Herring campaign recount team at the Fairfax County Courthouse – ground zero of the #VAAG recount.   After the official canvass of the Virginia Attorney General’s race, Democrat Mark Herring lead Republican Mark Obenshain by 165 votes out of nearly 2.2 million cast.

The Fairfax recount process started before 7 am on Monday, December 18, 2013 when the teams assembled – 40 groups of election officials composed of one Democrat and one Republican, and one observer for each party per group.  It took until about 10 am for work to start, but then 28 teams worked in the main room, 6 teams in an auxiliary (overflow room), and 8 teams in the Central Absentee Precinct (CAP) room in the basement.  I was the site coordinator for the CAP room.  The recount was overseen by two members of the Fairfax County Electoral Board – Republican Brian Shoeneman and Democrat Seth Stark, and staffed by the Fairfax County Registrar’s Office (with assistance from the Clerk’s office).

Fairfax cast over 300,000 votes in about 240 precincts, and these had to be recounted in various ways:  For the relatively few votes recorded on electronic touchscreen machines, the recount  meant checking the tape that reported vote totals; the numerous optical scan ballots (where you fill in the little oval next to the name) were re-run through a scanner/tabulator machine programmed to read only the votes in the attorney general’s race and to reject overvotes (where both candidates were marked) and undervotes (where the scanner did not record a vote) —  those rejected ballots, as well as ballots with write-in votes, and ballots that were damaged so the machine did not accept them, were recounted by hand.  In the CAP, we also had military and overseas ballots that were not of a regulation size and could not be read by the scanner and these also had to be read by hand.

We had a couple of issues when we started – the CAP room, although next to the cafeteria, was very small had no chairs for observers.  We complained – invoking the Americans with Disabilities Act – and extra chairs were brought in.  Then one election official asserted that only observers assigned to tables could be in the room and coordinators could not – she was promptly overruled by Brian/Seth.  The group settled into a routine relatively quickly, counted on Monday and Tuesday, checked the results on Wednesday and the official results went off to the State Board of Elections (“SBE”) and then to the Recount Court on Wednesday afternoon.  I suspect that by the time they got to the Court, Obenshain had conceded.  Ultimately, the statewide margin was certified at 907 votes.

The process provided that if the counters did not agree on a ballot, it would be sent to the Court in Richmond to reivew.  There were a total of only 13 challenged ballots in Fairfax and 122 statewide (most of these would have been counted for Herring).  In the CAP room, there were few disputes – a personality conflict between two election officials the first day resulted in the only two challenges from the CAP (the officials were reassigned to different teams the next day).  Observers were not supposed to interfere with the process or even talk to the counters, but conversations ensued, with only a couple of dust ups.

Some lessons:

Elections aren’t perfect.   We strive to make them so, but at every stage of the process human, mechanical, or just random error creeps in.  The state-wide recount total was 526 votes higher for Obenshain and 1,268 votes higher for Herring, so it came up with about less than 2,000 net candidate vote increase, out of 2, 209,283 total votes.  In Fairfax, the Herring margin increased by 368 to 67,906 votes.  The primary difference was that optical scan machines did not count “undercounts” the first time around, meaning that ballots that were marked with too light a check mark, or circled, or marks made outside of the oval, did not count.  But that doesn’t explain everything.  In the CAP, the two candidates total votes combined was 40 votes less than the initial certified result.  How did that happen?  Counting errors the first time around, maybe scanners reading and then rejecting a ballot which was fed in twice, ballots somehow misplaced?  The CAP finished counting all the ballots brought from the Clerk’s Office vault for the 11th CD, and we were two thousand votes off.  Staff went back to the vault and promptly found another two boxes of ballots.  About 25 additional ballots were subsequently located.  When we were done, the vault was empty.

Optical scan ballots work pretty well.  The SBE provided very good guidance on which peculiarly-marked ballots should and should not be counted.  See 12/14 post.  In the CAP, the most common error was filling in the (D) next to Herring’s name, rather than the oval.  Except the one voter who filled in the “D.” in “Mark D. Obenshain” (counted as an Obie vote).   Also note that the SBE apparently adopted guidance, which never made it to the basement, that were a candidates name is marked and written in it counts.  (2 Obie votes – one counted/one not).  There were very few mismarked ballots.  A bigger concern is the whole undercount issue – those votes were caught in the recount, but the optical scan machines don’t tell the voter that a vote has not been counted in any given race.  Recounts are expensive and can be traumatic – these votes should have been caught the first time.

Touch screen computer voting (direct-recording electronic or “DRE”) works well, we think.  DRE machines give the voter a screen summarizing the vote before pushing the final button, and so reminds the voter of any race in which he or she did not vote.  On the other hand, some voters leave before pressing the final buttons, the machines (esp. in Roanoke) may not be not calibrated properly so votes are given to the wrong candidates (rare, but happens), and machines can lose power, although allegedly not votes.  The biggest problem with DREs is that other than rechecking the totals for math errors, there is no way to know if votes have been lost or misrecorded (perhaps intentionally).  Some computer-knowledgeable folks insist that this could never happen; others insist it can.  Currently state (and DNC) policy is to phase out the machines.  A DRE machine that records the last screen that lists all the votes could generate “ballots” could be printed and reviewed in a recount.  On the other hand, the biggest voting problem we had in 2012 was long lines, and using optical scan ballots (whcih can be read later) clearly speeds up the voting process.

Military and Overseas Citizen Absentee Mail-In ballots work pretty well.  Voters abroad can request an absentee ballot, receive it in email form, and then print it, mark it, and send it in to the registrar.  The paper size can vary so, these are read by hand.  We had two ballots that had the candidates from 2009 – McDonnell/Bolling/Cuccinelli.  These apparently had been counted in the canvass, although we can’t be sure for who.

You can vote straight ticket in Virginia (under rare circumstances) – Military and overseas voters can request a Federal Write in Absentee Ballot (FWAB) – and write in candidates or parties, and if you write in, e.g. “Democrat” that counts as a vote for Democratic nominees in all offices.

CAPs are different – at least in Fairfax.  Fairfax absentee voters are supposed to be segregated by congressional district – the 25,000 ballots included ones that were mailed in and ones that were cast in person at eight satellite locations – each of those locations contained ballots for each of the 3 CDs.  The recounters first task was to open all of these separate envelopes and create piles by congressional district – this made it impossible to trace the initial source of the ballot, but it probably didn’t matter.  CAPs don’t have poll books and the reporting forms are prepared for usual precincts – there are some specific CAP issues that Fairfax will likely address.

The canvass is critical.  The “unofficial results” that are reported on election night are exactly that – they contain numerous errors – uncounted ballots, transposed numbers, misreported results.  Over the next week each local electoral board conducts a canvass and totals the results.  The canvass is where errors are to be corrected.  This year, because of the closeness of the AG race, and the vigilance of observers like Dave Wasserman @redistrict, Ben Tribbet @notlarrysabato, and Donald Brownlee, I think extra care was taken and thus the difference between the canvass and recount results was minimal.

Virginia has too many ways to vote.  Each of Virginia’s 137 cities and counties can choose their own voting systems (within SBE certification guidelines), and that results in a great diversity of systems.  The SBE has approved numerous different machines – both optical scan and DRE.  Alexandria and Chesapeake had to hand count their optical scan ballots because their scanners could not be programmed to read only the AG’s race and reject undervotes (which then could be handed counted).  The different machines lead to different technical difficulties on Election Day and thereafter.  A centralized system should result in the state using its buying power to reduce costs of equipment purchased and standardized techinical issues.

That Contest thing needs to be revisited.  The Virginia code allows a losing candidate to contest an election and have the Virginia General Assembly pick a winner.  See 11/29/2013 post.  The Code, however, does not provide standards as to what constitutes a valid grounds for contest.  Some have suggested that this power be given to a non-partisan group, e.g. Judges.  I disagree – we need non-partisan redistricting, but if a small group is going to decide an election, the voters should be able to hold them accountable.  We had one group of judges (The US Supreme Court) decide the 2000 election; not again.  Changing the statute to provide clear high standards (e.g. requiring criminal activity that changes an election result) should be considered.

We had a great group of people – thanks to Kevin O’Holleran, Adam Zuckerman, and Megan Tyler of the Herring campaign; DPVA counsel Steve Cobb and Liz Howard; Kip Wainscott and the excellent Perkins Coie legal team; John Ferrell, Ginny Peters and all the Fairfax Democratic volunteers, and, of course, Mark Herring.  Also thanks to Brian/Seth and the Fairfax registrar’s office staff and volunteer election officials.