SF Open Source Voting TAC

Draft Meeting Minutes: June 14, 2018

Order of Business

1. Call to Order & Roll Call

Chair Jerdonek called the meeting to order at 6:04 p.m. Present: Members Hage, Jerdonek, Kattouw, Wasserman. Member Bafundo had an excused absence. Also present: Secretary Chan.

2. General Public Comment

None.

3. Approval of Minutes of Previous Meeting

This item was postponed to the next meeting.

4. Administration

Chair Jerdonek reported that at its June meeting the Elections Commission reappointed all TAC members to a second term except for Member Bafundo, who chose not to reapply because his work travel has conflicted with meetings. The application period to replace his seat will open sometime after the election.

It was decided that Member Wasserman would represent TAC at the Elections Commission’s July meeting. The next TAC meeting will be held on Tuesday, July 10. Secretary Chan will confirm the availability of the room. Member Kattouw said that the next couple months will be very busy for him because of work and personal responsibilities, so he will not be able to take on much work for TAC during this time.

Chair Jerdonek mentioned the idea of video recording the TAC meetings for YouTube posting. Member Kattouw moved for TAC to try this at the next meeting if equipment can be obtained. Member Hage seconded. The motion carried 4-0.

Chair Jerdonek brought up the topic of using a custom domain for TAC for its site hosted on GitHub Pages. There was a short discussion of the pros and cons of this. No action was taken.

5. Member Reports

Member Kattouw reported that he was a poll worker for the election. He will be working on a personal project to write code to analyze the RCV ballot data after the election. He was also accepted to be a speaker at a conference on September 27, 2018 in St. Louis called Strange Loop. He will do a presentation similar to the one he did at LibrePlanet. Chair Jerdonek moved and Member Wasserman seconded to approve Member Kattouw speaking as a member of the committee.

Member Hage discussed election results he reviewed and some problems he encountered in gathering counties’ data.

Chair Jerdonek reported that he was also a poll worker and had a high turnout of voters at his precinct. He went to City Hall afterwards on election night. He said there was a technical problem that delayed the reporting of results by a couple of hours. He also said there was a good chance the Department could obtain a collection of digital ballot pictures for the committee as the committee has discussed at previous meetings.

Chair Jerdonek also reported that Maine just used RCV in their state election and voted to keep using it in future elections. The coverage of RCV by the San Francisco Chronicle has been negative.

On the funding front the City has only committed $300K to the open source voting project so far this budget cycle. The California Clean Money Campaign has on record commitments of support for future funding from Mayor-elect London Breed, Supervisor Malia Cohen, and others. Clean Money will be working in the coming weeks on getting $4 million in additional funding for the next fiscal year.

There was a discussion around how different types of ballots were dealt with in the voting process, how the vote records are relayed to the central office for tallying, and what percent of ballots had to be handled manually to resolve whatever could not be handled manually by the machines.

Member Kattouw said that he could add some pros and cons to the text that Member Hage presented at the last meeting regarding machine-marked ballots.

Public comment:

Mr. David Cary discussed Alameda County’s RCV reporting. He mentioned that the vote center in City Hall had long lines with wait times upwards of forty-five minutes for voters.

6. Voting System Component Development

Member Hage presented a demo of a results reporter that he and Chair Jerdonek started writing and pushed to GitHub. It has a number of features including the ability to output results multi-lingually, but it is not finished yet. The Department might be able to use it as soon as the November election. It is meant as a proof of concept. Part of the intent is to show that Slalom’s cost estimate for the results reporting component was too high and that, if done in an agile and open-source manner, the rest of the system could potentially be developed for a cost much lower than Slalom’s estimates.

It will be an ongoing task to refine and expand the functionality to be able to produce all the reports that the Department currently produces, including reports that the Department doesn’t currently do automatically. It was proposed that each month one person could be designated a lead in choosing who should work on the code, and then rotate the task to different committee members each month. The proposal was to put it as a repository in GitHub under TAC’s account with name “osv-results-reporter.” Member Hage responded to questions about how the input file is formatted and organized.

Public comment:

Mr. Jim Soper said that people in the election integrity community would like to see the precinct reports in spreadsheet format. He said there was a state law requiring that the data be available in such a format within 30 days after the election.

Mr. David Cary suggested they also demonstrate other aspects of the agile process and how it can apply to this project. He mentioned the continued work by NIST on setting up standards so that data gathering might be easier.

There was a short discussion regarding how to proceed with the project, including using a “gatekeeper” to oversee who was working on the project between meetings, allowing independent submissions, or giving separate pieces to individuals to develop, that can then be compiled and reviewed by the other members during meetings.

Chair Jerdonek moved that the Committee create a GitHub repository for an open source results reporter called “osv-results-reporter,” using the code presented tonight, and adopt this as an official TAC project, with the added condition that this action only be valid if the copyright for individual contributions by committee members can be retained by the individual committee members. Member Wasserman seconded.

Public comment:

Mr. David Cary suggested that the code be ultimately subsumed in the code used for the open source voting project and that the City be the copyright holder.

The question of how to do copyrights was discussed, including such issues as GPL, contributor license agreements (CLA’s), and only allowing contributions from members of the public if the copyright is willingly transferred to the City. Does the Committee’s intellectual property work qualify as owned by the City, or is it the copyrighted property of the individuals in the Committee? This question will be investigated further. Member Kattouw and Member Wasserman will bring language that could possibly be used.

Member Hage volunteered to be the gatekeeper for the next month. Member Kattouw moved that he be appointed, and Member Wasserman seconded. The motion carried 4-0.

7. Project Background and Terminology

Member Kattouw summarized the two diffs he prepared for the agenda packet. Chair Jerdonek moved to accept the patches. Member Hage seconded. The motion carried 4-0.

8. Equipment Decisions and Implementation Plan

Member Kattouw reviewed his “diff” for this item, which included pros and cons. Suggestions were made to revise some of the language around verifying machine-marked ballots. Member Hage moved to accept, and Member Kattouw seconded. The motion carried 4-0.

Public comment:

Mr. Jim Soper discussed Los Angeles County’s process of scanning QR codes on their ballots, which he said aren’t voter verifiable.

9. Committee Recommendations

There was no discussion on this item.

10. Slalom Report

Member Kattouw’s comments on the Slalom Report, which were included in the agenda packet, were reviewed and discussed. It included some more prominent points of contention about the report, but didn’t include a complete or in-depth analysis of each point.

Member Wasserman wanted “OSI-approved license” to be used instead of “a proven license.”

The Committee discussed various issues around licensing, software ownership, and if there are examples of governments having licenses for voting systems.

Slalom’s cost estimates were discussed. It was noted that the costs of open source weren’t consistently compared to the cost of the current system for all components. Moreover, the costs weren’t constructed based on any analysis. Rather, they were generic, boiler-plate presentations. They also didn’t compare the costs of current open source voting related projects in government use, including Colorado and New Hampshire, for example. Further evidence of the unreliability of the estimates is that the cloud-based hosting estimates were outrageously high compared to current, actual costs.

Member Wasserman asked if TAC should write a succinct response of these points to the Elections Commission. Chair Jerdonek said he will be introducing a resolution next week to the Commission recommending a rejection of certain aspects of the Slalom report. The TAC could make a preliminary statement at that meeting. Member Wasserman can highlight the points and emphasize that the TAC seriously questions Slalom’s conclusions.

There was a short discussion with Mr. David Cary about the cost of hardware scanners, either from Dominion or off-the-shelf.

Public comment:

Mr. David Cary commented that his last statement to the Commission to reject the Slalom report did not come with much specific justification, but his document included in the agenda packet attempts to provide some of that. He encouraged the Committee to give some high-level feedback to the Commission on the report.

Member Wasserman raised the question of the impact it would have if the Committee recommended rejecting the Slalom report. Would it take the open source voting project back to square one? Chair Jerdonek felt it wasn’t necessary to come up with a total estimated cost for the project because that didn’t coincide with the agile development method being advanced, where the modular development of components provides a better means to estimate and control costs at each stage.

Mr. Jim Soper said that the OSET Foundation could probably provide decent cost estimates for an open source voting system. He commented that New Hampshire used a derivative of Prime III, which is not open source. It passed certain tests in Ohio, but it is not totally certified. He believed that the Slalom report did damage to the possibility of the State legislature funding the project, due to the report’s high cost estimates.

Mr. David Cary reminded the Committee that at the March Commission meeting, there was a motion to move forward with getting funding and doing the next steps for developing the system. The budget request put forth by Director Arntz was not the $4 million requested by the Commission. There seems to be a disconnect between the Commission’s policy and what is actually happening. There needs to be a stronger statement about the Slalom report. He didn’t think there was anything new in the report beyond what was already contained in the Commission’s 2015 resolution.

Member Wasserman commented that the only cost numbers that people can consider are what Slalom presented. Mr. David Cary said that in 2015 certain groups presented various cost estimates to the Commission.

Chair Jerdonek mentioned that DHR has a technology project to improve the City’s hiring process. It will use an agile approach, has a project manager, and is a multi-million dollar project very similar in scale to the open source voting project, yet COIT gave them $1 million to begin their work.

The Committee crafted the following written statement to send to the Commission (at 4:33:00 of the meeting audio):

“Certain inaccuracies and basic omissions in Slalom’s Report lead us to seriously question the report as a whole. We recommend not relying on it for determining the future direction of the project. A few of the more glaring examples include: (1) None of the costs are realistically substantiated. They are based on generic cloud-based applications and not specific to a voting system application. (2) $1 million for cloud hosting is an absurdly high figure, orders of magnitude greater than what it would actually cost. The estimate shows a fundamental lack of understanding of what a voting system entails. (3) No consideration was given to open source election software recently developed or used by other states, including Prime III used in New Hampshire and Colorado RLA used in Colorado. The development process that they described is not in line with open source development processes. We are happy to substantiate these points further if needed. We encourage the Commission to assert its previous resolution to follow our recommendations.”

Member Kattouw moved to adopt the statement, with Member Wasserman seconding. The motion carried 4-0 vote.

11. Topics for future discussion

Public comment:

Mr. Jim Soper suggested looking at AB 2125 regarding risk limiting audits.

Adjourned at 10:45 p.m.