SF Seal

SF Open Source Voting TAC

Official site of the San Francisco Open Source Voting System Technical Advisory Committee (OSVTAC)


Meeting Minutes: November 16, 2017

Elections Commission
City and County of San Francisco
Don Chan, Secretary
Open Source Voting System Technical Advisory Committee
Christopher Jerdonek, Chair
Larry Bafundo, Vice Chair
Carl Hage
Roan Kattouw
Tony Wasserman
MEETING MINUTES
Open Source Voting System Technical Advisory Committee
of the San Francisco Elections Commission
Thursday, November 16, 2017
6:00 p.m.
City Hall, Room 421
1 Dr. Carlton B. Goodlett Place
San Francisco, California 94102

Order of Business

1. Call to Order & Roll Call

Chair Jerdonek called the meeting to order at 6:05 p.m. Present: Members Bafundo, Jerdonek, Kattouw, and Wasserman. Excused absence: Member Hage. Also present: Secretary Don Chan.

2. Project Management and Procurement

Vice Chair Bafundo introduced Ms. Jessie Posilkin, who was invited to speak on modular procurement and agile practices in government. Ms. Posilkin is an Innovation Specialist at GSA / 18F and a colleague of his. She spoke to the committee via video conferencing. A video recording can be viewed at: https://www.youtube.com/watch?v=aI1glu6O5ic.

Ms. Posilkin began by saying that the principles behind modular procurement are not far from the basics of software development, that is, when you are developing software it is better to use a modular approach. Points made included:

Member Wasserman asked whether you should have the same contractor do several “chunks.” Ms. Posilkin replied that you might not want to have the same contractor. She mentioned that the State of California has a couple of good vendor pools that we might want to try accessing:

The question was raised as to how you get government funders to fully fund something that will be built a “block at a time,” when the ultimate product details and costs are not known up front. She responded that you have to find a “champion” who gets it and has the power to allocate the funds.

If an RFP doesn’t detail outcomes enough, how do you require vendors to do “additional” tasks? She replied that you shouldn’t think of a contract as unamendable. Her experience has shown contracts to all the time have multiple change orders (with costs adjustments) as a common occurrence. She said the most important element is strict monitoring of the contract.

There was a feeling that in order to assure that a definite product was obtained at the end of each modular project, there needs to be a “buy-in” by the City beyond hiring an external intermediary (e.g. a technical lead to oversee the project). Contracts can be written in terms of target capabilities, key constraints, and key outcomes so that trade-offs could be made actively during the development of those increments.

When asked for examples of how specific modular contracts are written, she said it is not how it is written that determines modularity, but rather how you are making a format that allows for agile software development.

She mentioned a report from the National Association of State CIOs that lays out why governments should adopt agile.

There was a discussion about the “granularity” of a contract and how explicit one should be while giving the contractor the flexibility to work without restrictive conditions. The wording of the contract is key, but oversight by the City is critical, and not just contracted out to an outside body for the responsibility.

Chair Jerdonek said that he had a conversation with Robert Henning, an Assistant Director of the San Francisco Office of Contracts Administration (OCA), regarding the project in relation to agile procurement. He was given information on the City codes that guide the procurement process. He suggested the idea of the Committee holding a panel discussions where this topic can be explored with him, the City’s Chief Innovation Officer, and the City’s Chief Digital Services Officer.

Key “take-aways” from the presentation: the committee needs to define the key building blocks or modules, as well as the order of development. The project needs to be approached incrementally and not as a single deployment. There can be an accumulation of user stories that cover all potential scenarios of users (voters), which can then be sequenced to address each particular problem.

It was suggested that at the next committee meeting, examples of modular contracts can be examined and discussed. Ms. Posilkin offered to send examples for review. Member Wasserman said that he would check with Code for America to see if they have examples. The question of putting award limits on each contract was raised. Depending on the amount, the process can be more or less streamlined. For example, if chunks are under $50K or $100K, it might not need an RFP process.

The product of the contract should be a functional deliverable, and not just “progress reported”. How to price such a contract was discussed, in terms of the number of person-years, but no specific figures were agreed on.

Public comment: Mr. Jim Soper stated that the group he is working with is seeking funds from the California Secretary of State. Any funds from elsewhere would be helpful in advancing this project.

The idea of enlisting the help of 18F was discussed. No decision was reached.

3. General Public Comment

Mr. Jim Soper mentioned an article he read that said all Intel computer chips contain a hidden operating system (Minix) on them.

Member Wasserman gave a short explanation of what Minix was (i.e. what it is turned into on that chip).

4. Approval of Minutes of Previous Meetings

The corrections were as follows. For the September minutes: section 3 should read “pull request.” In #4 of the numbered list, the word “ballot” is misspelled. In #11, replace ”reviewed” with “talked through.”

With these corrections, Vice Chair Bafundo moved to approve the draft minutes for the September 21, 2017 and October 19, 2017 meetings. It was seconded by Member Kattouw. Upon voice vote, the motion carried unanimously.

5. Member Reports

Vice Chair Bafundo reported that when he worked on the LA project his team delivered a design document that detailed the philosophy of the project, how it should be built and why. He had reached out to LA to try and get a copy of that document but has not heard back. He mentioned a previous colleague, David McGivney, also involved with that project, who might be called on to give some insight to its inner workings, either through prepared questions, or being part of that panel discussion brought up by Chair Jerdonek.

Member Kattouw mentioned his report at the Elections Commission meeting and some of the discussion items that were dealt with, including a report from both BOPEC and Slalom, and a memo from Travis County explaining why they are not going forward with open source. He also mentioned that he submitted a proposal to speak at LibrePlanet’s March conference. Lastly, he said he heard that Concord, New Hampshire had used the Prime III system for a local election.

Member Wasserman raised the idea of Member Kattouw doing his presentation at a smaller conference as a “practice run” and mentioned SCALE (which will be held in Pasadena before the LibrePlanet conference).

6. Administration

Chair Jerdonek reviewed the meeting schedule for the rest of the year and going into next year. The December meeting will be on the 14th. Member Wasserman will attend the December 20 Elections Commission meeting, Vice Chair Bafundo will attend the January 17 Commission meeting, and Member Wasserman is available for the February 21 and March 21 meetings. For next year the committee will try a second Thursday meeting schedule, but remain flexible if changes are needed.

Chair Jerdonek reported on changes he made to the website. He said there was a suggestion that when changes are made to the recommendations document, the dates of those revisions be indicated as notes within the text.

Vice Chair Bafundo asked if it would be possible to take the document and work on them at the committee meetings. Chair Jerdonek said maybe it could be done a section at a time and do updates at the meeting.

Chair Jerdonek raised the topic of licensing the Committee’s work on the Committee’s website (code and/or content). There needs to be clarification of who owns the work being developed, and how those copyrights are held. Member Kattouw has someone he can ask. As a first step, Chair Jerdonek suggested that the code be issued under the GPLv3 license. He moved that the committee adopt as a policy that the code in the repository be licensed under GPLv3 or if a new version comes out, be upgraded to that version, and that the documentation be clear that contributions are accepted under those terms. Member Kattouw seconded.

Member Kattouw reviewed some of the aspects of licensing, what kind of license and updating or commercializing implications there are.

Upon voice vote the motion carried unanimously. Member Kattouw moved that the recommendations document in the repository be licensed under CC BY-SA 4.0 (the Creative Commons Attribution-ShareAlike 4.0 International License). Member Wasserman seconded. Upon voice vote, the motion carried unanimously.

It was suggested that the Committee’s policy should be to encourage members to license their conference presentations under CC BY-SA.

Member Kattouw moved to make it the policy of the Committee to encourage its members to license presentations they produce about this topic under the CC BY-SA 4.0 license. Member Wasserman seconded.

Member Kattouw said it was conventional to have a CONTRIBUTING.md document. Member Wasserman mentioned the possibility of public contributions wanting recognition. It was noted that the history of such changes is listed in the document so recognition is technically public. He said to satisfy the legal attributions of CC BY-SA, it is only necessary to list all of the authors or provide a link to such a list. If done right, we could reflect the authors of every change using Git.

Chair Jerdonek said that Member Hage would need to agree before the recommendations document can be licensed since Member Hage has contributed some text to the document and he is not present today.

He also mentioned that the next Committee report to the Commission is due in January. Vice Chair Bafundo volunteered to draft it.

7. Committee Goals and Assumptions

Vice Chair Bafundo asked if the Committee has had any discussion of open source and what needs to be open source, and whether the Committee adopted a policy on this. For example, can some components like the tally system not be? The main points brought up were: components should be prioritized for which are “critical path” components or part of the core of the system (e.g. the tally system), and those should be recommended as open source. Other components may or may not demand this (e.g. if an existing, private-vendor component is cheaper and can be interweaved with the other open source components). It was noted that sometimes using a private vendor’s product may limit what open source components can be developed. He said that the Committee should seek for as much of the system as possible to be open source, but if compromises have to be made, do it on the back end.

8. Project Background and Terminology

Chair Jerdonek noted the copy of the application from LA County to the Secretary of State for certification of their Tally System and found it very lacking in information. Member Kattouw said such applications are many times used as “place holders” where they come back and fill in the dearth of details. This might be an attempt to get a partial system certified and not a complete system.

Chair Jerdonek mentioned an article sent to him by Dr. Juan Gilbert about the Prime III system in New Hampshire. He also said he learned that it has been modified from its original form without those changes being public, so that it is technically no longer open source.

Member Kattouw mentioned seeing a task opened on the Colorado project that said “Discussion: precise choice for open source license.” It names AGPLv3 with classpath exception.

Chair Jerdonek commented that STAR-Vote (Travis County) did not take an agile approach and tried to do everything all at once.

Member Hage sent an email to Chair Jerdonek compiling research he had done on other open source voting system projects. There should be a section in the recommendations document that summarizes these projects. Member Hage’s document can be discussed at the next meeting.

Chair Jerdonek asked to put a link in the recommendations document to the LA submission. Member Kattouw so moved, seconded by Chair Jerdonek. Upon voice vote, the motion carried unanimously.

9. Committee Recommendations

At a previous meeting the subject of interpreted programming languages was brought up. Chair Jerdonek said he researched this and found out that the VVSG (voluntary voting systems guidelines) has gone through three iterations, and that the latest version adopted in March 2015 removed the prohibition on interpreted languages.

Member Kattouw brought up a question from Chair Jerdonek around his use of the word reproducible (vis-à-vis building open source systems), and he clarified that to mean something that could be taken and independently built by other government bodies.

Chair Jerdonek mentioned some other items to be discussed: 1) concern for patent infringement, 2) should the City own the copyright or dual copyright between vendor and City. There was a short discussion about types of copyrights and the benefits and implications of each (e.g. single, dual, or multiple). 3) He had another test recommendation where test cases are separate from the code. There was a short discussion regarding the use of digital ballot images for independent testing of results.

The term “digital ballot picture” was previously chosen over using the term “digital ballot image.”

10. Topics for future discussion

Adjourned at 9:34 p.m.