State Certification of Future Voting Systems — 3 Points of Departure

This is a more technical post than others here given the broadening of an audience visiting this Foundation web site in search for content like this article below rather than hanging out on our more geeky Project site (which is soon to be relaunched and be way more engaging for all audiences, we're excited to report).  Usually, you will find this kind of content over there, while here we'll talk more about voting experience innovations, policy matters, and progress of the Project.  So, for those who are passionate about elections reform and improving the voting experience, but are not as fluent in some of the technical issues, feel free to look this over, but do not fret if seems like gobbledygook.  There is more relevant stuff for your concerns to come.  Ready?  Here we go...

In advance of this week’s EVN Conference that just concluded this evening (and a great conference it was--more on that soon), we spoke frequently with colleagues at several election oriented groups (Brennan Center, National Conference of State Legislatures, EAC, NIST, and others) about the way forward from the current voting system certification regime.  One of the topics for the EVN conference is a shared goal for many elections professionals: how to move toward a near future certification regime that can much better serve State election officials in States that want to have more control, customization, and tailoring of the certification process, to better serve the needs of their local election officials. 

NOTE: For those unaware, voting systems--that technology which casts and counts your ballots--must be "certified" as meeting or exceeding a set of tests for integrity.  As we move toward new innovative solutions, based on open source principles and a more modular architecture, the antiquated (if not soon to be obsolete) testing and certification process will have to be brought to date to support such new component-based systems.  Part of our non-profit work is to help governments wherever we can to re-invent certification.  That is largely in the hands of the U.S. Elections Assistance Commission and NIST, but increasingly requires addressing States' needs at the State level in concert with Federal guidelines development.  Now, here's John...

As readers of our more Project-oriented technical blog  will have noticed, I am already in the process of laying out several recommendations for certification reform, in response to the open discussion on that issue, broadly, at the recent NIST/EAC Symposium in early February. However, for the road towards new State certification programs, we can relatively briefly make 3 main points. If progress on all 3 is rapid -- and we believe that it is possible -- then we could see a lot more flexibility for States, without any dilution of the fundamental controls for which certification is intended. The 3 points are:

1. process simplification,
2. voting system components, and
3. chunking of requirements (for want of a better term).

Simplification

The greatest single way to simplify certification is to reduce the scope of testing, by focusing on those requirements that are essential to evidence-based voting systems (here is a somewhat geeky article by 2 U.C. Berkeley professors on evidenced-based elections) and dropping those requirements that don't fit them. That doesn't work for every voting system, but it does fit the desires of most if not all States that are currently looking at more "localization" of certification, and new voting systems that are evidence-based. But what's that all about?  Details have already been discussed in this previous post on the older Project web site.

Components

So, if we first focus on evidence-based voting systems, then limit the requirements appropriately, then we can look at individual components of a voting system rather than take the entire system as a single thing.  A State may not want to certify a election management system product (EMS), because in its definition of evidence-based systems, and EMS doesn't create any evidence of the election, and doesn't cast or count ballots. But such a State may want to require a vendor to work with a test lab on testing a specific component like a ballot marking device or a ballot counting device.

With the smaller scope of testing enabled by simplification, a focus on individual components can take an even bigger step to creating smaller, more manageable tests and certification, more amenable to State adjustments, than the current monolithic multi-year scheme in place today.

But what would component certification amount to in practice? What are "components," how would they work together; and what are the requirements and standards?  Many questions, but also at least the beginning of some answers in recent posts like this one.

Chunking

Ok, fair warning: John offers some conceptual curves ahead; drive carefully.  :-)  For lack of a better word, "chunking" of voting system requirements is the term that we're using to describe how States could leverage future voting system requirements and testing guidelines that have been simplified and focused on voting system components in the manner we refer to above. Requirements could be composed of related but separable chunks, where some chunks can be used independent of others, in the context of one State's process of certifying a single component, based on testing of it, using the State's selection of chunks.

For example, a State may wish to certify a ballot marking device (BMD). They may require a test lab to test the BMD using chunks of requirements from a larger set of (future, perhaps federal) voting systems requirements. They might direct the BMD's maker, and a test lab, to focus testing on usability, accessibility, and reliability, but not other sets of requirements that might apply to (for instance) ballot counting devices but not apply to BMDs.

Are you still with us?  Awesome, then you've made it.  So let's have John summarize.

Summary

The current (Federal) certification model needs to be reinvented to support a modern more innovative architecture for voting systems--one based on components.  And such will be essential with a process for States' adopting their own variation of this certification model.  Reinvention means considering three aspects: (process) simplification, (voting system) components, and the more cerebral concept of "chunking" (of requirements). The first 2 of the 3 steps are in progress or within reach. My personal belief is these efforts might go faster if we also include the goal of having future voting system requirements that, thanks to simplification and component-based partitioning, would be  more amenable to States being able to pick and choose those chunks that are the most helpful to their objectives.  So, that's my broad vision as CTO.