Can We Test Voting Systems?

We’ve quite a bit lately about a topic that is central to election confidence. One way of asking the question (which we heard this week at the Pew Center’s “Make Voting Work” Voting in America summit) is:

How should voting technology be tested so that we can believe that it works right? If there is an “Underwriter’s Lab (UL)” tag on my desk lamp, should there be one on the machines that count my vote?

The answer is “yes, but.” Yes, there is a way that voting systems are supposed to be independently tested, accredited, certified as meeting Federal guidelines, and certified by states as well. And that’s important because for a voting system to be legal use, it must be certified. (In most places, in most cases – as with everything in U.S. voting there exceptions and thorny details.)

But here’s the problem. Not a single voting system product in use today meets the current Federal guidelines, and not a single one has completed the Federal certification process. Only one of the 4 major vendors is still trying – after a year of efforts and no end in sight.

There’s no point in bashing on the current system, because all the players have something to complain about – including the Election Access Commission (the part of the Federal government that is responsible) which (I am guessing) has a pretty clear view of the current system’s failure to product any results in 3 or so years.

At OSDV, we typically don’t like to complain (and yes I am complaining now) without also having some type of alternative in mind. Well, it turns out that there is a fine alternative that is already proven, that NIST has been operating successfully for years, and could be adapted for test, evaluation, and certification of election technology products. Keeners can read more on the OSDV wiki, but it is worth closing with the observation that NIST’s NIAP program (as it is called) includes one thing that is sorely lacking in voting systems: profiles of what an evaluated product is supposed to be.

That’s right, nobody has actually made a standard definition of a voting system product! But it’s underway. EAC has funded an effort to begin modeling how elections actually work, how voting systems function within them, and what the risks are. (OSDV participates in this team.) And OSDV is building reference systems and specification that define what the voting system components actually do. Both are a beginning, but they are work towards knowing what it is that’s being tested in a future NIAP-like program (knock wood) so that we can real meaningful independent testing and certification. “Can we test voting systems?” – today, not very well, but there is a better way.

— EJS