I wrote yesterday about accountable algorithms. When I say that a public algorithm is “accountable” I mean that the output produced by a particular execution of the algorithm can be verified as correct after the fact by a skeptical member of the public. Today I want to work through an example.
Accountable Algorithms
September 12, 2012 by
Ethan Zuckerman had an interesting reaction to his first experience with the TSA Pre-Check program, which lets frequent flyers go through a much shorter and less elaborate procedure at airport security checkpoints. Ethan’s concerns about unfairness are worth pondering, but I want to focus here on his call for more openness about the algorithm that […]
Opening Government: On the Limits of FOIA and the Metaphor of Transparency
April 26, 2012 by
At a recent symposium (“Piracy and the Politics of Policing: Legislating and Enforcing Copyright Law”) sponsored by the Cardozo Arts and Entertainment Law Journal, I was invited to respond to an excellent paper by David Levine on secrecy, national security, and the denial of public access to documents from the Anti-Counterfeiting Trade Agreement (ACTA) negotiation […]
Broken Ballots
April 23, 2012 by
A important new book has just been published on the technology and policy of elections. Broken Ballots: Will Your Vote Count by Douglas W. Jones and Barbara Simons, covers voting systems from the 19th century to the present, with particular focus on the last two decades. The authors describe the strengths and weaknesses of the […]
Who won the Iowa primary – and does it matter from a technical perspective?
January 19, 2012 by
As Americans know, the 2012 presidential season began “officially” with the Iowa caucuses on January 3. I say “officially”, because caucuses are a strange beast that are a creation of political parties, and not government.
Regardless, the Republican results were interesting – out of about 125,000 votes cast, Mitt Romney led by eight votes over Rick Santorum, with other contenders far behind. The “official” results released today show Santorum ahead by 34 votes.
However, it’s not so simple as that.
First, there’s the matter of paper ballots. The good news is that Iowa caucuses, unlike primaries and general elections in some states, are recorded on paper. So in a case like this, there’s paper to turn to, unlike all-electronic systems where the results rely on correct software.
Second, there’s the matter of proper chain of custody. In releasing the updated results, it appears that some of the records from the caucuses cannot be located. It doesn’t matter whether the records are paper or electronic – if the chain of custody is weak (or non-existent!), then the results are at best suspect.
Third, and perhaps most importantly, “the early part of the Presidential Primary series is the only case in American politics that I know of where the preliminary election results may be actually more important than the final certified results.” [Thanks to David Jefferson for this observation.] While this is not a technical issue, it points out that our technical solutions for voting systems must recognize the reality that timely accurate results are important – timely results that are wrong aren’t helpful, and slow results that are right will be ignored.
Finally, it’s critical to realize that caucuses, like primaries in some states, are run by the parties, and not by the election professionals. Perhaps if the caucuses were run by the pros, some of these problems might not have happened.
