Arizona Capital Times Editorial on Election Data - Right?
I can't recall a newspaper editorial on election stuff that I've agreed more with than the Arizona Capital Times "Improved election data would mean a better informed electorate" ...though with a title like you can imagine I'd be a fan. But before I pick at just one part they got wrong, let me pick 3 of the 24 sentences worth quoteing:
Sean Greene, the election initiatives research manager at Pew Charitable Trusts, argues that structuring election data in a uniform manner should be just as important as other government data, such as statistics in public health. ... The same attitude should be applied to election data. Ultimately, the ways that counties handle their election data affects the ability of journalists, researchers, lawmakers — or anyone else for that matter — to efficiently monitor our state’s election process.
Yup!
Now let me paraphrase the part that I think is wrong, where a there is a much brighter future.
Arizona Secretary of State Ken Bennett seems to be in favor of changes to (as we say here) "get the data out" for transparency and open-gov goodness (agreed) but pointed out that this is a county matter (true). Yet, based on Bennet's remarks, the authors expressed concerns over costs of the changes: "Counties are entitled to a level of autonomy and costs associated with changing any system can become prohibitive, especially if improvements are only being encouraged." (Our emphasis in underlining added.)
Here is the very good news: those costs are small, at least measured in dollars or in hours of staff time. What is not required is changes to any existing software or election management system -- which would be costly. What is required is some really boring basic scripting and querying to make a funky little tool that crunches over the existing database of election records (every county has one) and spits out the data in a public data format. That's just conversion. What is really not required is for any election office to create new systems for mining that data. Rather, just get the data out in some format that others can grab and mine.
But don't take it from me. I checked with a couple very senior election technologists from two ends of the country that I rreached out to of after reading the article. This sort of thing is really not a lot of work, and indeed election IT teams get called on to make various sorts of data extracts frequently -- and sometimes have the time to make them.
The constraint is not the dollar cost. It is the opportunity costs, which are relative costs. For a large county, this is a non-issue. For a small county with maybe a couple technical staff, the capability is there, but just not the time; they barely have enough time for their many existing responsibilities. Only 10% of one person's work week is actually 5% of an entire week - and probably a higher proportion of a week's output.
This is where standards come into play.
If there were an agreed-to nationwide standard for what this data should look like, some of the larger counties would use it, and get a couple people to spend the hours to build such an export tool that works right. And if the scripts and queries of that tool were published to share with other elections organizations -- especially those that use the same voting system product, as many of the counties in Arizona do -- others could benefit from the prior work with lots less effort.
But a standard is needed.
Even Better News: NIST (National Institute of Standards and Technology) is working on one for exactly that! And people are waiting for it to be finalized. Our TrustTheVote Project team is waiting. I'm sure that some of the vendors have teams that are waiting. A couple large election office's technical teams are waiting. And I am sure that public interest organizations -- that share the sentiments of the editorial's authors -- have been waiting for a long time for the election technology people to get the data out!
As I said, we are waiting too. When there is a standard, and the early adopters are getting the data out, then we will be collaborating with them on what to do with the data once it is out. For starters, it will be integration with the existing TTV Analytics (big picture here), but even more interesting -- TTV Election Night Reporting System, which takes that data and visualizes it for members of the public, like this:
We got work to do! But getting a bit sore from crouching in the starting blocks :-) More soon!
-- EJS