Hanlon’s Razor, which I first encountered in the Murphy’s Laws Book 2 listed at the link, goes something like this: “Never attribute to malice that which can adequately explained by stupidity.”   Other words can go in place of “stupidity.”  “Incompetence” and “ignorance” also work.  The gist of it is that human failure is more likely than human malice, so we’d rather assume that someone made a mistake, instead of deliberately slipping us the knife.

Political opponents often prefer to jump on malice, becuase it makes their enemies look worse, so it’s sometimes hard to separate malice and stupidity.  Secrecy can make it hard to do, and the recent resurgence of the voting machine debacle that was Diebold (and is now Premier Election Solutions) points to one of those cases.  The problem, according to Premier, occurred when a district with many memory cards tried to update a central repository.  In this case, votes can be lost.  This flaw had been in the system for a decade.

Of course, there was no way to actually know about the error.  There was no paper audit, nothing to compare against. Possibly the original memory cards, but I know how government works — if these machines were certified by the state, then the counties assumed they worked, and had been tested.  Perhaps they were, but it shouldn’t escape notice, that until recently, Premier had a factory just outside of Columbus, Ohio.  And now, Ohio is suing them over the losses.

Now, this sounds very much like an incompetence issue.  We’ve seen similar problems with Windows, odd things that never seem to be important enough to fix.  We even know that these machines ran windows, because Premier tried to blame Norton Anti-Virus for the problem.  That’s stupid for its own reasons.

Officials in Butler county lost 150 votes off a single card.  That’s a single machine, in a single district.  Premier claims that the effect is most likely to occur in “larger districts” where they have more voting machines.

So, in computing terms, this is mostly a random error.  That is, it doesn’t matter whether these votes were Republican or Democrat or Other, the failure is a technical one.  In general, that would be a good thing.  If you’re going to have an error, have it affect a random population.  That means that if, say, 51% of the vote went Repblican, 51% of the lost votes were also Republican.  This would not — most likely — affect the result of the election.

However, the failure isn’t completely random. It hits “larger districts”.  In Ohio, that means cities.  Ohio’s demographics reflect that of the country, in the sense that the higher populated areas are more strongly Democrat, and the lower populated areas are more strongly Republican. In fact, it’s largely polarized that way, with some notable exceptions, like Cincinnati.

That implies to me that the locations most strongly at risk to uncover this error are ones that are also largely Democratic.  Franklin County, where Columbus is, was evenly split between Bush and Kerry in the 2004 elections, according to CNN, while Cleveland was strongly Kerry, and Cincinnati was weakly Bush.

Also, according to CNN, the final vote counts between Bush and Kerry differ by only 110,000 votes.  Those 110, 000 votes decided the US election, potentially.  How likely is it that failed e-voting machines contributed to this?  Well, the affected voting machines are used in 44 counties — so not all of them.  But in one county, in one precinct — in one machine, we lost 150 votes.

According to Franklin County Board of Elections, there are 1212 voting locations just in the county, 95% of which would be in the Columbus Metropolitan Area.  I don’t know the statistics on what machines were where.  I know the machines that were used that year weren’t used in the primaries this year.  I know we didn’t have a paper trails, and I know there were enough other shenanigans going on that it leaves me with almost no confidence in the 2004 election.  Not that I had much in the first place.

And while accurate results are important, it’s even more important that we believe the results are accurate.  Instead, we get left with doubt about the results, and thus our leadership. Election tampering, even accidentally, undermines the confidence in elections, which can undermine the very core of what a democracy stands for.

This error existed for 10 years.  A little thought brought me to the realization that it would hit predominantly democratic areas.  Not completely, and I don’t have good numbers or statistics to support it, but it feels that way to me.  Surely the computer scientists who discovered the error could have seen it as well.

It begs the question in my mind.  When did Premier Election Systems know about this error.  How long did it take them to find it.  Just because it has been there for ten years doesn’t mean they didn’t just locate the error last week.  I’ve found errors that old, that no one knew about.  Any good debugger is going to be able to look at it forensically (or archaelogically) and guess when the bug was introduced.

But when Premier knew about it, and what they chose to do about it when they found out.  And that is the difference between malice and stupidity.