The Inquirer-Home

Diebold voting machines benchmarked

The INQ beats the blogs
Tue Nov 02 2004, 08:55
TYPICALLY, benchmarks are used by us to determine the merits of a piece of computer hardware. Benchmarks tell us the Intel Extreme Graphics and AMD's Athlon XP lineup are obsolete at best, and deceptive marketing at worst.

What I am going to try to do is benchmark Diebold's voting machines.

Assuming that Diebold voting machines have a random dispersion, county by county in the clearly undecided states, the vote tallies from Diebold counties should closely match the state's voting percentages as a whole.

Just as The INQ called foul on Extreme Graphics and Athlon XP's later "optimistic" model numbering scheme, the goal here is to call foul on another computer, if it happens to have some sort of odd floating point error.

Test Setup
One Presidential election
The candidate States, in no particular order, abbreviated for, um, brevity
WA, OR, NV, AZ, CO, NM, MN, IA, MO,WI, MI, OH, PA, NJ, NH, FL, HI

Now, some of these states, such as Washington and Oregon, do not use Diebold touch screen voting in any county, but are swing states and help establish our baseline.

alt='usisland'
The Island of the United States

In other states, such as Michigan, each polling place can have a mix of machines, where touchscreens are only made available to provide for disabled voters. Disturbingly, Michigan has have little information available about their voting machines in general, which may preclude their, um, inclusion in our benchmarking suite.

The excluded states, such as Georgia, are poor choices for the benchmark for several reasons. Georgia is in fact a perfect case example. Georgia is both strongly Republican and exclusively Diebold, which would dramatically skew our results. We are looking for deviation from a statistical average and the bell curve differences, if there are any, between Diebold voting machines and other voting systems.

Results
Well, that's for the US citizens to decide today and it will take me a bit of time to sift the data. If the Diebold counties deviate significantly from the average, then we can make the assumption the Diebold has a floating point error of a bit greater magnitude than FDIV, or that the distribution of Diebold's voting machines is not random. Either would be a bit disturbing.

Happy Voting! [PS, I am an American citizen old enough to vote, so leave me alone] µ

 

Share this:

blog comments powered by Disqus
Advertisement
Subscribe to INQ newsletters

Sign up for INQbot – a weekly roundup of the best from the INQ

Advertisement
INQ Poll

Heartbleed bug discovered in OpenSSL

Have you reacted to Heartbleed?