Texas in the works

Lately, I’ve been getting to bed very late. Yesterday, I woke up at one in the afternoon after too many late nights in the weeks before. This is partly because I have been looking at the Texas voter rolls lately, and found something interesting.

“Interesting” means not going to bed until it’s solved or I fall asleep at my desk, whichever comes first. I never had these problems in my work as an artist or art director. I could always go to bed without worrying about some problem from the day that needed solving. Searching for algorithms, and untangling them once found, is totally different.

Late last year, I was asked to analyze the voter rolls of a couple counties in Texas. At the time, I was working on the Ohio and Pennsylvania rolls. I’d looked at Texas the year before and hadn’t found anything. However, in the first pass, the data had been very noisy thanks to the way voter history is recorded there (one new record for every election). Since then, I’d cleaned it up, making it much easier to see what was going on.

Initial efforts yielded the discovery of an algorithm that shared some traits with an algorithm I found in Ohio’s voter rolls. I wrote it up and moved on to the next state. The goal at the time was to see if there was an algorithm or cloned records in each state, then move on to the next. There was no time to solve the algorithms before the election, so I focused on trying to find signs of suspicious activity in as many states as possible.

So far, I’ve looked at NY, NJ, OH, PA, TX, OK, WI, GA, AZ, CA, and a mystery state, to be revealed later. Of those, all but two have unnecessarily complex voter ID algorithms.

A few weeks ago, I was asked to consider submitting the TX study for peer review. To do that, I had to modify the report I’d made. Unfortunately, I didn’t remember how I’d arrived at one of the findings. This meant opening the data and looking again. That effort yielded more information about the Texas algorithm.

I’ll save the details for the journal article, but can say this much: Texas’ largest county uses a very sophisticated algorithm to assign one of the two voter ID numbers they use. This algorithm clearly segregates records, has the ability to carry information not contained in any visible field, and is well-hidden under multiple layers of obfuscation.

It has one thing in common with NY: the Texas algorithm loves repeated numbers. However, unlike NY, which uses repunits (1, 11, 111, 1,111, 11,111), TX uses 9 base repDigits (9, 99, 999, 9,999, 99,999).

Last night, I was looking at columns of repdigits. Clusters of 9’s all over the page, along with a few multiples, like 198 (2*99), 297 (3*99), and and 19,998 (2*9999). Then I realized it was 2:30 AM. This was a few hours earlier than I went to bed the night before, but I was tired. By then, I was typing with one hand, had both feet on my desk, and was slouched so low in my office chair I could barely see my screen. My heart beat a little faster than normal, telling me I’d better get to bed.

It was difficult to leave the algorithm behind, but it had to be done. We’ll see how far I get before sending it in for publication, but enough is known to raise some eyebrows regarding the presence of such a complicated algorithm in the Texas voter rolls.

All this is my way of sharing that I have a lot of MyPillow products in my house. We have 3 pairs of their MySlippers, which my wife and I wear all over the place, and loads of their pillows, which really are the best I’ve ever slept on. I’ve been using them for about 4 years. I’m so used to them, that I forget sometimes how good they are. They are what pillows are supposed to be.

However, last night, one of my old pillows suddenly appeared on my bed, in the middle of the rest. It was flat and miserable. I tossed it aside, moved the MyPillows where they belonged (under my head), and went to sleep.

Five hours later, I woke to the sound of either a herd of elephants on my stairs or my daughter getting ready for work. Now I’m back to work on the algorithm.