I'd like to guide you to the truth, so you can stop being such a hack.
Look here people, this is a great example of the power of visualization.
Votes, in math, are called a TIME SERIES. They're exactly like a stock ticker, votes are "events" that occur at certain "times".
If you have two voting machines right next to each other, you have two time series. Which can be combined into one, the same way individual stock trades are aggregated.
Crash courses on time series analysis are everywhere. This one though, caught my eye. Because of the visualization method. Check this out:
In this example they use a disease epidemic as a time series, but look at they way they expose this correlation, this is beautiful:
This is THREE YEARS OUT, they're looking for any relationship between the outbreaks of Dengue types 1 and 2 fevers.
I mean, CRYSTAL clear, right? Crystal. Undeniable. Self evident and obvious.
This is what I'm talking about. This is power spectral analysis on a time series, and clever visualization of the results.
If you peruse the link, they show some great examples of "periodic" behavior in the underlying generators. Like, if you have a voting booth and it takes 3 min to get through the procedure and cast your votes, then WHEN times get busy and there are lots of voters in line you can expect to see a periodicity of about 3 min in your data stream.
There are TWO issues here, the occurrence of a vote, and its authentication. In the scenario just mentioned which is occurrence, if you're a computer guy you'll immediately realize we want to tag each vote with the unique serial number of the machine it came from, because if we fail to do that, we have to use statistical methods (factor analysis) after the fact, to separate the data streams and recover the original individual time series.
And the SAME concept applies to signature verification machines, AND to signature verifying humans.
That Dengue graph is beautiful, ain't it? You can see the fractal structure just by looking at it. And once you know it exists, you can predict its criticality - this latter thing being POSSIBLY causal to anomalies in the aggregate data stream.
