Well the original idea was to have a library that used the law of finite differences to find patterns and decode information, and eventually it turned into and equation that did the same thing but with least-squares regression. I got an applet from Bryan Lewis, a mathematician at Kent State and I made a flow-chart out of everything that it did, and if I can figure out this loop/decision thing, I'll be able to express it as an equation, that uses a series of series to hold the data, and a variable that is incremented by one each time, to calculate all the possible elements of the 4th degree equation. If it's lowerthe unused elements work out to be 1 or 0, and thus hav no effect on the equation. For example the equation of a straight line would calculated (eventually. There are other equations which work out the accuracy of it) to be Y=(equation for the slope)x+(equation for the y-intercept). Ideally it can take any numerical values and eventually find a pattern that explains it - very useful.
