Establishing Benchmarks

When we build systems we want the first establish a benchmark like we did by following the process described up to this point.

Then when we are trying to improve the system we only change 1 item at a time, otherwise it is impossible to determine the impact from a single changed item is multiple items are changed at the same time.

Also, because of the use of random seeding, we need to run the same process several times to see how much each end-result varies.  If they are all fairly consistent that is good.

** this should be expanded to show the variation of 4 identical tests.