Harold was touched with a bit of nostalgia as he began to train the Machine. It reminded him of his interactions with Will when the boy was just a toddler. Nate loved his son, but was also an extrovert and head of the company and had many demands on his time. Olivia was not inclined to be a stay at home mom, so even though her relationship with Harold was somewhat cool she was not going to turn down free babysitting.

The patience he had learned when dealing with a hyperactive and inquisitive toddler served Harold well when trying to teach human behavior to a burgeoning AI. Of course, Harold never dreamed of the trouble a curious AI could get into...

The Machine was given a baseline of human actions and reactions to certain events. It then observed and gathered as much data as possible. When an incident violent event occurred it reviewed the preceding data to detect outliers and trends. It compared cases which escalated into physical violence with similar cases without violence to determine if there was a triggering act. Pattern analysis was used to create predictive algorithms.

As the machine used the information provided to refine its algorithms it became more proficient in predicting violent behavior and Sys Admin needed to spend less time on its development.

Left more and more to its own devices, the Machine took it upon itself to improve its performance.

While crimes involving forethought and planning were easy to spot, the Machine was less successful in predicting crimes where an emotional component was involved—when would the perpetrator tip over into violence and when would they cool down and reconsider their actions?

To improve reliability in this area, the Machine devised a series of experiments to map human response to irritating stimuli.

Initial tests with email spam gave unreliable results.

The Machine took to calling people at random hours to map their responses, as it had found that people were annoyed to answer a call with no one on the line. It could then gauge their rage response. While early morning and late night calls interrupting sleep were quite provoking, calls during dinner mealtime seemed to cause the greatest anger response.

An unexpected collateral effect of the calls was a number of divorces, as some spouses became suspicious of the number of hang up calls their significant other was receiving.

Nathan Ingram's was one such marriage. This did however give Nathan more free time to work with the Machine on the Irrelavants project.