The imperative Try Again, may be codified, formalized and packaged in a mathematical language, as a science and as precise protocols to assist us in practicing TryAgain efficiently, and to help us create artificial intelligence entities that would use TryAgain to solve the problems before them.
From a mathematical point of view the question is how to best use all the relevant information and all the important lessons learned so far (at the conclusion of the last failed try), in order to practice a best next try.
This challenge is carried out under time constraints: is it better to try again quickly with a good plan devised after a moderate measure of analysis and planning, or to try again later with a likely better plan devised from a longer effort of analysis and planning?
The answer for humans is emotionally charged; the answer for artificial intelligence is based on resource allocation.
Below we present the following topics:
Randomness is a means to effect changes without wasting effort on planning them, exploiting the power of chance to solve problems and open new vistas. After all, human kind is a case in point for 'randomness in action" as Charles Darwin taught us.
Following a failed 'try', it is necessary to allocate resources to an optimal plan between (i) trying again the same as the last try, and (ii) evaluating what to try again and keeping this evaluation until a 'perfect' TryAgain plan is identified. The challenge is to find the optimum between these two extremes.
Presenting the infinite roadmap for rational practice of the TryAgain imperative. It is based on three conceptual "branching": (i) breakdown, (ii) abstraction and (iii) extension.