1.13 Corp Games
Requirements
Modify the Team Games program to allow users to benchmark Strong Social Intelligence (i.e. teams that collaborate on each move). Create a new type of player called “Corp” similar to a Team. When selecting a member to serve as Explorer, also select one or more members to serve as Debaters (contrast to Weak Social Intelligence in which the only Debater is the Explorer itself). Increase the Research Budget max to permit more use of Debaters (research/collaboration overhead is an expected weakness of Corps, but might be addressed by reducing cost per experiment). The options for the specialization colormap on Corps are Explored, Debated, Accuracy, F1, LG, Empath, and Teach. For each event (and for top node), allow users to view a graph of the evolution of share (among all members) in exploring (a.k.a. hypothesizing), debating (a.k.a. hypothesis testing), and total participation.
At the end of each experiment, instead of the Explorer learning strategic/unstrategic outcomes if predicted by the Explorer on its last move (as in introspection), determine the strategic/unstrategic outcome via vote among all members weighted by each member’s Long-Game for the grandparent node in the Corp’s game-classification tree. At the end of each full match, the skill rating is updated for Corp as a whole (never for an individual member of the Corp).
Acceptance Test Plan
Like in the previous Release, benchmark Corps of various compositions against each other and against existing players (including the current grand master of Test). Do any fare well against the grand master in the first 100 games? Is the optimal composition of a Corp the same as a regular Team? Do Corps evolve the same specialties as regular Teams? How do the two types of Teams fare against each other on Train and on Test? Are different types of players optimal at different stages of expertise? Do different machine learning algorithms (or parameters) work better for specialists, generalists, explorers and debaters?
Compare favoritism by Corps vs other kinds of players