Today’s problem concerns the automatic improvement of each actor’s pValue for Katsinagon, Shialgon, and Tanagagon. These numbers are central to success in the game. Each actor starts off with a rough knowledge of every other gon count, with a uValue that is random between 0 and 1. At the beginning of each turn, that knowledge is improved by a random amount. The problem I note is that the value converges on the correct value too quickly; my algorithm is too generous. So I modified the code as follows:
Original code
float r = random.nextFloat();
newUncertainty[i][j]=r;
New code:
float r = random.nextFloat();
r = 1.0f-((1.0f-r)/2.0f); // this scrunches the random number into the range 0.5 to 1.0
newUncertainty[i][j]=r;
This slows the convergence, but I fear that the convergence will still be too rapid. Of course, these values change every night with dream combat, so that might insert enough uncertainty into the system to make the game work.