Many times previously I have pondered the problem of deciding an Actor’s preferences in deal-making. Now the time has come for me to grapple with and solve the problem. The current version I am facing is more complex than previous versions, because there are three different requests that an Actor can make: an auragon count, a bead, or a promise not to attack. How in the world can an Actor evaluate the relative values of these options? And within the the first option, how can the Actor decide whose auragon count to seek, and which auragon count to seek?
These alternatives must be considered simultaneously, so the list of alternatives is as follows:
third actor red auragon count
third actor blue auragon count
third actor green auragon count
fourth actor red auragon count
fourth actor blue auragon count
fourth actor green auragon count
bead
promise not to attack
The easiest evaluation is that of the bead. A bead has a fixed value that declines with time; late in the game, it is less valuable, but the decline per turn increases with time, because lying on the last turn is of greatest utility. Thus, the particular value of a bead per turn increases, but the total value of a bead at any time diminishes with time. Think in terms of a series like 1, 2, 3, 4, 5, 6 for individual turn values.
The second easiest evaluation is that of the promise not to attack. That depends on the relative strengths of the two actors. In particular, an actor who has zero auragons of a particular type is particularly vulnerable and will want to have extra protection. This is a complex calculation. Ultimately, much hinges on the degree of certainty that the actor has in his pAuragonCount values. Which brings up another issue:
Uncertainty Values
At one point in the development cycle I had extensive provisions for calculating uncertainties. However, I ripped those out at some later time. So I retrieved the old code and compared it with the current code, but I was unable to locate the code that handled uncertainties.
Uncertainty values provide the perfect way to determine the value of information; the more uncertain my current pValue is, the more valuable new information would be.