I am shifting to an event-based evaluation of statements. In other words, when A tells B and C likes B x, I don’t evaluate the truthfulness of the statement based on B’s current value of perceived Affinity of C for B. Perceived Affinity is, after all, the digested results of all statements made previously on the matter. Instead, I’m going to look at those statements themselves. B instead scans the entire history of what he knows about C’s feelings for him (B). B then assembles those statements into an array and evaluates the array as a whole.
Let’s try an example: B has two records in his history and hears a new one. Here are the three statements taken together:
A tells B that C likes B x.
E tells B that C likes B y.
F tells B that C likes B z.
B uses his current affinity values for each of the reporters to assign weights to x, y, and z; B then assigns the weighted average of those values to be his perceived Affinity of C for B. Next, B must re-evaluate his affinities for A, E, and F based on this new value of perceived Affinity, as compared with what each one of them actually said. B calculates the agreeableness and suspiciousness of each statement, then bSums the net acceptability (equal to agreeableness - suspiciousness) with the old value of affinity of B for the reporting person to get the new affinity value.
Yes, I think that will work.