PAMB is essentially a programmed opponent, mixed with some rule changes to TMWWBK to compensate for the fact that the programmed opponent will never be as smart as the player. In general I do not like modifying rules, even for the programmed side. As a solo gamer I acknowledge that gameplay will never be as challenging as a live opponent, so I solo game to try out tactics or strategies, refine lists or decks, or to create an interesting or amusing narrative.
So, how is PAMB programmed? Basically it is a simple rule-based system with priorities identified for each unit type and general actions to take if the unit-based ones do not apply. The first system is for programming the natives, but includes a second system for programming the Imperial forces. Here is a small sampling, to give you an idea. (But not all. You should buy the book.)
- I like my shooting units to find cover and stay there. Failing that, I like them to advance to short range and fire, avoiding melee.
- I like my melee units to advance quickly into contact, towards the closest enemy they can see, except as noted in rule 3.
As stated previously, PAMB also has rules that modify the TMWWBK rules, such as randomly appearing native units (similar to the Two Hour Wargames' PEF concept), native action modifiers (which include the unit leaving the table), and recycling native units.
My dream has always been for solo gamers to be able to 'trade' programmed opponents. My schemes have always been too complex - essentially requiring other gamers to have computer programming skills to define the rules - and I can see that PAMB's method is a good start towards codifying such rules for programmed opponents. The one change I would make is that rules have to be in order of precedence and that when you hit a rule that applies, you stop evaluating the rules.
I still think that generic opponents would be harder to create than ones for a specific scenario for a specific time period. Maybe that will be my next project.