Lovely post Michael!
A certain immediacy and intuition seems to be prerequisite in games made today. Why would the average customer choose high investment entertainment over a lower one? I think this has done a lot of damage to the strategy games I used to enjoy.
The development of intuitive games falls easily upon realism. If a game looks and acts like the world we know, we discard the investment of learning its rules. This raises the minimum development cost across the board, striving for realism in physics and graphics. This also pushes out the interesting worlds, a grimy war zone more familiar than even a historical setting.
Emergent programming presents a nice balance. The designer can set up a system capable of more scenarios than they've explicitly considered. Though we do find difficult (and more mathematical) problems very quickly!
You mention Dwarf Fortress, which is quite happy to let systems interact unchecked. Both the design and the agents in the game face problems from the lack of predictability.
How does one program intelligent characters for unforeseen situations? It's quite easy to establish an oscillation in discrete time systems like these games. How can a player sympathize with a character surprised by longstanding routine events? How can the computational cost of forecasting the future reflect the randomness or regularity a character has actually observed?
Further, how does one control gameplay once off piste? In a scripted game the pace of events and challenges can be mapped out at the designers leisure. Allowing emergent challenges could result in unwelcome fluctuations, long periods of boredom sparsely populated with moments of literal impossibility.
Please keep us up to date as you build the mind of the fisherman!
David