This is a continuation of my reports from the simulations I run in order to test the functionality of the AI. The first post can be found here: Simulation 1-8
Lets talk a bit about patrolling your territory. The AI uses all un-claimed armies to patrol its territory to fend of enemy scouts and smaller enemy armies. For the best effect these patrolling armies should spread out (not move together as one super patrolling army ;)) and they should stay relatively close to your own territory. In one simulation below I actually had one AI capture the capital of another AI and some turns later a third AI captured the same settlement, now a province capital. This might sound great when a patrolling army manages to capture an important enemy settlement but in 99% of all cases this is not an optimal behavior as you leave your own borders unprotected. Patrolling armies should not run away and wreak havooc among the enemies…
I’ve done some good advances on the behavior of patrolling armies and when I’m completed with this (only need to make them stay closer to home) this will be the first version of the AI – Anomander Rake.
But now to the simulations:
Simulation #9-11
In these simulations a lot of behind the scene changes has been made. Especially simulation 11 is interesting as I’ve optimised the editor (processing goes a lot quicker) and I’ve also added the influence map and it now calculates influence for all AIs (prcessing goes a bit slower again).
Simulation 9 – 3 min, 49 sec. 124 combats. 8 capitals, 12 cities, 9 large towns, 5 small towns and 40 watchtowers.
Simulation 10 – 4 min, 57 sec. 141 combats. 8 capitals, 9 cities, 7 large towns, 9 small towns and 32 watchtowers.
Simulation 11 – 4 min, 4 sec. 113 combats. 8 capitals, 11 cities, 8 large towns, 5 small towns and 39 watchtowers.
Simulation #12-15
Here I’m starting to make changes to patrolling and the first change is to make armies actively use the influence map allowing them to ‘feel’ the presence of enemy influence even if the enemy is actually outside of their sight. I’ll only put down the details of one of these simulations as it was a bit odd… In simulation 13 we got more combats than usually and for the first time an AI lost its capital. First to one AI and then a third AI came and captured the same settlement. The simulation ended with those two aggressive empires having a far better ranking then all the other AIs.
Simulation 13 – 3 min, 52 sec. 169 combats. 7 capitals, 14 cities, 8 large towns, 3 small towns and 43 watchtowers.
Due to the heavy fighting in this simulation the total population in the world actually dropped with 20 000 or so.
Simulation #16-17
In these simulations I’ve implemented that the AI must spread where his patrolling armies moves. Previously we’ve often had the case where most patrol armies move together in a big stack hunting the same target. Now all patrol armies cover their own region and this made the number of combats rise dramatically. Similar to simulation 13 the total population in these simulations is lower then we had when we started running simulations (a drop from 500 000 to 480 000). Combat takes it toll on the population.
Simulation 16 – 3 min, 33 sec. 240 combats. 8 capitals, 13 cities, 5 large towns, 5 small towns and 32 watchtowers.
Simulation 17 – 3 min, 52 sec. 222 combats. 8 capitals, 10 cities, 8 large towns, 1 small towns and 33 watchtowers.
Next, as I mentioned above, I’ll make patrol armies stay closer to their home territory and this should again lower the number of combats but also make sure that less enemy scouts penetrate their borders.