Investigation of AI reassessment upon reload

[Bromley]

EDIT: Apologies, but I have missed out Rhodes (Greeks) in all of my tests. It's immaterial to the conclusions though.

As far as I know, only DimeBagHo has tried to quantify the effect on the AI of different intervals of save/load. I've just finished a series of tests to provide some similar evidence. Unlike Dime, I just ran one test for each interval, so I don't have averages. That weakness is offset by running the test for all intervals between every 2 turns to every 10 turns (inclusive).

Just to be clear, when I say an interval of, say, 4 turns I mean that I click End Turn 4 times and then save/load.

As intervals of 4-6 turns seemed to be key, I did perform a second run for each of those, so I have basic averages there. I also ran a no-load game. The first column in the results, labeled "Start" is the starting position in 270 BC. I started all the runs from this same saved position.

So:
RTW 1.2, no mods, M/M, Brutii, 20 turns, various intervals, everything moved to cities at start, no other player action, answer "no" to any diplo/marriage.

I recorded the number of territories owned by all of the factions (except rebels). The total of these show the total AI expansion, which is the key result for this test. The data on the individual factions is still there though, so others can draw conclusions from those if they'd like.

I also specifically recorded the state of Sicily.

Conclusion

Observations of Sicily show that the AI has real trouble expanding if loaded every 4 turns. The cusp seems to be 5 turns, where the Scipii in one run took one city yet in the other remained static. With 6 or more turns, the Scipii managed to overrun Sicily every time (although it took them longer at 6 turns than at 10).

Total AI expansion seems to really suffer until you hit intervals of 4 turns. It seems to require an interval of 7 turns to maximise (or get close enough for there to be little difference). Obviously, more runs would generate averages that would better pinpoint this, but I'm happy that the general trend is pretty obvious with what we've already got.

Note how that last paragraph disagrees with the first. i.e. in the first I'm saying that 6 turns will allow the Scipii to conquer Sicily happily and yet in the second I'm saying that intervals of 4 and 7 are the key ones. I believe this has to do with things like the distance between the cities and the level of opposition in the area. So extending the interval beyond 6 makes no difference to the Scipii in Sicily because they only need 6 clear turns to reach and overrun a city. It does help other AI factions reach the cities that are further away though. Likewise, the Scipii are completely boned with an interval of 4, yet other factions manage to do okay. That's because the Scipii are facing walls and armies rather than rebel settlements.




If I haven't explained something properly, pm me (to keep this thread as uncluttered as possible) and I'll fix it. Likewise, if people want any of my saves pm me and I'll upload them somewhere.