I don't see the value of random levels of aggression. I mean if you randomly got a few normal AI next to a few Aggressive AI... you'd just end up with a lot of dead normal AI and a few Aggressive AI with nice big empires.
Perhaps you would in some cases (I would argue in others you might not, because if those non-aggressive AIs don't' get attacked right away -- the aggressive AIs are busy with other prey -- maybe the tech lead they build up means THEY eventually become the nice big empires (with really advanced units instead of a lot of them). But the point is to increase variability and unpredictability of the experience, isn't it? Doesn't this do that? And if you think the point is more simply for it to be harder, and you believe what you wrote would always be true ... wouldn't "a few Aggressive AI with nice big empires" be harder to beat than a lot of Aggressive AI with small empires??? Wouldn't this possibility partially defeat the canned strategy of beating the Aggressive AI, just survive the early eras and come & get them later when they are hopelessly outteched? If they can get to be big enough, maybe they can keep up in the tech race just through sheer size of empire (vs. quality of infrastructure).
Aggressive AI doesn't means every one is a bloodthirsty killing machine. It means that warlike civs will kick ass, and peaceful civs will have larger armies to defend themselves from said warlike civs. Random Personalities changes this around.
With Random personalities and Aggressive AI, the AI aggression is still random because each leader has the personality of someone else. The aggression of any given AI is determined by their personality. Therefore, with random aggression, Gandhi may be a bloodythirsty killing machine one game, and docile the next.
"The aggression of any given AI is determined by their personality." I'm willing to bet that this statement by you is simply not true. Again I ask you (and I see others have asked), how do you know these things?
I don't know all the details of how the CIv AIs are specifically programmed, but as I wrote earlier, I
have programmed AIs before (both decision tree pruning and neural net). I assume that the
different AI personalities in civ are a result of different weightings applied to the same AI ruleset (determining probability of a decision path being chosen). In other words, maybe gandhi has a higher weighting towards peaceful pursuits in the probabilities. But the options (the choices the AI can make) are still the same, it's just that one personality is more likely (has a higher probability, as a result of a weighting multiplier) to pursue one path than another one may. Everything I have ever read in these forums supports that conclusion, in fact dj_anon is publishing these weighting factors in his excellent reference sheets.
A
different AI (aggressive or normal) is not just an other set of multipliers applied on top of the same AI! To call this a different AI instead of just anew set of personalities is a ridiculous claim. Reading what Blake has written, I think he has done exactly what he claims he's done, he has written a different AI implementation. I have no reason to believe he's lying. If you do, I still say: much of what he claims the aggressive AI does can't just be explained by different personality weighting multipliers!
This is why I say your apparent claim equating Aggressive AI gandhi to standard AI Alex is silly. That's like saying a woman bodybuilder has the same physical strengths and weaknesses (upper body, lower body, etc) as a more typical man. Uh, no ...
---
Anyway, other than winddbourne, who I think also recognizes that the key to improving the game experience is not necessarily increasing absolute militarism, just increasing variability and unpredictability, it doesn't look like my idea of random AI algorithm usage setting (random type of AI used) is gathering much interest
