What I don't understand is why more games/companies don't emulate Stardock and its "the AI learns from successful players" idea. In their games (at least GalCivII), successful players can submit their game record and the AI learns from them the best build queues and other strategies. I have great hopes for the Civ5 AI, but I think all AIs could benefit from this type of data gathering/cloud-ish computing.
Why wait for users to submit game records? I'm sure Steamworks could be used to gather this data without the player's knowledge.
0h n032 H4x0R2 fr0m Fir4xi5 4r3 in my c0mpu73r 57341ing my 57r473gi35!!111!!1!!1
Seriously this sounds like a really good idea, but I'm not sure we ever saw improvements in the GalCivII AI that could be directly attributed to this data.
Even after the latest (final?) patch it was far from challenging.
As for the other items discussed here I like the idea of the Revolutions mod. I will have to look into that further, but the idea of keep a stable civilization (and possibly promoting stability overseas) seems like an interesting late game challenge - a sort of post-scarcity challenge (at least in terms of civ elements like buildings, technology, etc).
I have tried various revolutions mods and they have some interesting mechanics, but I could never get past the fact that keeping my empire stable felt less like fun and more like repetitive busy work. I don't think the ideal way to do this has been found yet. In particular I found the stability issues arose too early and before sufficient means to control them had arisen. Even as a peaceful builder (in some cases especially as a peaceful builder) it was hard to manage and kept forcing me to war.
If we are looking for a late game equivalent to exploration and research I think we should be looking at things like prospecting, economic development and some variation on corporations to boost production, late-game research, and introduce new diplomatic options. I especially like prospecting (as in locating and exploiting important mineral deposits) not just on one's own land but through diplomatic and economic agreements with smaller Civs and City States.
Which actually introduces another late game change I would like to see. There should come a point (around the time a human player quits and starts a new game
) that an AI Civ should accept they are not going to win but instead looks for a larger Civ to support, based on things like proximity, power and diplomatic history. This would have the effect of creating a host of new 'city state'-like dynamics as the larger Civs vie for the affections of those smaller Civs that have switched from an objective of winning to one of surviving and hopefully picking the side that will win.
But my favorite idea that i have come up with, based on some ideas here, would be some "general victory conditions" for the late game. In my view these general victory conditions would be a mix of the mini-objectives you get in Civ4 ("be the first to build 8 knights") and the social policies/cultural victory in Civ5. I would have some late game technologies, or late game events, or UN actions enable these general victory conditions. Such things could be major production, science, military, cultural expenditures like "give 15 tanks and 15 infantry for global UN peacekeeping operations" or "donate 2000 gold to the WHO to fight hunger and disease" or "establish technology pacts with a majority of foreign governments", etc. Each victory condition would be worth a certain number of points toward total victory (weighted for balance and difficulty), with total victory coming at, say, 10 victory points. The normal victory conditions could be included here by simply having each of them be worth the 10 needed points. (If any of you have played Twilight Imperium 3, my idea is similar to their victory points system.)
I think this type of building up to victory would be complementary to the current system - builders would have something more complex to do over the final span of game, cities won't have to sit idle if they aren't specialized toward the major victory condition you are pursuing, no single achievement ends the game so you have to have a balance to your capabilities, and some could be made that would enhance diplomacy in the end game (for example, the troops donated to the UN could actually be deployed to war zones on the map as voted for in the UN) more than just getting a majority of the voting nations in the game to vote for you (I have always been disappointed by the UN and the diplomatic victory conditions in Civ games).
I think there is some real merit to this idea. I have always felt there should be some ability to aggregate achievements throughout history as part of measuring the success of your game. In fact I wonder if it would be possible to award victory points for the most successful Civilization during each era, not as an absolute measure (i.e. largest at end of era wins) but using a relative measure, so the most successful Civ during an era would be determined by which was the most improved.
I particularly like the sentence I have bolded in the quote...the mastery mod for Civ4 and later
BTS, which introduced "
a new victory condition that approaches "winning" from a more comprehensive perspective", was always one of my favourites; That mod combined with some accumulation of victory points over time would be a winner for me.
). I'd prefer a more gradual victory-where you achieve points for being the first to "launch satellites", put a person in Orbit, land on the Moon & set up an extra-terrestrial base/colony. 
