I think I can safely say that the experiment was a success:
38 Hours?! Good grief (and no, I'm not trying to sound like Charlie Brown here). I mean, I knew this was taking a lot of time, but that's almost as long as it took me to win Epic 27 - which was an Always War game! My games are always spent "on the clock", I'm not exactly sure why some people leave Civ3 running for so long when they aren't playing. To each his own though. 
Some interesting stuff emerged from watching the replay. The Zulus got a settler from a goody hut in 3750BC, which probably is part of the reason why they seemed to be doing well in my game. They might well have been much weaker in other games, I'll have to see how that played out. I of course got my own settler from a hut in 3500BC, and used it to found Tarsus. Since I expanded west in this game instead of east, I got the whole western and northern parts of the starting continent and left the east mostly to France. Watching the replay, I expect that in most games France won't get quite so much land to the east of the Hittite capital - but Spain will probably get more of the west. In my game, Spain and France were largely equal in size and power; in most games, Spain will probably be much larger and stronger, France a bit of a runt civ. Also, aside from Babylon and Iroquois fighting in the early medieval period, the AI civs never fought each other, only me. I'd be shocked if that was the same in other games! It will be interesting to see how conditions differed under other circumstances. Oh, and Babylon triggered their golden age in that war and got no fewer than THREE great leaders. Too bad the AI doesn't know how to use them to form armies! That's clearly the reason the Iroquois - the early leader on the other continent - got roughed up so badly and became backwards in tech.
Speaking of technology, after the game was over, I clicked "let me play a couple more turns" and checked out where the AI civs were on the tech tree. Most civs had Rocketry, Computers, and Fission as their Modern Age techs at game's end, but the Zulus were the tech leaders (amazingly) and also had Miniaturization. Iroquois were far behind the others, not in the Modern Age and still lacking Atomic Theory/Electronics. But, seriously, the Zulu as the tech leaders?! I'm not sure I've ever seen an AI Zulu hold the tech lead in anything other than a runaway AI situation. And Shaka lacked coal and had no rails either! I can't even begin to figure that one out. 
Maybe it's because the Zulus got that settler from the hut, or because they were furthest away from Rubber Foxhole.
One more picture to show, take a look at the culture graph right before I launched the spaceship:
The surprising thing is that I start off even with the other civs, then just keep increasing my percentage throughout the game. Usually, the player gets a good percentage early on, then that share decreases as the AI civs get around to building their own cultural buildings. But nothing like that happened here, my culture just kept going up and up! And - I wasn't expanding my territory either, which would also account for it. No, instead the AI civs locked themselves into mobilization and saw their culture plummet as a result. Spain's culture is the best example of this; they were mobilized for so long that their culture was clearly much lower at the end of the game compared to before all the fighting started. Very unusual - does the AI always mobilize this much in Conquests? I don't think that's such a smart thing at all for it to do.
So, what did I pull out of this game? Well, for one thing, removing the ability to prebuild wonders actually made the races much more enjoyable (and nerve-wracking!) to play out. The AI civs were able to put up a much stronger challenge as a result, although this was of course a variant game. In short, I liked having wonder races which couldn't be affected by prebuilds - but it would be nice if the AI civs had to play by the same rules! (And it would also be good not to suffer such a huge loss for a failed attempt at a wonder, which fortunately I was able to avoid here).
Another thing that clearly jumped out from this game was just how bad the Civ3 AI is at managing its cities and laying down tile improvements with its workers. The screenshot of Madrid in the Ancient Age is extremely instructive in this regard. The bonuses on the higher difficulties tend to hide this fact, but the AI is basically completely incompetent at building wonders. This is one thing that will definitely need to be addressed in the next Civ game. *crosses fingers*
As far as AI diplomacy... Well, I wasn't able to affect diplomacy in any way in this game, so it was interesting to see how things played out. The AI would occasionally offer me a deal, some of which I accepted, but most of which were laughably bad. I was able to trade for French furs at one point, but after 20 turns Joanie would not renew the deal, and I of course couldn't ask for an extension myself. The AI civs, of course, traded their luxuries back and forth for pennies on the dollar (when I investigated AI civs with my diplomats/spies, they inevitably all had 8 luxuries once Navigation opened up worldwide trade). I'm sure they were doing the same thing with techs too - although, to be fair, Civ3 is still LIGHT-YEARS ahead of Civ2 and earlier games when it comes to the AI playing as a team against the player.
The biggest problem though with the AI when it comes to diplomacy is how the AI deals with war. Simply put, any civ will happily go to war if another civ is willing to pay them enough money for it. They are always mercenaries on sale to the highest bidder, in other words. Now in most games, the player simply signs a bunch of AIs to do his or her dirty work and gangpile one individual civ targeted for elimination. Here, I couldn't affect diplomacy in any way, and so the inevitable result - which I saw coming from the moment that the Zulus sneak-attacked me - was the entire world getting locked into an Always War situation with my civ. The logic of the game dictated that I would reach that situation eventually; I'm just surprised it didn't happen in my first war. This despite the fact that my civ was by far the strongest in the world and had enjoyed peaceful relations with almost every other civ throughout the game. The AI needs to be able to decide whether or not it's in its own interest to jump into a war instead of ALWAYS be willing to fight for cash.
I could talk more about the problems of the AI in terms of actual fighting, but Sirian already explained that in great detail in his Epic 47 report, so there's not much point in elaborating further. One thing I will mention is that the Civ3 AI has no clue how to use its naval units whatsoever. The AIs all build these huge navies, and they even know to use them in stacks (which is good), but they then do nothing but bombard cities or tile improvements to no effect whatsoever. I killed them just to get rid of their bombardment animations. The AI does a little better with planes, but still is mostly clueless. The AI certainly doesn't understand what to do with carriers; they might stick some planes on one, but never to do anything other than bombard another unit to no point. And we all know what the problems are with the Armies in Conquests (which the AI doesn't use), so no need to go on about that.
One final note - the AI civs seem to like war mobilization WAY too much in Conquests. Declare war on them, watch them mobilize, and see their culture go into the drink. If there's going to be wartime mobilization in future Civ games, the AI had better be a little more cautious about how to use it, because the performance here was just sad.
This game was a microcosm of what is good and bad about Civ3. The Ancient Age was a blast, balancing expansion and discovery with some tight wonder races and slick city management - I loved it! Couldn't stop playing for several hours when that Pyramids race was taking place. The Middle Ages were also enjoyable to play, as I was able to continue developing my civ and engage in more challenging wonder races. Taking prebuilding off the table forced me to plan both research and construction much more carefully in order to get those wonders. By handicapping myself on wonder construction, it was almost like facing intelligent opponents, heh. The scenario design for this game worked very well for me all throughout the early parts of this game, and I really did enjoy it.
As always though, Civ3 simply breaks down around the Industrial Age, and my enjoyment rapidly dissipated after reaching that point. In part, it was due to having to fight all those wars - but that's part of the problem, in that I couldn't see where the industrial/modern resources were and the AI civs could. There has to be a better way to implement things than that (well-known) gripe about Civ3. And... once I got factories and rails, the game just couldn't challenge me any more. This isn't exactly a new revelation, but it's such an important point about Civ3 that I don't feel bad about making it again here. Conquests only made things worse by foolishly adding things like super-strong armies and lethal bombers, with the net effect that wars (and the game in general) is simply easier. It should be clear that the last 50-100 turns of this game were just an absolute pain, no fun at all to play through whatsoever. And the fact that THIS is what happens when the player's diplomacy is taken off the table just shows that the AI needs to be reworked substantially for the next incarnation of Civilization, particularly to prevent the game from becoming such a runaway for the player after the start of the Industrial Age. The Civ3 AI is quite strong in many respects, but it would be nice if it could somehow understand not to fight wars that it can't possibly win.
Well, that's all I have to say for now. Nothing ground-breakingly new to report, but a good summary of some of major problems with the Civ3 AI. I'll have more commentary on report day.
