I really don't like using rationalism in this game. The problem is that, with research agreements, multiple cities running max scientists, and a high population, research speed is already so fast that my production can't keep up.
You're racing the AI to your win condition. If it isn't culture, your win condition on a sufficiently large Deity map involves advanced tech. Either you're going to win in the UN, or you're going to build a spaceship, or you're going to drop a bunch of GDRs (or similar massive military disparity) on the AI, or you're going to nuke the snot out of it.
The more you can accelerate the time to reaching your win condition, the less opportunity the AI has to do anything about it. Rationalism increases the efficiency of your Science specialists by 67%, which is substantial. It also triples the Science productivity of any pop unit working a Trading Post.
Clearly you like to build things; your ICS approach demonstrates that. But when you can end the game right after turn 200 and stay a full tech level ahead of even a juggernaut AI, there really isn't any reason to build a lot of stuff. If the juggernaut is on another continent, it will never have time to reach you. If you share a continent with it, early Artillery will make anything it throws at you look just silly.
Couldn't disagree more. I don't see any diference between winning by culture and any other win in civ Iv in terms of paradigm. Care to develop this point ?
Optimal behavior varies. In a cultural game, you have a minimum threshold of number of cities that you need to achieve. Anything beyond that is superfluous. You need a few late game techs and not much else, those techs result directly from the standard Deity beeline anyway, and you pick up sufficient military techs along the way. So expansion is not really a priority; the marginal impact on completion time is minimal, because you don't really feel the effects of +pop -> +research all that much.
For any other victory condition, expansion to the limit of your ability to finance it (and still keep the research bar moving) is always highly desirable. Domination requires doing so. Spaceship requires keeping the Science bar moving, and
appropriate expansion through conquest (no lengthy wars against hardcore unit-spammers) will always increase research speed. Diplomatic wins require votes; conquering citizens directly increases your vote count, and there are ways to game who you target to improve your diplomatic relations as needed in the process of expansion.
I can't speak to anything after vanilla, as I didn't play it. But that doesn't really matter here, since comparing Civ4 + xpacs to Civ5 is a blatantly unfair comparison. We should be comparing vanilla to vanilla.
I really disagree with this. Sure it's possible change civics in Civ IV, but you try to avoid doing it as much as possible. Not only because of anarchy, but also because your empire is usually set up to take advantage of a particular set of civics. For example, if you've built a lot of workshops to take advantage of state property, it's not easy to change into something else.
You're talking extreme late game, and you're stacking the deck by discussing a Civic that requires significant investment to exploit. Most Civics don't work that way. You would prefer to run many of them for brief periods, but not always. Anarchy is indeed a significant cost that precludes brief swaps for non-Spiritual leaders; a Spiritual game plays quite differently. Some Civic choices are obvious (Hereditary Rule, Slavery on acquisition); others are much less so.
There are just a lot more opportunities for meaningful choice along the way under the Civics system. With the Social Policy system, optimal play always requires determining your endgame early and selecting policies accordingly. If you want to win by Domination, you need to build a lot of stuff in a lot of cities. If you want to win by Diplomacy or Science, you need a small number of big Hammer producers and a lot of Science output. Culture is its own SP animal; the proper approach is fairly obvious.