Is that really such a terrible issue? Different games play out in different ways. Sometimes you get peaceful AI neighbors, sometimes you get warlike AI neighbors. Sometimes you get a strong starting position, sometimes you don't. Some games are easier than others. I don't think that's a problem with game design, it simply shows that there are always some random elements to each map.
It becomes a problem when it combines with other gameplay elements to create un-winnable situations (either completely or for a given skill level on a chosen difficulty) or when it breaks the original intention of a "fun" and/or "role play" AI in the first place.
It's not like "fun" by Soren's definition = unpredictable, either. A while back, DMOC posted some deity games where the player was hidden off in a part of the world the AI would never, ever see. He would then simply run the game until one of the AI won. He only got a few games posted that way, however I was able to accurately guess the winner in each as well as several trials of my own. If they're that predictable, how is it they are "fun" exactly? Having AIs randomly do nothing/not try constitutes a "fun" AI? In a game where there can only be one winner, arbitrary activity that props one side up isn't "fun", it is fake difficulty. Maybe you disagree with me, but I don't see a place for fake difficulty in civ...or most games for that matter.
It was also particularly annoying that super AIs would deliberately defend culture (via peacevassal) AIs in what amounted to an active attempt to lose the game and take the player with it. That's..."fun"?
Far more engaging would be AIs that actively attempt to win...in separate, viable ways. Civ V misses on the "separate, viable ways" bit. Combining forces against a stronger opponent, making temporary alliances which mutually raise chances of victory against the remaining opposition, and pursuing different kinds of victories can make competitive AIs a lot more compelling than "ohai I'm gandhi I build no units and go for culture consistently...come kill me quickly before shaka does it or instead kill shaka and save me for later".
Deliberate AI stupidity + bonuses do not a good game make. Competitive AI + reasonable bonuses + legit balance of options makes a much better game. Going back to the "fun AI" model would doom civ V...thankfully they know better.
This. This random factor, luck, what-have-you, is the one thing that makes me play Civ4 for years.
Luck should only have so much impact. Entire outcomes should not be based on luck alone. Veteran players ran into this problem however; on difficulties where luck had no chance of deciding the game on its own, there was no challenge for them. On difficulties that truly tested the player, it was possible to lose on turn 0 if the RNG felt like it...or some close approximation.
It's like a more extreme version of events. Most events were pretty benign, but Vedic Aryans early were a guaranteed kill shot, and things like bermuda triangle = "LOL your 100 destroyers/battleship escorting 40 full transports sinkz nao" were also beyond stupid and a great way to insta-lose in a game that takes people hours (fortunately I was one of the fastest players). Things like this start to matter more when a XOTM or MP game is stupid enough to leave them on and they single-handedly decide the outcome.
For every person who likes IV's luck-based gameplay, I hope you get Vedic Aryan'd 10 games in a row. In MP.
And in the Civ4 material, Soren and Dorian sounded more like "We thought stairs would be a good idea for our building, but after we tested it, we found out how important it was to provide easy access, so we scrapped the stairs and built ramps instead." You can see why my preference is with the latter approach? 
