tl;dr version is that whilst one unit per tile (1UPT) and a hex system can be good ideas, they don't lend themselves well to a game with Civ's scale. A hex game of the Eastern Front in WW2 isn't going to be seriously if there are fewer than several hundred tiles between Odessa and Koenigsberg, but Civ V doesn't come close to this.

So by the end of the game you have a ridiculous 'carpet of doom' where every single tile in the world is occupied. And they knew this problem existed, so they tried to make it hard to build those big armies, but that only leads to different problems, specifically that you don't have any reasonable sense of progression, indeed progress is hindered more than benefited by, say, buildings.

Efforts to patch the game thus far have largely revolved around this. If X is underpowered, they solve it by nerfing Y down to X's level. They have to, or the carpet of doom issue would arise even sooner than it does. As the article says, it works after a fashion, but it's hardly fun.

AI seems entirely random. Granted the very obvious math of Civ IV in terms of leader attitudes might have been a bit too simple, but this is not the correct solution. Even though you can now see leader attitudes to some extent, it doesn't feel like these have any bearing on how you act. Obfuscating diplomacy to a degree I can get behind - this just feels like you're dealing with dice rolls.

Also the game has regressed to pre-IV Infinite City Sprawl being by far the best strategy.