One important “nerf” for Chinese was the Definitive Edition. Earlier in team games, positions were random, so flexible civs like Magyars, Chinese, maybe Huns and Mongols, were often picked, because pocket or flank, they will do the job.
Chinese are a bit like Burgundians, you don’t notice the fact that they are OP in terms of the sheer army numbers they can pump out (they can do that too but other civs like Mayans or Huns are better at this). You notice that in spite of killing army, and taking fights at his base as opposed to yours, etc., you always seem to be behind, the damage to his eco is never quite enough.
They have a lot of cheaper things which allows them to nearly always have a far smoother boom than you while making army (e.g., 1 TC all-in by Chinese might add a second TC shortly after due to the resources saved on Blacksmith and eco upgrades, which you typically can’t do with other civs)
We seem to be fixated on the extra vills to start.
Cheaper transitions are a huge deal. It means if your opponent throws a curveball, you can tech into a proper counter quicker and have them ready to go with less investment. The defensive capability should not be understated. Their very basic gameplan is also very, very good, with one glaring danger in Onager to consider.
1 Like
Thats also true. But the effect is the same. Transition efficiency scales also with elo exactly. But it’s more a subtile improvement whilst the special start will be trained by most players at some point a bit.
I also never pick chinese, but for the case if I get them, I played like 10 x the start so i know what I have to do.
The standard methodology for analyzing a causal effect like this (i.e. what is the causal effect on win rate for Chinese) would be:
- Ensure all players know chinese (and the game) very well
- Randomly assign players Chinese or have players always go chinese with random assignment to opponent or control for non-uniform civ matchup distribution.
- Use some (generalized) linear model to extract the win rate with some controls like map type and elo diff.
This would give you the ATE for assigning players chinese. You could also with enough data break it down by civ (i.e. giving chinese to the opponent vs civ Z has an effect of Y).
Why the first bullet point is there is because in mechanism/game design you cant realistically include unknowledgeable agents. Including the behavior of unknowledgable agents is inherently not robust w.r.t. time/more knowledge. But we want this estimate to be robust in this respect as you cant make balance changes based on estimates which arent robust in this respect (though you can do design changes and side-grades).
Realistically no one but civ-pickers (maybe like 50 games in short succession) and the pros (maybe not even every pro) know chinese well enough to use in this estimate. Playing chinese once every month even at like 1800 elo is going to leave a lot of holes which can be exploited.
I doubt such an estimate would prove chinese are OP simply because they have weaknesses like MAA and strong siege. But at the same time this data isnt really the way to go about it. You could certainly use this data to justify that chinese need a slight redesign to avoid this massive gap that no other civ has though.
2 Likes