That's what a mod IS. XML modding does not require huge changes, it's mainly about small updates to the existing database. This is why we point people to Kael's guide; your first instinct, to say that it doesn't really apply to X, is just wrong. It's all just a matter of scale, but the mechanics are the same.
GameInfo/CIV5Eras.xml sets the costs, which depend on the current Era. In my own mod, for instance, I've basically doubled the cost of RAs across the board.
See, the problem with RAs isn't that they're overpowered. It's that the AI has no concept of savings. Let's take some hypothetical numbers for an arbitrary point in the game:
An RA costs 400 gold.
A unit costs 600 gold.
Bribing a city-state costs 500 gold.
So a hypothetical AI has 350 gold, and gains 25 gold per turn. On this turn, he can't afford anything, and next turn (375) he still won't have enough. But then he'll have four turns in a row where his only choices are an RA or nothing. The AI is probabilistic, so it might set the RA likelihood low enough that it'll reach the more expensive things fairly often, but what if it doesn't? The AI doesn't know that four turns of not buying an RA will unlock better options, because he can't plan that way.
End result: lots of RAs, few rushed units or city-state bribes. These AIs are tech threats, but no real threat in the long term.
Now, double the cost of the RA so that it's more expensive than rushing a unit in the same era. Now the problem's inverted; if the AI thinks he might want a unit then he'll rush it, and if he wants a resource he'll bribe a city-state, but he'll buy RAs far less often, only buying them when he thinks he doesn't need more units or more allies. This creates a far more dangerous environment for you the player, because the AIs won't skimp on resources or units just to get techs faster, and once they DO have a strong military they'll then buy plenty of RAs.