One issue with making games, at least as a commercial practice, is that the finished product will not be perfect. This lack of perfection rarely affects the bottom-line, unless it's imperfection of a particularly fun-killing kind. But it's always amazed me at how willing end-users are willing to put up with buggy, clunky, and otherwise lousy products so long as it doesn't interfere with whatever it is the players keep buying. However, part of the nature of games is that they're complex objects, and often more complex than the players playing them are prepared to understand. Even simple toy-games like Hawk-Dove can confuse players about the optimal strategy to play, and there are plenty of games out there with multiple strategies, of which most of those live strategies are considered 'dead' by the players because they only see their own personal experience and the anecdotes of others, rather than the entire structure of the game.
So when a designer needs to go back and 'fix' whatever bugs exist in a game, they need to consider more than what might work in terms of the mathematical model. The designer needs to consider the usability of a fix, as well as its mathematical properties. The player's perception of the fix is probably more important than the problem it fixes, or fix itself. Anyone who's had to wade through articles containing the terms "nerf" and "buff" probably understands why that it is a problem. Because players have a diverse range of interests, and the more interested they are diverse it gets.
Designers handling usability testing often understand that they need to filter the feedback they get. Some players will assert ownership of the game and take umbrage that they aren't given the game to redesign to suit themselves, and other players will simply be too inarticulate to provide any useful feedback even when they actually have some beautiful insight into the game. Sometimes, if you're really unlucky, those players will be the same player. And this player, for whatever reason, will have a disproportionate effect on how the game is perceived, because this is the sort of person with the time and inclination to interpret their NDA selectively (because, after all, it's their game) and play whatever it is they imagine to be the devil's advocate.
Part of re-designing a game is going to be selling the fix to both the tests and to the greater public. It's not enough to say: "Here it is!" You can have the greatest game in the world, but if people don't know that, don't feel inclined to try it, and have been warned against it by a jilted tester, it will fail to be experienced by its end-users. Admitting to a mistake and offering a fix is the wrong way to do it, which isn't to say that doing so is bad. Ideally people can admit mistakes, fix mistakes, and move on. But there's usually a background of stuff going on that makes it irrational to do any of these things. It's the nature of a plurality of goals and methods, as well as intelligences, that what's optimal in a vacuum is often sub-optimal when indexed with other games. Instead a designer needs to explain the fix, and not how it works, but how the fix offers value to the end-user.
Balancing a set of rules is not a matter of balancing columns in an account book, or the sides of an equation, or the payoffs in a table. Balancing a set of rules is a matter of selling end-users on a product. Which is problematic if the value you are selling is not easily apparent to a plurality of end-users. Take Warhammer 40,000, for example. Many fans complain that Space Marines, in all their variations, are un-balanced. The Space Marine armies are too powerful, in some vague and never-fully-explained way. It doesn't matter whether this is true or not, but whether it's a matter of popular perception, because what matters is who is buying the game. And if the Space Marine line of products is financing the rest of the products which actually balance the game, despite those other lines of products being sold at a loss, then the other lines need to be sold better, rather than doing anything to undermine sales of the Space Marine line of products.
Because that's what sets of rules are, ephemeral objects that they are: Marketing. While people buy games, and buy products associated with games, and play games, they are just doing so as a buy-in to rationalize owning all of the gaming products being sold. Games companies don't sell games, they sell a common culture or wavelength that put the owners of gaming products together for a social experience. Re-balancing a set of rules goes some distance towards churning the market for such things, as even the greatest social experiences tend to devolve into boredom. Doing it properly puts more people on the same wavelength, creating a space in which customers become clients, and want to come together. Doing it badly may result in a better game, but ultimately narrows the space by appealing to special interests at the expense of the broader market.