After my last post, I've been thinking about an issue that's sort of related to "Exception Based Design", but it moves in a different direction.
I'll explain it with an example.
3rd Edition D&D (and 3.5) released under the open game license were great for the hobby in some ways, but terrible for the hobby in others. They were possibly the ultimate incarnation of "Exception Based Design" where every sourcebook created new exceptions to the core rules, and every character filled specific niches in game-mechanisms and storytelling potential. In the mid 2000s, there were dozens/scores/hundreds of companies producing sourcebooks with the core System Reference Document of the OGL as their basis.
Every one of these wanted to have a point of difference about it, so it created new exceptions.
Eventually, there were exceptions enough to cover everything. If the fighter had a range of feat that worked off the strength attribute, now the sorceror and bard had mechanically similar feats based on the charisma attribute (ie. you could fight people with your words).
In a similar example...
Magic: the Gathering started with green basically being the colour of creatures, white the colour of protection, red the colour of direct damage, blue the colour of weird stuff that you did to the deck, and black the colour of graveyard manipulation. There was always a hefty bit of overlap between adjacent colours and a little bit of overlap between all colours (eg. everyone had access to creatures of some type), but generally the colours kept their themes intact.
The problem comes with new expansion sets, and the fact that eventually there are enough card variations floating around that any colour can produce almost any type of deck. I've seen green decks that work through deck and graveyard manipulation, blue decks that are predominantly creatures, and all manner of effective decks that have gone against the original principles of the colours.
This seems to mainly be a problem when the potential of a game expands beyond its original parameters, possibly a symptom of success that most small press companies wouldn't need to worry about. But at the fundamental level, it's a factor of design choices from the beginning.
If a designers specifically wants different aspects of the game to play in different ways, this creates an instant level of strategy within the mechanisms. Some players will be drawn to playing one way, and telling one type of story, while other players will be drawn in different directions. You don't want too much of the one flavour, because it narrows down the potential audience; conversely, you don't want too many competing flavours because it just dilutes the whole thing into a jumbled mess.
I don't really have an answer for how this balance can be achieved, or how it can be maintained as a game expands beyond its origins...it's just something that I'm thinking about.