I have long held that in RPGs there are three levels of design - the Designer level, the Group level, and the Individual level. All games incorporate all three levels into the design, but proportions of each vary enormously. Traditionally, the GM does almost all of the Group design, where the other members of the group concentrate on the Individual aspects of the design - their characters, backgrounds, and things their characters control; but this is not my decision. As a GM, I prefer to include the rest of my group in the Group level design process, but I leave that open, as how it is done is a Group level decision.
In recent times - I have been running games since 1977- I see more and more auteurism in game design - designers who dictate in various ways how the game is played, and what the game is about, in increasing detail. I am not an auteur. I feel my job as designer is to help you play the game you want to play.
In my designs I do my best to push as much as possible down to the Group level by various means, such as providing tools to make it easy to generate setting within a framework rather than dictating setting. I make the system as flexible as possible to accommodate what the group is interested in doing rather than what I, as designer thought of when designing. In different games I approach things in different ways - in some games the method of resolution can be swapped out as required, in others I only frame things, leaving interpretations open. Always I define from center - for example skills generally have a definition which amounts to "stuff like this", spells may be only an evocative name, leaving the actual effects open to discussion, and lately have focused on the group defining what the game is about by designing their own company or association, then designing their characters to fit the association, and defining by their association what the campaign/adventure will be about.
When I develop settings, I try to keep the detail level as low as possible - working in broad strokes for the group to refine and detail themselves. I try not to prejudice group decisions by giving examples which are as broad and fundamentally different as possible. By this I encourage groups to think of these examples as arbitrary directions in a sea of possibilities - for instance in StarCluster 3, my example companies included among others medical groups, insurance adjusters, police and law enforcement, and religious groups.
What I set out to provide is ideas and settings and cultures that fire up one's imagination on what one can do with them. That is the living, breathing center of a designer's job, and always has been. This is why pure toolkits have never done well - most people do better within a framework than without.
The problem comes when the framework becomes a cage, when the only way to fit your ideas in is to rip something else out.
There are many ways in which designers do this - lock the group level imagination in a cage, I mean. One common way is to provide too much detail in a setting, then interrelate everything. Another is to tightly intermesh carefully balanced components of a system mechanically, so that one daren't tinker lest the whole edifice tears itself apart. Yet another is to circumscribe the setting and system 'til the game becomes a single scenario with a sparse attached system and many times pre-genned characters, just competent to deal with what is in the scenario - though this is better than the previous two as it's easier to add than to tear apart, then add! Still another is to lock down the variation possible between characters, making strict types, each with a specific function.
The net effect of all of these is to increase the Designer's share of the input into the game at the expense of Group and - less commonly - Individual input. There are big benefits to these methods, which is, of course, why Designers go this route and GMs stand for it. One can pick these games and play them as is without a lot of setup. One doesn't need to think too much, as choices are controlled and managed. They are great for one-shots. They are uniform across instantiations. They limit bad games by limiting possibly bad Group level choices. These are all solid, tangible benefits. The "caged-in" effect not a goal, but an unanticipated and perhaps unwanted side effect for designers working toward the benefits listed.
Umm, I guess there's more, but I could write pages and pages about this subject, and I'm salmoning - constantly swimming upstream. I find most GMs/groups - the vast majority actually - don't mind the Designer doing all the designing, they are just there for the ride. This makes me sad when I think about it, and I lose energy.