A Matter of Trust
Trust is a complex, multi-faceted issue, but there are a few aspects that seem easy to identify. We expect the system will be well defined, commonly understood and fair. We expect the GM will take us somewhere interesting. And we expect the other players will positively contribute to the story.
The interesting thing is, in some ways, one type of trust can replace the others.
Players who don't trust the other participants will often want a system they can depend on. In many ways, explicit social contracts are just a method for building trusted systems.
Alternatively, if you just mistrust the GM, you may try to distribute much of his power to the other player, or eliminate the GM entirely. If you mistrust the other players, you may want a strong GM to keep them in line. Finally, if you don't trust the system, you may de-emphasize it, relying on the GM and other players to just "wing it."
Note: none of these compensations are inherently bad. But, they all have a degenerate form.
Trusting the system may lead to rules lawyering as each participant tries to use the system to defend their concept of their character, their story and even the flow of the game itself.
Social contracts, by definition, limit the type of stories that you can tell. If you sufficiently trust the other participants, they may be able to escort you outside your comfort zone. This can lead you to more powerful, more interesting stories than your explicit social contract would allow.
GM-less games tend to emphasize emergent stories. Unfortunately, it can be hard to maintain the story's structure or any real consistency. There's no guarantee that the gun I placed on the mantle piece in act 1 will actually be shot in act 3. Most likely, it will be forgotten, leading to an inconsistent and somewhat scattered story.
On the other hand, relying on a strong GM forces the GM into a dictatorial role. By definition, this limits the type of choices and the amount of influence you have as a player.
Finally, "winging it" makes it more likely that the participants will have inconsistent or incompatible views of the system. Even with a single, well-defined, hard-and-fast set of rules, each player will have his or her own interpretations. Add another layer of fuzziness, and the players' mental concepts begin to drift.
Clearly, this is not a clean or concise science. And, to make things worse, every gamer has a unique set of trust issues, with varying levels of tolerance.
By default, most players seem to trust the system over the other participants. In many ways, this is the most pragmatic choice. You can alway buy a good system--finding good players can be much harder task. By emphasizing the system over the other participants, you can play with the widest range of other gamers. Unfortunately, your games will become severely limited to those things that the system handles well. Typically, this means tactically heavy combat.
For myself, I want to play with a GM that I can trust. I need a GM to manage the structure and consistency, and to make me feel like the story is going somewhere interesting. This represents my own mistrust in emergent stories.
Not surprisingly, I have a lower bar for the other players, but I still demand a high level of trust there. The system is the least important to me. I want it to be fast and transparent--but I don't really need to trust it. Not if I trust the other participants.
Obviously, many sets of trust issues are incompatible. I should not play games with people who mistrust their fellow gamers. That's just going to end badly.
Additionally, I think we can get a feel for a game designer's personal trust issues by examining the type of games they build. A lot of indie game designs are reactionary, at least in part. This is particularly true for designers who grew up on mainstream games. We all carry the scars of bad experiences, and we design our games around them.
Personally, I want games that heavily focus on story. This means the rules I design tend to be a bit fuzzy--They need to be fast and they need to be flexible. I trust the GM and Players to keep things fair and consistent. I always want a role for the GM, but will strongly empower the players. I feel the best stories can be found in the chaotic borderland between purely emergent stories and heavily plotted ones. The GM still monitors and maintains the structure, but the players make real decisions that have real consequences. The story's not allowed to drift randomly, nor is it railroaded to a particular destination. The player's trust that it is going somewhere, but they also help define the final destination.
Now, not everything in that statement can be boiled down to an issue of trust--but I'm sure some of my own, personal trust issues definitely shine through.