Is it advisable to run a campaign where the GM occasionally rotates to be a regular player while a player becomes the GM? I've seen some mentions of this strategy to prevent the GM from getting burned out from always planning, but it seems like having an alternative game going would almost be better. That way you don't have one GM giving out treasure beyond what the PCs should have and the other GM having to deal with it.
It's been a long time that I've played in a group where only one person ever GMed anything... As specific games rise and fall, usually someone else will run something sometime, unless they are a control freak or everyone else is totally slothful (this was the case when I was in high school, though, to be fair). But nowadays, we always have multiple concurrent campaigns going on with different GMs running them.
Rotating GMs within a campaign is different - to a degree, it depends on the nature of your campaign. I've run campaigns that are very coherent stories, with loads of secrets, that I'd never rotate in the middle of. If there's a concrete vision, you don't want to rotate. If there's less of one, it's easier - kinda the "Babylon 5 model" vs the "Star Trek model." So rotating within one actual campaign is possible but is more or less desirable depending on the type of campaign.
In one campaign we proactively said "Hey, let's deliberately rotate every player in as GM." We wanted everyone to get a shot behind the screen, learn what it's like, and give us all insight into strengths and weaknesses, so whenever one adventure finished up, we handed off to the next person. Even the ones that really sucked at GMing had something specific they did great that we learned from - dialogue, pacing, whatever. The campaign conceit was just that we were all pirates and were roving around on random adventures, there was no huge metaplot. We never had any problems with canon conflict or whatnot as a result, and everyone got some GM trigger time. It also made those who seldom GMed appreciate the GM more.
Try the Run Club group structure described by Ben Robbins.
Here's how Run Club works:
1) Every month (or two weeks, or whatever works) someone takes a turn and runs a game. One-shot, short game. No campaign. No big picture. Just a single game.
2) Everybody who plays will GM. Everybody. This is the core principle of Run Club. You cannot play if you will not GM. That's the pact.
3) When everyone has run a game, the round is finished and you can start over again.
That's it. Simple on the surface, but in that simplicity a number of complex issues are addressed.
That may not be a great fit if what you're looking for is a long-running, coherent campaign rather than a series of one-shots, but it's a solid idea if your primary goal is to share the GM load around and make sure everybody gets a chance to be a player.
Gary Gygax and Rob Kuntz rotated referee duties in the old Lake Geneva Greyhawk campaign in the 1970s. In almost every long tabletop campaign I've run, and in most of the ones I've heard of, sooner or later one of the players has an adventure he or she has wanted to run. Without any exceptions I can think of, it's been fun to step down from the referee seat for a while and play, and to let the other players get a break from my style of running things.
It does require loosening up a bit. You also have to be willing to accept that someone else may have a little control over your setting for a while, and you may have to deal with changes not of your making. If you're an "amateur author" type of referee, this may be difficult. I've always thought of the unpredictability and shared control as design features, not bugs - for the same reason, sometimes I ask other people to design dungeons or the like for my settings.
Every famous setting has, sooner or later, become "everyone's." It's part of the fun.
In most of the groups I've RP'd with over the years, we rotated GMs periodically, as there's always someone who has a neat idea for a campaign.
However, we've never shared campaigns. Personally, I'd feel to constrained as a GM. How much can I alter the setting? Which NPCs can I kill off? Can I flood or otherwise destroy cities?
My group runs 3 separate games for when the GM gets burnt out. It sometimes gets hard to remember all the different stories, but the players like making different characters to play.
On the other hand, if you want to blow your players mind, work with the other DM and have your two stories come together in a big explosion that melds the worlds. After that, kill the other DM so he doesn't destroy your world.
Rotating GMs is more of a modern development; a lot of the new-style GM-less games effectively rotate GM on the boundary of every scene or every conflict. What you're talking about is certainly a lot more traditional architecture, but I've seen it done extremely well.
What rotating GMs per-story or chapter does require is that the setting as a whole be accessible and either understood by future GMs or be accepted as a very plastic set-up by everyone. This could be as straightforward as the first campaign having been very clear and revelatory, so everyone's on the same page for the next story through, or as deliberate as using Universalis, where everyone constructs the setting together very deliberately, or simply by creating and maintaining a group-wiki as if doing a round-robin writing exercise. Or you could use a well-known, already extant setting like Star Wars, Star Trek, or Lord of the Rings.
Ultimately, its going to come down to figuring out how much you can bring people into understanding the setting and create a desire for them to tell stories in it.
We have rotated GMs within a single setting and did not feel it worked that well .. it felt disjointed.
That said everyone takes their turn GMing using their own settings. We have a couple of people who love running long campaigns and others who dont really enjoy that and run one off games.
The best GMs are ones who still play and the best players are ones who also GM.
One game method that endured a summer was referred to as a "Fractured Multiverse", but it took a lot of preparation to run properly. Each player at the table was responsible for running one game system, and every player made a character dossier that was the basis of making that character in each system present. Whenever a DM/GM/ST had a session to run they would take over and the party would be shunted into that player's universe and the story would continue there. So the Bard in the D&D game would be a rock star in the OWoD setting, then get pulled back to being a musically talented priest in L5R, then catapulted into being a one man Daft Punk in a Cyberpunk game.
This way everyone gets their own arc, the ability to keep the same characters, and a modular state for whomever shows up.
I run a long, complex campaign with hundreds of recurring named NPCs, tens of recurring subplots, lots of politics (many PCs are nobles) and some mysteries. Sorry, it wasn't exact: we run a long, complex campaign with hundreds of recurring named NPCs, ten or twenty of longer plots, lots of politics and some mysteries. And it works.
I ran the game for four years alone. I participated on the creation of the game world years before start of the campaigned (it have some 10 e-books issued officially, but there's much more I have prepared and didn't write, or didn't finish) and I'm an "amateur writer" GM, so everything's quite complex. But then I had less and less time for game preparation and I decided to stay in the city I studied in (quite distant from the town where we play, though I still visit my original home quite often), so I realized that with me as the only GM the campaign is not sustainable. Now we are in the "test" phase - I'm still the main GM, but my brother runs some sessions too. Other GMs might join soon.
How do we do it: I control the main area of party's operation and most other lands and my brother has everything to the east from the "old" region (a crime-infested city, wild plains and on of foreign countries). We regularly discuss the big picture: gameworld (meta)physics, plans of important NPCs/ organization, NPCs and events affecting both areas, preparation of sessions affected by NPCs from domains of both of us (I'm looking forword to the next session: assasins including a group created by my brother will invade PCs' home, I'm the GM) etc. We are brothers, our gaming style is very similar and we don't play every week-end I'm at home, so maintaining a common big picture is no problem. Even as a GM who tend to strictly control "the big picture" I feel perfectly comfortable with the plan that my brother will become the main GM for whole gameworld one day.
For other potential GMs, this amount of coordination is not possible. But we solved it: the PCs found a gate to other worlds, turning the setting to something like Planescape. As long as everyone follows few basic rules (similar gaming style, never have one's PC there, plus few fiction rules like "no time travel"), there should be no problem as long as each GM controls their own world (we didn't test it yet).
Few things to consider:
gaming style - the styles of all the GMs need to be compatible. There's no problem if one prefers mystery solving while the other one prefers hack'n'slash, as long as the players accept that this is the same campaign. And most players accept a lot, at least if it's just a session or two followed by return to old order. Topics that might ruin the campaign are equipment and thus power the PCs have (we have introduced mechanism of "translation of power" as a part of travel between worlds, so if the PCs found a machine gun in some other world, it will be a piece of scrap metal or at best a moderately powerful magic staff in any mediaval fantasy world), deadliness or how likely is a PC to die (we have a houserule that a PC can die on a mistake or bad rolls, if the player agrees - otherwise the PC gets some points in disadvantages or reduced stats and nearly escapes or is ressurected; in some of the other worlds, the PC's body is just a projection of astral body, so death means to awake wounded at home) and how "safe" or "unsafe" the game is (we don't have any strict rules yet, but I would insist on discussion with other GMs before introduction of any "problematic" topic into game). Insisting on "better roleplaying" or failing to keep with roleplaying standards (most of my players are or were swordsmen, so a bossfight without good description of combat maneuvers would disappoint them) is almost as good way to disgust players as pushing players used to freedom to a railroad. It's good to discuss these things before the first session of any new GM, plus to have some guidelines.
are different systems OK? - the previous paragraph should be copied here, but everything is more difficult. Having a different charsheet can break the identification with the character much more than learning that the PC is hopeless outside ordinary hack'n'slash campaign. Make sure that the other players will find what they like in the other system too, and that they understand the concept. Mixing DnD with Wushu is a bad idea, even though the PCs' everyday adventuring life wouldn't change so much. Even if the games are compatible (I know a guy who played DnD 4e, GURPS and Exalted, still in the same campaign, so it is possible), frequent changes of the system make the "translation" of the charsheets between systems very important. If any of the systems have a concept of "experience", you have to give the PC experience in each such system after every session. This means that either game realists will be disappointed that characters are too different between systems, or powerplayers will be disappointed that their excellent build in one system is not so excellent in another (munchkins will complaint for this, but they will exploit the system anyway). So, I wouldn't recommend the game system rotation proposed by CatLord.
how much will knowing something ruin my immersion to the game? The more mysteries in the game, the less should each GM know. It's also specifical for the player and for the system (some systems are better suited for director stance anyway). If every GM has their own world or region, most mysteries are region specific, so there's no problem.
how are the areas governed by different GMs connected? - the more connected, the more information you have to exchange, the bigger need for communication and the more likely some point will be spoilt. Planescape or a dimension-travel campaign and a game world per GM is best - there is very few you need to share. A planet/ country/ city per GM is worse, but the more empty space or wilderness between them, the better. But cultural "space" is what does matter here - 20 miles through Iron Curtain is much more than 200 miles within the same country. War is the best area-of-operation separator after wilderness or a mountain range that is hard to cross (the only drawback of a warzone is that there must be some referee or good rules, because GMs on both sides want their kingdom to win). Be careful with seas or empty space between planets - is there's enough traffic, they don't separate as they should. Anyway, any plot concerning travelling between the GMs' areas is hard to coordinate and has to be discussed with others; avoid it if you are not all experienced enough, or if you are not used to teamwork.
level of detail - the more persistent game facts and the more mysteries, the harder. Not the worse, just the more is at stake. If you improvise everything, a starship from another GM's planet is not very important, even if it has human crew, although the planet is inhabited by lizardmen. If you prepare everything, a starship with human crew coming from the lizardman planet might mean a hard time for the other GM - but if you discuss it beforehand, the GM owning lizardman planet might already have (or invent from scratch) a plan with human spies working for lizardman emperor. That's why I think this might be good - detailed universe means that you have to discuss everything, but more heads know more and can create better story.
level of power - if the PCs are ordinary rangers, most NPCs they meet (and can overcome) means no difference in any GM's planet/ country/ city. If the PCs are one-man armies or polititians, it's not so easy to invent their new peer without breaking plans of the GM controlling the NPC's area of operation. The more power level, the harder it is.
EDIT - this is what I forgot:
one adventure = one session - have a chance to escape after each session, unless the GM proved their skill to run the game. The players might not be satisfied by the new GM's performance, or the new GM might find out that this task is too difficult, or the GM can have not opportunity to participate for few weeks. If the PCs can take the first ship going home, it's not a big problem. If the PCs are trapped within a dungeon in the middle of nowhere, with a cave-in behind them a three-sessions-long dungeon in front of them, it's not easy to avoid anticlimactic end of the campaign.
trust the new GM, but not too much - always think before allowing a player to run a game, and think twice before allowing anyone who didn't play at least ten sessions with you. The person might not be a good GM, or might be a good GM for a style incompatible with yours, or might simply misunderstand your campaign. The point of previous paragraph is this: always let the new GM (new for your campaign) to prove their qualities in a session where they can't make too much damage to the campaign.
Having this in mind and having good players open to discussion, the GM rotation within one campaign shouldn't be a problem.
This is usually very hard because it exposes the players to the story giving them knowledge that can spoil the game for them.
There is an excellent section written on this in the D&D 4.0 Dungeon Masters Guide. While targeted at D&D, the information is relevant to all tabletop games.