So I'm seeing a pattern here (which oddly, I wasn't looking for in the thread I started on a similar topic back before the crash--a slightly different focus can bring out interesting ideas). So far, the categories look like:
1) Homebrew with Selective Borrowing: You see the published material mostly as inspiration for creating your homebrew world. If you keep to a theme or a map or something you consider important, you consider your game to be a <Insert Setting> game.
2) Alternate Universe Setting: You start with all of the assumptions of the published world (according to whatever materials you have available to you) and then selectively make massive changes to a variety of things, which might include well-established world assumptions, maps, races, NPCs, past timeline, maybe even playing with a different ruleset.
3) Canon with Selective Changes: You start with all of the assumptions of the published world (according to whatever materials you have available to you), and assume anything that doesn't come up during your campaign adheres to those assumptions unless otherwise stated. You make selective changes to parts of the world you feel should be changed to better fit your personal vision, including past timeline, NPCS, and other relatively minor elements. Alternately, you might stick closely on pretty much everything with One Big Change (like One Big Lie on Mohs Scale of Sci-Fi Hardness).
4) Temporally Fixed Canon: You start with pretty much all of the assumptions of the published world as of a particular date--timeline or product publication--or you accept all of the canon from certain foundational materials but not others. Anything after that date or outside of that product set can and will deviate due to the developments of your campaigns or DM's (meta)plots.
5) Established Canon: You follow all of the assumptions of all of the published materials you have available as closely as possible. Naturally, once a campaign has begun, things will deviate from ongoing official canon as in #4, but you attempt to keep up with ongoing canon and implement any elements that you can.
I generally go with #3, with leanings toward #4. I set a date whereon canon up to that date is generally assumed to be correct, but I make subtle changes--most of which are invisible to players not highly steeped in the lore. I welcome input from players who might know more about a setting than I, with the caveat that I might over-rule any element I dislike.
Here are selected examples from this thread of each of the types. Correct me if I'm wrong.
When I talk about running a GH game, or an Oriental Adventures game; or when I say that I am running a module; what I mean is that I am using some maps, some characters, some tropes and themes, taken from the setting or module.
But I don't pay much attention to the "canon" of the setting or module. I've run OA using homedrawn maps and the Kara-Tur boxed set. I'm currently running a GH game, using Burning Wheel mechanics, and I move between my old folio maps and 2nd ed and 3E era ones - whichever happens to be at the top of my folder - without worrying too much about it.
What makes this game a GH one is the basic geography and history (Hardby is a city ruled by a magic-using Gynarch, across the Wooly Bay from the Bright Desert, which is populated by Suel tribesmen). Not the minutiae of canon: the details of the setting I make up as needed for play or determined during the course of play.
I approach my 4e games - one using the default cosmology (but the "world map" is the map from the inside of the old B/X module Night's Dark Terror) and one in Dark Sun - the same way. Follow basic outlines, and use published material where it seems useful, but otherwise without too much concern for what is "canon" in the setting.
I used the above to define #1.
I often create an alternate version of the setting. In my FR, the sundering was very different, for instance. Netheril is still a power, but has mostly ditched Shar as bad deal, Myth Drannor is still intact, there are still earthmotes floating around. Returned Abeir, all the Genasi and Dragonborn, etc, are mostly still there, and aunt her is back, but greatly diminished, because I don't retcon away the fact that Tymanchibar sat on it. Instead, the two worlds were aligned very closely for about a month, and moving between was very easy. This lead to the Unther contingent that was on Abeir to come back, and use the chaos to take "back" a large chunk of their old geography. Except, it's not the same land.
Netheril no no longer has flying cities, but only Shade is gone. The other one is nestled into some mountains, it's once great magic now used to keep the desert at bay. Post war, they are eager to put the past behind them, and the other nations are mixed in their response to that.
Sembia, Netherl, Cormyr, Myth Drannor, The Dales, and some other groups in that area have formed an organization of Knights, which include a handful of orders, ranging from Paladins, to corsairs, to ranger-types, and swordmage types, attached to a school on the southern Dragon Coast. Both organizations are international, and partner nations send the children of their most powerful families there, as well as recruited commoners and petitioners who prove themselves.
Recently, (ie, the year my current campaign is starting) students from many new nations are enrolling, and the school is being expanded. its built into a canyon, where it opens into the ocean, with earbtmotes having been moved to hover between the cliffs.
A new knighthood is also being formed, that will bond with and ride on the backs of giant birds
I'd say the one above is #2.
I like to keep up with the Forgotten Realms and its lore, but I don't mind changing things around. As in art, you should know the rules before breaking them. If Luskan becomes a halfling village in my campaign, I want it to be a creative decision and not pure ignorance.
Hard to say due to brevity, but this one above seems to lean toward #3. See my own example at the beginning for another #3.
Importance of canon and what I consider canon depend upon the setting in question. Canon is also only important for establishing the base setting and lore (pantheons, races, cultures, etc.). It is to be diverged from once play begins.
Greyhawk: It is the original folio, original boxed set, the deities Gygax presented in Dragon Magazine (so no Suel Pantheon), and bits of information on various cultures that appeared in Dragon (e.g., the culture information that appeared in the Barbarian class).
Forgotten Realm: Canon for me is the original grey box set, the 1e FR series of supplements, some early 1e articles by Ed Greenwood, and a 2e supplements from the 2e FOR supplement line. Nothing from the Time of Troubles onward
Al Qadim: Arabian Adventures book, Land of Fate boxed set and anything by Jeff Grubb
Darksun: Only the original boxed set and 1e articles by Timothy Brown
Ravenloft: Only the Realms of Terror boxed set and a few of its supplements.
Furthermore, any additional canon added by Planescape, Spelljammer, novels or late 2e and beyond is automatically ignored by me.
The above is a perfect example of #4, I believe.
I care deeply about canon right up to the start of a campaign. The point of using a setting, rather than making it up on my own as I go along, is to provide answers to questions about the world the characters live in at the start of the campaign.
But once the PCs start actually doing stuff, my concerns firmly fix themselves to that stuff and how it changes the world around them, and if it isn't compatible with some bit of canon - especially a bit of canon that comes along in a published product released sometime during the campaign - then it is the PCs actions and their consequences that stay true.
Also, there is one special circumstance in which I care not the slightest for canon and toss it entirely to the wayside: if a player knows it better than I do and tries to dictate to me along the lines of "Um... didn't you mean [this]?" while we are at the table trying to play the game.
The one above seems to me to be a #5.
In addition, I would add one more category:
6) The Strawman Canonist: You follow all of the assumptions and elements of the published materials, making sure to acquire as many resources as possible to avoid missing anything. Once a campaign has started, you do your darnedest to make sure nothing that happens contradicts current or future products that are/will be coming out. (I don't believe anyone actually does this, but I get the feeling that sometimes it is used (I would say misused) as an example of following canon for contrast with going more free-form. I would say #4 and #5 are better actual examples of following canon.)
