312 Gaming on the Fly

Dayminkaynin posted this on our forums

…and one on how to run a game on the fly? You sit down as a player and the group makes you GM. You have to use player queues to run the game.

We record tonight at 8pm cdt on Twitch, coming to podcatchers later in the week!


Any tips that you want to share about how you may run an impromptu game?

1 Like

9 times out of 10, this is how I prefer to run anything. Even if I wrote and published the adventure, which is some strange Freudian shit right there. Right now, I’m running a play-by-post game that’s become a mystery/investigation adventure and I have no clue who did it, how it’s going to end, or anything else beyond the current turn. In many ways, I find this a more liberating way to run a mystery because there’s no attempt to force players to comprehend the clues you’re laying down. It’s all about picking up what the players put down and making up pieces based on what they give you.

It all goes back to that trust issue from the previous episode. Gaming on the fly requires everyone to trust the GM has their best interests at heart and will make our main characters to central focus from which the story develops. It’s the reason why I turn to gaming when small town Canada offer absolutely zero improv troupes. Fucking snow-covered hillbillies and their damn tractor pulls!


By nature, I’m an over-prepper. Sometimes I’ll take an existing scenario and heavily modify it to fit in my campaign. To me, it feels deep with a lot of detail I might not have thought to add. But then I find myself having to pause to reference my notes or reference material and that can break the flow of the game…or the group will go in an entirely different direction. But sometimes I find myself not really prepared for the game and I ad-lib the session. Almost without exception, these are the more popular sessions. Plus, I don’t have to pause or look anything up…just jot down notes as I go for future reference.

Trying to learn from this, I now do minimal prep…mainly to keep in mind what the various factions not related to the group will be doing. Then I run mostly off-the-cuff. I admit I find it easier and more enjoyable and my group seems to enjoy it as well.


I currently run for two games: old school D&D (Swords & Wizardry) and Conan 2d20.

For S&W I do a lot of prep: multi-level dungeons, city and town descriptions (sometimes maps), some names of prominent NPCs, brief descriptions of what might be in an overland 5-mile band, some notes concerning what various factions are up to. But chances are you’re still going to be Gaming on the Fly. PCs notoriously do not go into the dungeon (but it’s there, breathing and attentive for whenever the PCs finally want an adventure). They might decide to travel to a town for which you have nothing prepared. They might do absolutely nothing but make plans and have meetings with the local NPCs (my players tend to do this last one a lot. It’s their way of avoiding putting their characters into danger).

My point here is that something is prepared, even if it doesn’t necessarily see use during a particular session. For old school D&D, I disrespect the “Quantum Ogre.” For me, narrative choices matter: there absolutely should be a difference between going down Passage A vs Passage B. Something else to note is that this mode of play is very dependent on player activity. This is why my favored *master term for this mode of play is Referee. The focus of play is on player agency. If an “adventure” isn’t happening, well… “You did all see the broadside advertising for the Tomb of Horrors, right? Well, are you going to go there?”

For Conan 2d20, in contrast, I tend to select a pre-1950s Weird Tale for inspiration, use that for the story situation, stat out some NPCs, then throw the PCs into it. That kind of game is also reliant on PC activity, but my GMing is much more reactive to emerging narrative. Most notably, I don’t bother mapping what I may presume are going to be key locations. Conan 2d20 is almost a super-crunchy Story Game. “Dungeons” can organically assume the proportions of narrative need, not simulationist exploration. In short, the Quantum Ogre is welcome here (in exchange for my GM resource of Doom).

Now, were I to run S&W “on the fly,” with absolutely no prep… That’s why I have a binder of old standards, some mini-dungeons. I’m late to the old school scene, so my standards aren’t the usual suspects, I’m sure. Three are Creation’s Edge’s “The Cursed Fountain,” Jeffrey Talanian’s “Rats in the Walls,” and Matt Finch’s “Tomb of the Iron God.”

For Conan 2d20 I’d rely on memory. A really good scenario has been based on Clark Ashton Smith’s “The Charnel God.” But I can see spontaneously creating from whole cloth. Conan characters are built bristling with plot hooks. It would be easy enough to look over the characters, drop them into something, then observe what happens.

I expect different games require different degrees of “preparedness.” For my two favorites, whether maps are important or not seems to be the chief consideration. I expect it would be fun sometime, as an experiment, to burble something out of an online random dungeon generator and just run it!


I don’t think that the “quantum ogre” is a useful tool to talk about player agency. The entire conundrum makes a lot of assumptions that drastically change what the premise means. In a vacuum, if the only choice is going down path A or B and there is an ogre on one path and none on the other one, it means very little for player agency beyond side stepping an encounter unwittingly. If the game master knows it’s there but the players otherwise are unaware of the ogre, they won’t know that they’ve made a meaningful choice by taking Path B and side-stepping the encounter. Even if the players do know of its existence, is there really that much drama in such an easily avoidable obstacle?

Running a game on the fly means no unnecessary preparation. Creating a branching path of the plot which requires you to prepare two encounters, one of which the players are guaranteed to never see is counterintuitive. I find that personally, the best way to encourage player agency is to present the players with a problem, and do not have a single solution in mind when you create it. In this way, the players create their own narrative branches in the ways that they resolve encounters.


I can see how you can say how this has little to do with player agency, but my point here was how to “GM on the Fly” while maintaining expectations regarding two different modes of play, two modes in which I happen, currently, to engage in on a weekly basis. In this mode of play, a possible player challenge naturally is to learn (or not) about potential adversaries and to avoid them, if they choose to do so. This mode is not a tightly constructed narrative that is anticipative of PC actions.

I guess knowing about an Ogre and consequently avoiding it might be construed, by some, as lacking drama, but that’s not my purpose in old school, sandbox play. My purpose is in what is there and how the players choose to engage with it, be their choices dramatic or not. Besides, I think that learning about an Ogre and avoiding it (likely by some similarly perilous passage) can be dramatic. Perhaps it’s a point of view thing.

Anyway, my point is that GMing on the Fly for this mode still needs to adhere to expectations, and some of those expectations necessitate a greater measure of “preparedness” or preparation.


great episode. I love gaming on the fly it’s my preferred method. I find it makes my games much more interesting and dynamic. I tend to make up some crazy shit and honestly my players have always said that those game are their most fun. Even if I run a published module it’s usually off the rails fairly early on and by the end it rarely resembles the end product b/c I’ve improved so much. I miss it a lot since Covid and my games have went to VTT, Fantasy Grounds is our groups preferred VTT. I find it much harder to do inside a VTT though, since my players have gotten spoiled by the battlemaps, tokens, the dynamic lighting and line of sight, even the face to face games in the before time they’ve gotten used to having the minis and maps, granted the maps are usually tiles I’m just improving and tossing down or a just a giant sheet of graph paper that the players are mapping on. It’s been really hard to get them into a theatre of the mind, mindset so i can GM on the fly again.

As an aside if anyone is aware of of VTT that allows you to map on the fly, ideally with the players being able to do the mapping please let me know. I’ve mentioned that we are big Fantasy Grounds users we started off with Roll20 and lately I’ve been looking at Foundry and Owlbear Rodeo, but I don’t know enough about the later two yet to know if those are going to be viable.


For my COVID-online Swords & Wizardry game (which I mention a bit above), I use Mipui. When my PCs go into a dungeon, I start mapping 10’ squares. My players won’t map, but yours can if you give them permission to edit. My own players have permission to edit so they can place tokens, etc., for this “aerial” game mode. When things get tactical we usually switch to Roll20. I don’t use any pregenerated maps, etc., outside of what I have in my notes. Mipui shows PCs the larger context, and, during tactical moments in Roll20, I simply slap down some shapes, tokens, and define spaces.


1 Like

This is my answer to Brett’s question to my sense of setting for Gaming on the Fly.

In short, the answer is yes, I know these settings intimately. I never considered this sort of knowledge as an important factor for effective spontaneous gaming. For Swords & Wizardry and the introductory adventures I could pick up and run, setting mastery doesn’t seem important: the world of my homebrew D&D always grows, if and when necessary, in collaboration with my players. But the Conan Hyboria setting I know very well. And, like Brett, I also could run in Middle-earth. And in the Star Wars universe (first trilogy only as canon).

Interesting observation. Thanks!

1 Like