I can think of three major “plot” structures for an individual game off the top of my head. I put plot in quotations because I think a properly run game shouldn't have plot so much as theme, but then again, I'm a big fan of sandboxes.
The three structures are:
Dungeon Crawl- the core components are combat, exploring and retrieving loot
Investigation- exploration comes in the form of information and social networks and the”loot” takes the form of clues integral to the investigation
The third is a blend of the two styles, the Heist.
Heists usually deemphasize combat and emphasize loot. Like in an Investigation, exploration takes place outside of the “dungeon”. To put it another way, while exploration in a dungeon crawl takes place in the dungeon, with the players/characters slowly learning the lay of the land as they travel through it, a heist typically involves the players/characters getting access to a map of the location before venturing inside. The primary challenge is obtaining information about the location and then using it to plan a route or method of metaphorical attack.
The three styles exist on a spectrum of method to goal oriented, with dungeon crawls being primarily method oriented, investigations blending the two, and heists taking the position on the far right of the spectrum as mainly goal oriented. As an aside, assassinations are structured similarly to a heist, but the end goal is someones death.
What does all this junk mean for a game?
Handouts are important. Especially maps. This gives players information they can all see and reference while planning. As part of this, an easily accessible reference to character abilities is very useful.
Improvisation is back loaded on the part of the GM. The guts of the heist shouldn't be thought up on the spot, as it may make it difficult for the players to generate a plan. Complications and antagonist reactions to the plan may be improvised.
Speaking of complications, in other pieces of fiction, the more the audience knows about the plan of a heist, the more things are going to go wrong. To make this a gameable concept, the more the GM knows about the plan, the more complications are going to take place. This could take the form of a token economy, players buy information and the GM uses those tokens to buy complications, possibly in the form of a random roll table.
Heists usually have a complication. Everything can't go smoothly for the players, or there is not challenge during the heist itself. To offset this, consider the concept of the Tilt from Fiasco. While Fiasco games don't usually follow the narrative structure of the heist, midway through the game, it introduces a complication called the Tilt. This is something that helps turn everything on its head. These are short phrases like “something valuable catches fire” or “magnificent self destruction” that guide the players into setting up a situation where everything goes wrong.
Another concept that Fiasco embraces is the flashback. It doesn't come up much in the rules of other games, but it is a perfect way for a character to say “I saw this coming” while keeping the player in suspense.
The GUMSHOE system, with it's concept of the preparedness skill could easily translate into a set-up for flashbacks.
Action scenes aren't as important in the heist model. Think of the Thief video games, combat is dangerous and loud. If you have to fight, you want the advantage over your opponent. So while you could run a heist game using something combat focused like D&D, games focused on investigation might be a better choice as they hopefully inject more detail into the use of skills.
Now that we know what a heist is, what is a heistbox?
Well, like a sandbox game, it is player focused. It is similar to the ocean-centric saltbox in that it is a sub-type of the sandbox structure. A heistbox is a game structure focused around player devised/selected heists and/or games with the heist structure.
In my regular game right now, I'm running a sandbox in the hexcrawl style, where wilderness is traversed much in the same way as a dungeon crawl. I chose this style because it allowed players to drop in and out without disruption to the overall flow of the game. A major facet of the game is that the players wanted to create an organization, in this case, a guild of adventurers.
A heistbox is the perfect setup for this. There are multiple examples in fiction and real life of criminal organizations: the crew from a heist movie like Ocean's 11, organized crime families, gangs, thieves guilds, and so on.
The framework of a criminal gang also makes it easy for players to slip in and out of the game. The idea of a gang comes with the implication that there is a headquarters, or hideout. My previous sandbox game divided the headquarters role between the town and a central inn/tavern. Town was a safe area, and characters couldn't come to harm during sessions where they weren't present. During play, the characters always started, and planned their activities for the session from this central tavern, it also was the location where they gained information about local events and interacted with some recurring NPCs. My thought for the Heistbox was to combine the two. Since the action of the other sandbox took place in the area surrounding the city, it made sense that the city would be a safe zone. Conversely, the action of the Heistbox takes place entirely in a single city so they cannot be safe there. Additionally, when I spoke to my players about the idea of the Heistbox, one of the notes I made was that they were very invested in the idea of a hideout, and one of them even mentioned the potential for raids. So safety is not an option for anyone present at a session. As for the other players, as thieves and rogues of all kinds, I would imagine personal safehouses and boltholes would be a common investment, making their safety a moot point.
In my previous Sandbox, I made use of the faction turn from the Sine Nomine game line. In between sessions, I used rules to simulate the power struggles of the various factions of the region in order to generate content. While interested, it was time consuming. For the Heistbox, I want to create a set of tools to easily generate Heists, especially at the player's direction. An example of this would be, one player saying “I want to pull a bank heist” or “I want to rob a mafia poker game”. While I could try to predict the wants of my players, or have them wait a session before embarking on the heist of their design, it would be best to have the tools available to generate these situations on the fly. The second system paralleling the faction turn would be something to simulate the power struggles of the local factions. My current thoughts are to create two 'currencies' Reputation and Heat. Law enforcement factions distribute their Heat among the various criminal factions, making jobs harder for them. Criminal factions gain Reputation at the expense of the other factions (representing the transfer of power and standing in the criminal underworld) for their jobs. Jobs would be simple, one chosen per 'faction turn' and from a specific list of general jobs, like burglary, heists, protection rackets, or smuggling. Law enforcement would then distribute Heat and the factions would make a single roll. The players would also interact with this subsystem as a group, build/losing Heat and Reputation based on their performance which would apply to the other factions before they took their turns. These are just my current thoughts and it's likely to change as time goes on.
As I develop the setting and rules for the Heistbox, I'll try to post them here so you get a glimpse of what I'm building in real time, and my process for designing them.
Writing Excuses Podcast: http://www.writingexcuses.com/2012/06/17/writing-excuses-7-25-writing-capers/
Transcription of linked episoder: http://wetranscripts.livejournal.com/60653.html
Makers of the GUMSHOE system: http://pelgranepress.com/
Sine Nomine: http://www.sinenomine-pub.com/