Dread, EPOCH and the GMIS
We tend to think about the SIS in two different ways - either as a Sandbox, where the PCs are more-or-less free to roam, and the LockBox, where the locations are constraint. Similarly, we think about the events inside the SIS as being either sandbox (unhelpfully conflating the space the characters experience and the activities they undertake) or a railroad. I want to focus for a moment on how players interact with the story mechanics - the rules of story formation, to explain how Dread and EPOCH work differently from most games, and so how it is possible to run these games incorrectly.
In almost all gaming environments, whether sandbox, lockbox or railway, the basic operation is for the scene to be set, and then for the GM to ask: What Are You Doing? Thus it is the player who provides the impetus for action, even if that action is constrained by the lockbox or railway. Thus the stories are inherently centred around player characters. If the GM doesn't allow the players to inject story energy, you tend to get a situation where the PCs become observers of NPC stories - god knows we've all been there and it sucked! Most indie games are in some way or another focusing around preventing that basic scenario on every level, allowing the players access to directly alter the SIS through mechanics as well as empowering them to make decisions for their characters inside the SIS.
The Dread environment is structured completely differently. The game begins with player questionaires - these are a lengthy set of questions intended to provide the players both with instant buy-in to their characters and useful information about the game world. The answers to the questionaires will often determine critical elements of the story infrastructure, without which the game simply cannot function. Players who arrive at the table thinking purely about a character are missing the point, and missing an opportunity. In essence, while there are a number of basic premises built into the scenario, the meaningful contents of the SIS are player-generated.
In the three examples scenarios in the core book, this is often obscured by the specificity of the premis. Under a Metal Sky's premis is "you are the characters in Alien". The impersonality of the foe they face thus means this game functions with a crippled SIS, and the tower, rather than the players, does most of the work. Beneath the Mask begins to allocate the real story responsibility where it belongs with its leading questions about the psychology of the characters. The ideal Dread scenario is probably one in which the GM provides only the thinnest skeleton of the structure while the questionnaires provide the bulk of the setting information, at least in terms specifically related to the story.
These questions seem like they should have the same effect on the opening phases of a story as a "kicker" does in Forge terminology. They provide an immediate impetus to character action, and hence build-in a player responsibility for moving the story forward. In a conventional game, this runs in parrallel to a GM-authored story, and the two interact. For a conventional game, that still retains the basic operational sense of the GM saying to the player "what now" and then adapting and inserting in response. In Dread however, the GM then takes over the direction of the story and explicitly begins framing all action as a block in response to this initial momentum.
Hopefully a quick example can make this clear. In a conventional game, the GM describes you feeling hungry in a room with an Orc and a pie and says "now what?" The player decides whether to eat the pie or the Orc. In Dread, the GM describes the same scene and then says "the pie eats the Orc, unless you pull." Player agency is automatically withdrawn, and you fight to get it back through the Tower. If you try and run Dread in the permissive "what now" style, then you are constantly going to have players try and do nothing in order to avoid a pull, so you need to push the Tower and you need to play the villain.
What this means is that the GMSIS is now explicitly intruding into what would usually be thought of as the player's part of the space: their character. The GM pro-actively imagines the characters getting doomed, unless the player is willing to interfere. This is a powerful technique, and because players are usually unaware of the actual mechanism for their disempowerment, they become afraid of the Tower instead of recognising that the Tower is actually their weapon against the GM, repelling the GM's intrusion into the player SIS.
The emotions generated by pulls are therefore more complex than described in the rulebook itself. Yes, the Tower is potentially deadly, but it is also, crucially, your most potent tool for story power in a scenario and game system designed to make you powerless. Naturally the greatest expression of this is deciding to knock down the Tower in order to achieve your objective, albeit at the price of death.
This is why Dread actually works - and that leads me to the problems with it. Players are often used, in con environments, to being essentially passive - cutting directly against the basic model of play I discussed above. To accomodate this, the best one-shot games both allow a player to be vegetative, and incredibly pro-active. In Dread, the passivity renders the GM's basic strategy obvious, because the GM ends up having to force the character into a situation where they need to pull, rather than the player levering themselves into a position where a pull gets them what they want. Passivity cripples this game, although even then, the fear of the Tower can actually mean that on a simplistic measure of apparent stress, the game still works, so it can limp to a conclusion. Just as deadly, is if the GM over-plays their intrusion, making the action feel too heavy-handed, making the players feel like their punching bag. Players who feel like the GM is trying to "get" their characters can always resort to refusing to pull - opting out of the game, making the GM powerless.
So let's roll EPOCH into this discussion, because it's basic mechanic uses a similar tool to Dread's. In EPOCH, the characters have a limited resource that they expend - like pulls of the Tower. Also like Dread, the results of any given mechanical encounter are basically that the PCs succeed if they live, but the GM has the option to put a stinger into the situation, and actually, EPOCH may be a little stronger on that than Dread.
Unlike Dread though, there is no way to opt out of the process, so passivity is not an option. Even more than that, the resource that you expend can be gotten back - by being "interesting" to the rest of the group, and so proactivity is rewarded by life, and passivity by death. Thus it restores to the player their familiar and natural role as the people driving the action, but with the same kind of tension-driver that Dread uses so well. It therefore feels a lot more natural both as a GM and a player, while delivering the tension that you need for that physical response.
Where Dread uses the GMSIS as a hard boundary that is used to direct the characters, EPOCH achieves the same ends with a more usual GMSIS structure.