Linearity in scenario design is meant to be one of the bigger no-nos. I can see why. Player participation is key, and if the players don't feel that they have a role to play then any interest they may have had in the adventure quickly dies. This is a problem that affects not just tabletop but also video game design; this week's Zero Punctuation takes Medal of Honor; Warfighter to task for, among other things, carefully scripting the game so that player input is almost nil.
Scenario playtest feedback for my work often contains the complaint, "feels too linear". I particularly liked one report for a short piece due to appear in print soon, that started with the feels-too-linear statement only to say, once the players had become involved, that it didn't shake out in a linear fashion at all. In fact, the players floundered around a bit before finally working their way to the McGuffin; according to the report some of them liked the sensation of not being in control, while others said they could have used more pointers.
I'm not going to downplay linearity. It's definitely a script problem, and it gets all the worse in an environment like a video game, where the player literally cannot escape the boundaries of the plot, whether because invisible walls are blocking off all escape routes or for some other in-game reason. However I don't think it's as much of a issue as people seem to think, particularly in tabletop, and this is why:
Players are kittens.
I mean that. Anyone who's ever actually dealt with a kitten knows that they can be psychotic one minute, biting the hand that feeds it, and docile the next. Kittens wander off into any kind of disaster, wide-eyed and innocent, because they don't see any boundaries even if boundaries are there. They'll cheerfully walk off a table edge, given half a chance; but there's just as much likelihood that they'll find someplace they like the look of and put their heads down for a nap. Kittens do as they please ... and so do players.
The Keeper, knowing this, has to structure the scenario environment to withstand sudden flights of player fancy. The basic premise has to be strong, and supported by useful scenes that support the basic premise. That's the only way to prevent accidents. You wouldn't leave a kitten unattended on a table; why leave players without a safety net?
Say for the sake of discussion the scenario revolves around a murder, a good old-fashioned locked-room body-in-the-library situation. Once the body is discovered, there are certain things it is reasonable for the players, as the detective heroes of the hour, to do. They will want to inspect the body, the room, and the lock. They will want to talk to anyone who might have known the victim, or anyone else who happened to be in the house at the time. They will want to know how the victim died, and if it was something relatively unusual, like a poisoning, they're going to want to find out how easy it is to get hold of the poison and where it might have come from. Each of these issues can be a scene in itself, and each scene can contain clues that lead to the next logical scene. So if the players talk to the victim's lover, and get nothing, but talk to the maid and discover that the victim and the lover quarreled, they're going to want to go back to the lover again. None of this is brain surgery; anyone who's ever seen or read a murder mystery, ever, knows - at least in broad terms - how a murder investigation works.
Now, the Keeper could easily come close to linearity in plotting this situation out. Discovering the body leads to finding out how the murder was done, and from there to who might have done it, and from there to who did, in fact, do the deed. It seems so straightforward, when written out like that. There are variations on the theme, of course; many murder mysteries focus almost exclusively on motive, with an eye to a dramatic confession in the final chapter, while other investigations worry less about that and more about the mechanics of the crime. "That's How," Lord Peter Wimsey might say, "and now we know How, we also know Who."
However linear a scenario of this type might read on paper, it probably isn't going to be linear in practice. That's because you can't rely on the players to be perfect detectives. They're kittens, remember; they'll poke their heads where they please, and hope for the best. Some of them may not try to work out how the victim died, or content themselves with superficial seemings. Some of them may even invent wholly implausible theories and wed themselves to those theories beyond any hope of amicable divorce. The least likely NPC, having wandered into view, suddenly becomes Number One Suspect without even the smallest shred of evidence to convict them. Some will spend all their time chasing up exotic poisons without once confirming the actual cause of death. Some will convince themselves that, against all appearances, the person whose head was bashed in with a shovel actually was eaten by a lion - and this despite there being no lions in the scenario whatsoever.
The only thing a Keeper can do is write a fairly solid plot, and cover all the most reasonable bases. Anything beyond that is up to luck. Linearity in tabletop is, all too often, an illusion, and the Keeper believes in illusions at the Keeper's peril.
There is one other point to make here, and that is that even if a plot is completely on-rails, it doesn't have to be a problem. In fact, it might be the most enjoyable part of the experience. As evidence, I direct you to Telltale Games' survival horror puzzler, The Walking Dead.
If you haven't already played this, I highly recommend it, particularly if you enjoy zombie apocalypse settings. The game is based on the comic - not the TV series - of the same name. You play as Lee Everett, a man on the way to the penitentiary who manages somehow to struggle out of a car wreck into a world gone mad. Along the way he picks up a fellow survivor, an eight-year-old named Clementine, and together they do their best to pick their way through a very hostile Georgia.
It's grabbed a lot of people's attention this year, particularly for its excellent writing, but if there's one criticism that stands out above all the rest, it's aggressive railroading. You, as the player, have very little choice in what happens next. Scene follows scene like soldiers on parade, and if there is a puzzle in the way before you can get to the next scene, it usually isn't that much of an obstacle. Before you know it you're back on the train again, chugging away towards the next plot point.
Yet I, as player, don't have a problem with that; I've been enthralled by The Walking Dead since I learned I could pick it up on iPad, and just finished episode 4 this weekend. I'm on the edge of my seat waiting for 5; at least two major plot points reaaaaaally need resolving, and I won't be happy until they are. I suspect most fans of the game feel the same way.
It wouldn't have nearly the same kind of impact if it wasn't masterfully written, and that's the key issue here. Even if you are marching from scene to scene without deviation, it's your interaction with the world and the characters in it that makes the experience so enjoyable. What happens next to Ken, to Carly, to Lilly or Doug, how Clementine reacts, how you react to her, whether your morals are strong - or flexible - enough to withstand what's coming next, is what makes this game a stand-out.
The straight line between two points isn't an issue, and while linearity can be an easy plot criticism I'm not sure it's entirely valid. To my mind, using "linear" as a scenario descriptor is actually a cover for something else, and that something else is "badly written." We know bad writing when we see it; flat characters, unbelievable plot points, disjointed or facile settings, are all issues that fiction has had to deal with over the years.
But those issues are the problem, not the linearity - or otherwise - of the plot. A straight-line plot with excellent writing can work brilliantly; a badly written sandbox will often fail and, where it does succeed, anything good it manages to pull off is a matter of luck, not design.