"Let's see what this thing does."
The object in question is something simple and unassuming--a spinning wheel or a harmonica--and the day is bright and sunny.
If a sandbox D&Der says "let's see what this thing does," the phrase might not mean anything to whoever's listening, and even if using the spinning wheel does lead to a 3/4 party kill and the remaining players reversing alignment and/or gender, those words--"Let's see what that this thing does"--will only take on significance retroactively.
If, on the other hand, a Call of Cthulhu player says "Let's see what this thing does" you can hear the mood music and the crackling of thunder already--sunny day or no. The very set-up of the game is a kind of foreshadowing.
This isn't just because Call of Cthulhu is a horror game and therefore has the atmosphere of horror. It's also because Call of Cthulhu usually takes the dramatic form of a horror movie or thriller. All paths sooner or later lead to doom. The players know that and they know they are inviting trouble by playing that harmonica.
This gives all CoC dialogue, even the most banal--especially the most banal--a second significance.
"Your uncle seems to've had a terribly vivid imagination, Chauncey!"
"Leave me alone, I want to watch the birds eat."
In this way, Call of Cthulhu not only (like all role-playing games) allows acting, it automatically deepens it--even without trying. The more you can "act" (in this case, largely a matter of pretending to not know whatever you do know about the Lovecraft mythos and the fact that this game tends to end in tragedy and madness) the more fun this double-meaning is. "I don't trust this cheese, Lady Crustingham". In effect, the Call of Cthulhu player is both a participant and a knowing observer.
This isn't true in most sandboxes because what loads your Cthulhu PC's words with meaning is the intimation on the part of all players that certain future events will occur. Of course, the whole point of sandbox D&D is you have very little intimation of of what sort of future events will occur.
Duh. They're different games, with different roads to fun. Anyway...
The Rewards of Acting When You Already Know The Plot
Consideration: In effect, if certain elements of your characters' future fate are sketched out, and everyone at the table knows it, then there is an automatic extra reward for role-playing in the sense of "acting"--you get a "double effect" every time you role-play. Not only does everyone get whatever aesthetic pleasure there is from hearing a character say "Ho ho! They'll only let us go to the party if we stay for drinks and a lecture? I guess we'll just have to live dangerously, Pubblemayer!" you also get the aesthetic reward of knowing that that party is probably a cult meeting and the drinks are probably drugged and so that remark is a little funnier.
I could draw the obvious parallel to another genre: porn. Improvising porn dialogue (which--shocking I know--is often how we do it) is fun because, likewise, whatever you say is automatically given a second significance by what everyone knows is about to occur--"Where'd you put the melons, Cecilia?"
In a sandbox game, this isn't automatically so. If you say something your character "would" say, the players get the pleasure of acting in character and thereby contributing to creating the characters, and that--depending on how future events play out--might be all you get. You could step on a rusty caltrop and die for no reason 3 seconds later. If you act bold it might pay off, it might not, if you act sneaky it might pay off, it might not. No-one knows precisely what genre they're in (except, broadly, "adventure") and that's part of the fun and the challenge. It's a different kind of fun.
Point being it is the intimation that future events of a certain character will definitely occur that creates these kinds of double entendres in role-playing.
The intimation that future events of a certain character will definitely occur is also one of the elements that can turn a game into a railroad.
Now DIY D&Ders are often to be found howling (rightly so) against railroads. This is why we like sandboxes--it's almost impossible to railroad someone if they have total freedom of action. Good, problem solved. We just do location-based adventures and we can be sure we're never, ever railroading.
I think, however, it's important to realize that there's quite a lot of territory between the total sandbox and the total railroad and some of that territory allows for unique aesthetic effects (such as the double-meaning) unavailable at the other extremes and I would like to try to help create a way to talk about plot construction in a more detailed way than just having people argue about whether a given published adventure or type of campaign "is a railroad" or not. So here it is...
It'll be helpful to nail down the kinds of techniques that can lead to railroading. Basically, railroading usually proceeds from excessive and continuous use of certain GMing techniques which are, in themselves, harmless.
These techniques are techniques designed to limit the PCs options in a given situation, and I'm going to call them "chokers". Chokers can be good, for example: An adventure including a trick room where the PCs can't talk is not a railroad. The problem comes when chokers are used to elicit specific courses of action from the PCs for a decent chunk of a session or for whatever the group thinks of as too long to be fun.
Here are the chokers--if you think of any others, let me know, but make sure your choker doesn't fit into one of these categories--(for example the Interfering NPC is not so much a choker as a device capable of providing other chokers for the GM at will.)
Limiting Resources: Basically, this means creating a situation in the game where the PCs can't get their hands on things or people or abilities that they normally would in order to solve a problem. Nearly all RPG scenarios involve limiting resources in one way or another. A problem (like a riddle, for instance) with only one or two possible solutions is a kind of limited resources. Limiting resources or solutions so much that only GM-approved solutions to a problem are possible ("...the only way to break the curse is to..." etc.) is not, alone, railroading, but it can lead to it if combined with other restrictions.
Limiting Information: The PCs are only given enough information to make one course of action plausible. The familiar example is an investigation-type plot. If there's only one clue and it suggests only one possible course of action, this can turn into railroading. Again, like limiting resources, limiting information is a part of nearly all RPGs.
Unanticipatable Events: The PCs are affected by an event they could not have foreseen. Maybe they have to then deal with it, maybe it just affects them in some way indefinitely. Note that if the PCs could reasonably have foreseen and prepared for or avoided the event but didn't it's not a choker. A trap that might be detected isn't a choker. A trap that the PCs have no option but to fall into is.
Anticipated Uncontrollable Event: "It all goes down on the 16th!" What goes down? If the PCs don't know and can't know, or know but can't do anything about it, yet still know it's important, then it's an Anticipated Uncontrollable Event. They know something important will occur but have no way of addressing it. This can create a sense of (obviously) anticipation and it can generate excitement or mystery, but it can also make whatever the PCs are themselves doing seem irrelevant since they have no idea how or whether it'll have anything to do with this obviously important other thing the GM is going to all this trouble to let them know about.
Mind Control: Because of magic, science, insanity, brain damage or the like, the player loses control of his/her actions temporarily and the GM gets to tell them how they behave. Note that in some cases there's a fine line here: a player who is suddenly informed s/he is a kleptomaniac or a servant of an evil necromancer may actually welcome the opportunity to change personalities for a while and not see it as a restraint at all. In cases like that, the question of whether the PC enjoys and finds things to do in the new role or basically just has to grudgingly do things s/he would rather not can affect how much of this choker someone can take before it's perceived as railroading.
Unique Reward: There's something cool which the PCs can only get through a given GM-presented course of action. Presenting PCs in a game of AD&D with the prospect of lots of x.p. isn't using this choker, since there are other sources of xp, presenting the PCs with the prospect of attaining some specific artifact or relic is. Thanks to Menace 3 Society for bringing up this choker in the comments.
The next 4 chokers are worse than the above chokers--that is, they're less often justifiable and will lead to players feeling railroaded more often...
Nullification: The PCs take an action that they know should, by rights, lead to another event further down the line in the scenario working out differently. The GM, in order to preserve his or her idea of the scenario, thinks up--after the fact--a reason that the event still happens pretty much the same way despite the PCs' action.
Example--The PCs are going to fight Superman, they get some kryptonite, it turns out Superman is wearing a radiation suit that day.
In order for this to be nullification, the PCs have to know or suspect that the future event is coming. If they don't it's just, essentially, an unanticipatable event (i.e. the PCs run away from the monster that the DM planted a secret bomb on, so the DM puts the bomb on some other monster in a different adventure somewhere else). Now unanticipatable events are chokers, too, but this difference matters because nullification is the worst kind of choker, and is rarely, if ever, justifiable.
It's important also to note that, unlike limiting resources or information, nullification occurs after the fact of the PC's action. If the GM can hold up her notes and go "See, right here: Superman--in his radiation suit--is in room 6b, fixing the Orgulator from 6am-7:30pm." then that's just limited resources and limited info--which are chokers, too, but not quite as bad as nullification. Why not as bad? Because the PCs will at least be able to see that the reason that the GM stymied their plan was because she out-thought them ahead of time. Which feels more fair.
It'd feel yet more fair if the PCs had a way they could've known that--like if they had scouted ahead or checked Superman's calendar, which was hanging on the wall in the room they just went through--"April 12th--fix Orgulator--Get new rad suit first!".
Nudging: The PCs "should" take some action, notice some clue, make some choice, etc. They aren't doing it. Someone or something comes along and points them in the "right" direction. Nudging is bad because if the idea is that the PCs need to do A in order for B to happen and they don't do A, then they've failed and there should be consequences--otherwise the players aren't really being challenged and there was no point in putting the clue, choice, etc in there in the first place. At least in-game. (Obviously you could rig it so the PCs don't get xp if they don't notice your clue on their own, but that still means their actions had no effect on the story.)
Another reason nudging is bad is it has the possibility of eliminating interesting alternate solutions before the PCs have time to think of them.
Pointing By Punishing: A kind of nudging. The scenario is constructed such that the PCs "should" take some action (urging specifically toward not a goal but toward a specific tactic), notice some clue, make some choice, etc. They aren't doing it. There are consequences such that leaving the vital thing ignored or left undone makes things harder for the PCs and tells them that they should've done the thing.
For example: if nobody realizes they have to stop Orbach The Despised, then everything metal in a 20 mile radius slowly begins to rust. Pointing By Punishing lets the PCs know what they should've done, but also makes the thing they're supposed to accomplish harder--that way the PCs get a second chance, but there's still consequences to their actions. It's a choker, but it's preferable to straight nudging.
Nursing: Similar to nudging, except it's usually about combat or some other character-skill-based (rather than choice-based) thing--that is, fighting or rolling or otherwise doing something involving chance and tactics. There's a challenge to the PCs. The PCs fail or are going to, so the GM invents (or prepares in advance) some plot device that helps them out so they can win and the plot can keep rolling along.
Nursing essentially makes it so that the PC's failing has no in-game consequences. Essentially, it's a form of nullification except that sometimes the PCs actually like it.
The theory is that even if the players want to be nursed, eventually they'll get bored and frustrated in the long run if there's no consequences to their failures. I don't know if that's necessarily always true (someone should take a poll), but I do know that the GM will get frustrated and bored if there's no consequences to the PC's failures, since the story becomes predictable.
So those are all the techniques I could think of that could lead to railroading.
Now I'm gonna take a look at the different kinds of adventure structures you can set up, and how railroading might appear in them. I'm gonna start with the kinds of campaigns that have the most freedom (the least chance of railroading) and move to the ones that have the least freedom (and, therefore, the most chance of railroading).
NONTRADITIONAL, SUPER-OPEN GAMES
It's probably impossible to get railroaded in these kinds of game except maybe in some whiny psychsocial way, they are...
Total Improv/ Brainstorming "Game"
A bunch of people sit around and, with no rules or formal agreement, dream up a story and take parts in it. (Note that many more structured games often begin this way, especially with really experienced players who are trying to think up something new to do.)
Common Law Game + Common Law Story
A bunch of people sit around and draw up rules for a game before or while playing it (one possibility is the way it's described here), then play the game following these rules, with everyone contributing to the story both in and out of character.
(The next two types of games are, dependng on the levers and dials, of varying, but relatively high--levels of player freedom)
Certain Story Games, Like 'Shock', For Instance
A bunch of people contribute plot and/or setting elements according to a given rule scheme, and play characters in the game. The players also have an ability to affect things in the game out-of-character once the games already started.
Common-Law Rules Within An Otherwise Traditional RPG Set-Up
This is a game with a game master and all that and a fixed setting, but the actual rules are made up by the group as they go along, like this.
Certain Other Story Games
A bunch of people contribute plot and/or setting elements according to a given rule scheme, and play characters in the game. However, the players don't have the ability to affect things in the game out-of-character once the game's already started. There's a GM.
Now we come to the more familiar types of games in D&D. Railroading can happen, in extreme situations, here.
This has, for the most part, a traditional RPG set-up. GM and players playing PCs. The PCs can go anywhere (and there are lots of meaningfully different places to go) and do anything their characters could according to common sense and the rules. The villains act mostly like traps--they don't make their own schemes at all, and will not proactively do anything creative or involved that might affect the PCs other than, basically, fight them if they show up. No natural disasters or other world-affecting events occur unless triggered by the PCs.
Any kind of dungeon that is mostly traps and/or animal-level monsters is like this. If there are intelligent NPCs but their behavior is somehow randomized rather than intelligent DM-inflected scheming, then you're still in Extreme Sandbox territory--point is, the DM 's ability to shape events even through the NPCs is limited.
The only way you could possibly get railroaded in a game like this is if there were extremely limited resources or limited information in the initial conditions, or if you slip down into some area of the sandbox that it takes a long time to get out of which is full of automatic nudges, nurses, resource limiters, etc.
PC prep is important here--if PCs are given the opportunity to prep before going somewhere, it's hard for them to be railroaded.
(The next two, depending on the dials and switches, are at approximately equal levels of player freedom.)
Sandbox With Triggerable NPCs or Plot Events
As the Extreme Sandbox except the PCs can "activate" an NPC by choosing to interact with him/her and thereby place that NPC in a position to affect the plot in the way an intelligent, thinking, proactive person would. The PCs can also trigger events which have the ability to alter the plot (or create a plot)--like, say, if they steal a certain gem then children all over the gameworld starting eating their parents.
There's a thin line here: In Raggi's Death Frost Doom, for instance , the players can inadvertently wake thousands of corpses from their slumber--if the DM treats these undead as the kind of monsters that'll get in the way for a bit and then disperse after a session or two, it's an Extreme Sandbox, if the DM keeps the skeletons around but randomizes the movements of the army of undead, then it's still an Extreme Sandbox. If the DM decides when and where the skeletons go for some time thereafter in order to spice up the game (as I myself have done) then we've moved at into a genuine Triggered Plot Event.
It's possible for a DM to use the thus-triggered NPC or event to then start forcing all kinds of actions onto the PCs, so, technically, the possibility of railroading begins here. Once an NPC is unleashed, all chokers become possible.
As the Extreme Simulation, except the world has natural disasters and intelligent NPCs that (theoretically even when off-screen) do things which the PCs may or may not, depending on their own actions and perhaps some probability tables, intersect with. To be a 'perfect' clockwork simulation, all the NPCs must be directed by the GM strictly on a "What would this person actually do" basis and never ever ever on a "What would be interesting for this person to do right now" basis. Whether a totally clockwork sandbox would be desirable or even possible is another question, but it's here as a sort of platonic marker on the freedom scale. It's certainly possible to have a fun game by going for it.
Basically, this would be a sandbox that attempts to act as much as possible like a real world.
Again. as soon as a sufficiently powerful (and intelligent) NPC takes interest in the PCs, all chokers become possible.
Sandbox With Interested Forces
As the Clockwork Simulation except some NPC, force of nature, or other entity is definitely and unavoidably going to act on one or more of the PCs or has been designed specifically so that their interests overlap or are in competition. However, this force does not necessarily demand events unfold in any given way.
For example: in episode 2 of I Hit It With My Axe the medusa and the city guards want the party to take on a mission, and threaten them if they don't, but the PCs can still remain alive and have adventures if they just fight the medusa and guards or just run away, or take the mission and then say screw it once they're outside the city gates and go do something else., .
In order to fit here, the PCs should be able to deal with these interested forces in many different ways (not all of which are obvious to the DM)--including violating the "rules" of the genre (e.g. the investigators saying "Who cares who murdered the monk?", deciding to let the tsunami crush the city and just move, deciding to train a giraffe to fetch the coin from the haunted well., etc.) and still have a satisfying game.
(This, incidentally, may be why sandbox D&D so often goes gonzo--the wacky solution is often the only one that the DM hasn't thought of. Letting the players do whatever works pretty much encourages gonzo. Which is awesome.)
SANDBOXES WITH IMPOSED PLOT ELEMENTS:
The danger of railroading begins here in earnest, however, so does the ability for the DM to create a plot or to create a game existing consistently in a specific mood/genre. You can't, for instance decide to do a noir detective story or a horror story or a thriller on purpose without going into this zone.
Sandbox With Unanticipatable Events
As any of the sandboxes above, except the DM proactively introduces distinct events into the PCs' lives. This "event" has to be more than just introducing a NPC, trap, or monster that the PCs can immediately dispatch if they play their cards right (or be dispatched by if they play their cards wrong), it has to be an encounter which leads to a predetermined end. For example, the DM decides that the encounter with the rust monster will destroy the paladin's +5 sword or the pickpocket will get the dwarf's healing potion and secretly replace it with a gender-switching potion no matter what else happens, and then arranges things so that occurs.
This can be ok--like an adventure where the PCs will be shrunk down to the size of peanuts before entering the lizard king's palace is not necessarily nothing but a boring railroad. However, I would argue that an adventure where the PCs ended up shrunk down to the size of peanuts before entering the lizard king's palace because they went and opened the wrong drawer is slightly more fun for everyone because that way the PCs know that what happened was the result of their own curiousity (or whatever) and the GM isn't sure whether s/he's running a shrunken adventure or not until it actually happens. However, that's a minor quibble--particularly if you just finished having a really fun shrunken adventure...
How railroady this feels depends on whether these events prevent the PCs from doing the kinds of things that they individually, are playing RPGs in order to do.
If the player loves acting and doing voices above everything else, then an endless stream of unavoidable unfortunate events might be just what the doctor ordered to give their player something to chew the scenery about. If the player enjoys plotting and scheming against the DM, then a bunch of out-of-nowhere events which keep interrupting her while she's trying to build her Invincible Squash Golem are a problem. If the player enjoys solving mysteries, then having every solution but one be a dead end might be frustrating, if the player involves solving puzzles on the other hand, having every solution but one be a dead end might be exactly the kind of challenge s/he wants.
Sandbox That Changes By Itself
As the Clockwork Simulation above, except the GM (or whoever wrote the scenario) knows things are going to happen which will fundamentally alter the setting in ways that will matter to the PCs. Basically a big unanticipated event that affects the whole sandbox. Since a Sandbox That Changes By Itself is not yet a full-blown Plotted Adventure, these changes do not necessarily require the PCs to do any specific thing, they just require them to do things in a different way from now on. That is, they change the realities on the ground. A cataclysmic earthquake that affects the whole gameworld is an example.
Sandbox With Miscellaneous GM Intervention
This is arguably the most common sort of sandbox--it's an open game where pretty much any of the things in the above five categories can happen and occasionally do. The GM can introduce interested NPCs, decide there'll be disasters, impose events by fiat, and generally muck about in the machinery.
Again, the danger here isn't any one thing, it's simply taking care that whatever sandcastles they PCs are building in the sandbox get as much (or more) chance to actually amount to something and change the shape of the game as ideas introduced after the game's started by the DM. Sandboxes depend on the PCs having information about their surroundings. Keeping the sandbox a sandbox means that whatever the GM does to the game or gameworld, this information must, for the most part, continually to actually matter and be usable by the PCs.
Plots Requiring Use of The Sandbox
The PCs don't get to do whatever they want, they have a problem to solve. The DM (or scenario) formulates this problem such that having an adventure at all requires trying to solve it. However, the problem is long-term and open enough that the PCs can pretty much pick and choose how exactly to go about solving it and they have any resources of the entire gameworld that they can get their hands on at their disposal.
A simple and relatively "closed" form would be the Epic Sandbox, a more open form would be the scavenger hunt sandbox, and a middle form would be the Preparation Adventure--an adventure where the PCs have a given amount of time before a certain climactic confrontation takes place and they have to do as much as they can between the beginning of the adventure and that confrontation to prepare for it or otherwise tilt the odds in their favor. In order for it to work the PCs will need lts ofinformation about what's around. Like: you have five days before twelve far-flung evil wizards converge on the city and you also have tons of vital information on where all twelve of them live and the layouts of their homes, go.
Plot Requiring Use of the Sandbox With Miscellaneous Occasional Chokers
From the point of view of Gandalf and Aragorn, this was what most of the Lord Of The Rings trilogy was. Timetables, NPC Hobbits randomly doing or undoing your bidding, various interested parties interfering, having to use hard-won gameworld information to solve The Big Problem.
This points up one way to throw in DM-generated plot stuff without ruining the PCs' sense of freedom--spread the DM interventions way out over time.
Small WorldSomething's going on, the PCs have to address it or die (or otherwise be rendered not-playing). Only certain tactics can be used, either because of limited information (like in an investigation--no use questioning anyone unless they had something to do with the crime) or limited resources (stuck in a shopping mall when the zombies attack, wake up in a dungeon with all your stuff missing, etc.).
In this kind of plot, there are no DM-imposed events or interfering NPCs after the initial set-up, it's just, in a sense, a very small sandbox--either because the PCs have very little information about the world to use or because their access to the world is physically limited. There's no danger of this being a railroad unless the only solutions possible are ones the DM has already conceived of, in which case, see Unipuzzle, below.
The Raymond Chandler
This is a typical noir or Cthulhu set-up. Another example is The Big Lebowski. Lots of unanticipatable events., basically. The PCs are trying to do something, and then someone interferes pretty much out of the blue (as far as the PCs know at the time) and creates a new problem, and then while the PCs are trying to solve that problem (or the original one still) there's another and another and another until the PCs basically realize that Unexpected External Events which they have no capacity to prepare for are pretty much a feature of the landscape around here.
Is this a railroad? Not always.
If the PCs have the ability to influence events outside themselves in ways the GM didn't anticipate, then it's possible to escape the railroad. "Outside themselves" is an important phrase here: in pretty much every game, PC's know their decisions can affect whether their fellow PCs live or die or solve the problems or don't--so that's not a test of railroading. The test is whether their responses to the problems are capable of creating unanticipated changes in the nature of the conflict itself.
If the GM has to think about what the PCs did and then meaningfully change the plot to reflect that--and the PCs can see that change, and see that it was a result fo what they did, the railroad tracks are gone. It's true that Lebowski's rug got peed on by DM fiat, but his troubles really began when he decided to talk to the other Lebowski about it. And he knew it.
In other words, this kind of event-heavy plot can avoid being a railroad if all these unexpected events are shaped by the GM to reflect actions the PCs have taken and if they don't push the PCs ineluctably toward some events whose important lineaments are predetermined.
The quibble here is the word "important". If, no matter what the PCs do, they're going to end up fighting the villain alone in unarmed combat on a moonless night on top of the Empire State Building, there's some railroading going on. If PC actions can change it so they fight the villain on top of a zeppelin with a swordfish (in a way UNanticipated when the scenario was conceived), then it's not a railroad. But there are obviously shades of grey in between. If it's hand-to-hand combat on a moonless night in a dark alley is that a meaningful difference? When has the change become "meaningful"?
My plan would be: overdo it. Every time the PCs make a decision, find a way to make the next event reflect it, even if you don't have to. This is harder than sticking to the plan and requires quick thinking, but it's also more fun for everybody involved.
I don't think it's that hard to run a railroad-free Lebowski adventure--however, what is hard is to publish one. The problem is: a published scenario has to anticipate player actions, and the longer the scenario is, the further into the future the scenario writer has to project the game events. Unless the scenario is endlessly filled with stuff like "make up something clever to happen here depending on how the PCs dealt with things so far" then someone writing a 20-session event-driven campaign for commercial publication pretty much has to assume the GM will railroad by nudging or nullifying in oredr to spit out a continuus stream of encounters. People with published scenarios need to learn to improvise if they want to get around this.
Waiting For Godot
Basically a plot consistently hinging on Anticipated Uncontrollable events--someone or something over which the PCs have no control must show up on its own and do things in order for the adventure to end and the PCs know it.
If you were playing R2-D2 and every other character in Star Wars was an NPC, Star Wars would be this kind of adventure. If you were in the middle of playing an adventure module you knew was called "The Coming of Glorp" and your PC hadn't heard anything about Glorp yet, that would also be this kind of adventure. There's more going on than you can encompass.
One of the big problems with this situation is that because the PCs know they'll never possess all the keys to solving the situation himself/herself, they aren't sure how fully they need to engage the gameworld. In a Preparation Adventure, if you don't do it, clearly nobody else will, in the Waiting For Godot, the PC doesn't have to truly examine or grapple with the world in detail because there's always the possibility someone or something else will shape events for him/her.
This is pretty much a railroad no matter what. Something has to happen and the PCs have no known way to hasten it happening. They don't know what to do to prepare, either. All they know is they have to pretend to be interested in whatever their immediate task is until the DM makes it actually happen.
The best, and perhaps only, way to make this good, is to use it as a source of tension and mood. It begins! Beware! Soon! This tension won't last long, however, so it's best not to let it drag.
As Small World above, except the resources are so limited that it effectively actually limits the solutions to any problems to a few things the GM's obviously already thought up and provided for (as in a video game) and is waiting for the PCs to figure out.
Examples: there's only one suspect that actually knows anything and only one way of approaching him/her that'll work or there's a puzzle door with only one or two solutions, and this suspect or puzzle door in turn points to one specific room with only one specific clue which in turn points to one specific device that can be used in only one way to take the PCs to...and on and on for the length of a whole session.
Contrary to what you might believe from hearing discussions about railroading, players do a lot of things during an RPG besides sit around and try to find a plot. A party that does a lot of inter-party dialogue and role-playing, or really enjoys combat, or is enjoying the snacks, etc., can probably handle a day of UniPuzzle without bucking, but it will probably begin to feel like a railroad if it goes on much longer than that. If the UniPuzzle also manages to prevent the players from doing the non-plot-problem solving things they like to do (talk in-character, hit monsters, etc.) then their patience will be even more limited, since, in effect, there really will be nothing to do but jump through hoops.
Personally, as a GM, I avoid writing this stuff, since I really don't like sitting around waiting to see if the PCs manage to learn to think like me, but as a player I don't mind running through this kind of scenario once in a while, since I think it's fun to try to figure out what whoever wrote the thing was thinking.
There's a fight. Winning automatically leads to a clue, the clue points unambiguously to a new encounter, which means you fight. If you survive, you'll automatically find another unambiguous clue, which leads ineluctably to another encounter, etc.
The only meaningful choices you have are what tactics to use in the fight, what to talk about between fights, and whether to keep playing or not. You either have no access to the outside world or the fight is concocted in such a way that nothing from the outside world could possibly help you. Again, a party that enjoys just being a party (dialogue, rolling dice, etc.) can probably tolerate this for a session before it feels like a railroad.
Essentially identical to a UniPuzzle.
UniPuzzle or Gauntlet + Unanticipatable Events
Like one of the adventures described above, except in addition to being made to jump through hoops, the PCs also get interfered with at intervals by events they themselves did not initiate. Whether this is automatically any more of a railroad than the UniPuzzle, or Gauntlet is arguable, but it easily has the potential to be a good deal more confusing and frustrating, since the PCs are not only trying to figure out what's on the GMs mind, they also have to stop in the middle of doing that to think about what else is on the GM's mind.
There are things to do, if the PCs don't do them, someone or something will do it for them. If they make a choice that isn't planned for, it's somehow nullified.
These always suck unless the players consistently think of it as a point of pride to try to surpass obstacles in the scenario before the GM does it for them. Even then I think they still suck, from the point of view of the GM who wants to be surprised.