The thing is, I'm not sure a linear dungeon counts as "being railroaded". Let's say there are 17 different routes you can take through the dungeon to reach the dragon at the end. Aren't you still being directed to reach the dragon? So it's still a railroad, right?
Now you disappoint me. Apparently 'railroad' is a loaded term and you appear to have a more restrictive view of it than I do.
Regarding your example: Who says that the 17 different routes all lead to a dragon 'at the end'? I'm not even sure what 'at the end' is supposed to mean. In a non-linear dungeon layout there is no 'end'.
One of these 17 entrances might lead directly to the dragon. In fact I consider that highly likely. The dragon needs a way to quickly get in and out of its lair, right?
And who says that the dungeon needs to be static? Shouldn't the monsters roam around in it?
I agree in as much as that it isn't as important how many different routes there are, as the question whether these routes result in the players being able to make meaningful choices. E.g. many official modules contain a 'maze' of some sort. I hate that, because a 'maze' usually doesn't offer any meaningful choices. Do you go left or right? That's a non-decision.
I don't think so. A railroad, in my opinion, is the DM trying to prevent players from making choices they want to make. (If you force them to fight the dragon, even though they are doing a good job avoiding such a thing, that might start counting as a railroad.)
No one is disputing that. But this is a thread about dungeon design, right?
Getting back to the origin of the term 'railroading': People started to call this 'technique' that way because they felt the adventure was on a single-lane track. If your dungeon layout represents a single lane from the entrance to the dragon's lair, how would you call it?
Now, what happens when the party reaches the dragon's lair (does it simply always attack them, or could it be bargained with, etc.) is a completely different question and not the topic of this thread.
Furthermore, each time you add an alternate path in a dungeon it becomes more difficult to make all the parts of the dungeon work together, and you can easily end up with a less-interesting dungeon. E.g., what happens if the heroes skip Room A, where the key is:
- They don't actually need the key
- They can always go back and look for it later...in which case the point of the non-linearity kind of evaporates.
So because it's 'more difficult' you should avoid non-linear dungeons? Come on! I think you can do better than that!
Did you even look at the articles about dungeon design? It's perfectly fine to have 'choke points'. If you have a dungeon with three sub-sections, each of them can still be non-linear.
It's a different question whether you should include a door that cannot be opened or circumvented by any other means than a single unique key that you have hidden in one of the rooms. Personally, I'd try to avoid that. I'd also like to note that even in a completely linear dungeon the players could potentially miss the key.
And regarding the 'They can always go back and look for it later' part: this isn't actually a given. Complications could arise to make this difficult or even impossible. Did I already mention monsters that move about? If the key leads to an area that even just one of the inhabitants sometimes needs to or wants to enter, they might pick up the key (or have a duplicate).
Non-linearity doesn't 'evaporate' just because reaching one room requires something that can only be found in one particular other room. It's already an improvement over a linear design if they can skip one or more rooms at all. And non-linearity is the most fun when a combat or chase scene happens that involves moving around the dungeon. Even if it's a combat that 'cannot be avoided' it will become a lot more interesting if it takes place in an interesting environment. And unlike a set-piece combat in a carefully designed room, the players actually have some agency in deciding where they want to take the fight.
Finally, it's quite clearly a fallacy to consider something a railroad just because at some points all of the different branches lead to a single concluding scene. If that was the case, a campaign module like 'Masks of Nyarlathotep', which is widely considered as something like the holy grail of sandbox design would also be a railroad because in the end the investigators either stop Nyarlathotep or the world's end has come. It's all about the journey to get there, and it's unlikely that any two roleplaying groups will solve it entirely in the same way.
