Won't that lead to players just assuming there are 3 always 3 doors? AKA, if they enter a room with no doors, they know there are 2 secret doors. Not 3, not 1, but 2.
What is it with theRPGsite wanting to take every useful rule of thumb and analyzing it on the assumption that it must be used 100% of the time?
"Make sure you drink lots of fluids."
"MY GOD MAN! IF YOU STICK A GARDEN HOSE IN YOUR MOUTH AND TURN IT ON FULL BLAST FOR 24 HOURS YOU WILL DIE!"
Maybe I am dense. I absolutely support breaking up linearity. However, the concept here is just make sure the majority of points have a minimum of three lines leading to them and I don't know what we are gaining from THE MAJORITY having THREE OR MORE lines.
It doesn't feel like architecture.
I'd say the biggest oversight in this particular rule of thumb is that it ignores the common hub-and-spoke design.
When you're talking about jaquaying, though, what you generally want to do when analyzing the navigational routes in your dungeon is to simply ignore dead ends. (You can still have them, but when you're doing structural maps like
these, you simply ignore them.) Once you've done that, the rule of thumb becomes more useful.
Or, alternatively, you can flip that around: Any room which has less than three exits is
navigationally irrelevant insofar as non-linear/jaquayed design is concerned. Build a ball-and-stick diagram accordingly to see what the fundamental structure of the dungeon is.
For example, consider this dungeon map:
There is only one point in that dungeon which is jaquayed: The initial crossroads just inside the entrance where you can head in three directions. If you draw that as a ball-and-stick diagram it would like this:
And nothing more.
By contrast, look at this dungeon map:
If you map this dungeon out while eliminating any room that doesn't have at least three exits you get:
Which makes the structurally significant rooms obvious.