Friday, June 13, 2014

A Minor Observation on Adventure Design

At North Texas RPGCon this year, one of the things I did was a first playtesting run for Cyclopean Deeps Chapter 9: Hidden Worlds of Jupiter Kwan. This particular bit of the Cyclopean Deeps has been an open question for a long time with me. It is a very complex map with lots of possible locations (at least 300, I'd estimate) and multiple possibilities for several things in each of the areas.

When you're approaching an adventure that contains locations with multiple nested possibilities, there are basically three ways to do it. You can key all the locations, you can create tables to generate the various internal possibilities on the fly, or you can pre-generate a huge number of locations using those tables.

Option 1 (Key ALL the rooms!) had to be discarded immediately, if for no other reason that when you've got 300 keyed locations, your map looks like it's nothing but numbers. You can barely tell which numbers go with which location.

Option 2 (Multiple Tables) was what I playtested. Each time the adventurers entered an area, I rolled on the several tables required to generate the area. I immediately discovered that it wasn't going to work. There was way too much page-flipping and fast reading going on as I tried to assemble a description of what the characters could see. So, I am now in the process of creating a large number of pre-generated locations. There is still some die-rolling required, but this will cut it in half and minimize the need to flip pages around.

Normally I consider pre-generated locations to be the worst of both worlds: by numbering everything you can get well-crafted descriptions that are highly unique, and by creating a full set of tables you can get an even larger number of descriptions, although they will of necessity be a bit less well-crafted than if they were all written out straight.  

What are the Characteristics of an Adventure that Works Best with Pre-generated Random Locations?
However, Chapter 9 of Cyclopean Deeps has a particular set of characteristics that happen to work best with pre-generated locations: the movement is fast from place to place (in real time), the areas involve several different possible components, and there are many of these locations.


  1. What about doing a combination? Give the charts, and some pre-generated locations. It's not ideal, but it's better than trying to do 300 locations yourself, or leaving DMs having to do all the work.

    Also for the areas you're pre-generating, maybe include something that they can roll for within the description? Thus making each description somewhat reusable.

  2. What David said. What if you did 20 or 30 rooms, just to give an example of what should be done with the overall tables? This way, an on-the-fly game can just run with what you've done but a campaign DM can craft his or her own set of descriptions based upon what you did.

  3. I agree with David and Steve, Matt: pregen your locations over time, and each time you test the level, create sufficient #s of locations for that session. Within running it 7-10 times or so, you'll likely have your 300 encounters all defined.

    Then you can take those encounter areas and use them as the basis to tweak the whole---move some around to fill or create gaps, space out similar encounters for better distance/proximity from one another, de-dupe your specials and traps and tricks, etc.


  4. The scale of the problem is, I think, the only thing new here. This is just like striking the right balance in conventional keying (placed contents vs. wandering monsters), except that it sounds like you have more dimensions in the matrix of possibilities, plus a large number of locations to fill up. Taking the lesson learned from conventional keying, I agree with the others: Present a set of fully-keyed locations (probably more than just 10% though), and then offer the tables to handle the rest. (First of several thoughts on this...)

  5. Take inspiration from the old Necromancer Games practice: Free web downloads for their modules, with more content!

    Even if you need to keep the book itself light (either by limiting the number of areas you fully key, or limiting the size & complexity of the random tables), you can offer more via downloadable pdfs:
    - Collections of areas with (nearly) complete keys, written by you *or others*.
    - More complex, nested, multi-dimensional tables, again written by you *or others*.

    That way, the off-the-shelf book is fast to use, but people can ultimately access a different level of detail if they want, whether pre-rolled, or do-it-yourself.

  6. "... when you've got 300 keyed locations, your map looks like it's nothing but numbers. You can barely tell which numbers go with which location."

    This tells me that your map might be too small, and that it might benefit from a size increase.

    After all, if I'm running this, I'm gonna end up with numbers written all over the map, anyway; because otherwise how will I remember what randomly-generated thing ended up in each location?

  7. Solve the numbers-on-the-map issue by using grid or hex coordinates. For example, the building at R-25 in Izamne doesn't actually have "R-25" written on it; but I bet you can still uniquely identify the building I'm talking about, right?

    (It's hard to know whether this technique would work for your location, obviously, because of scale; but throwing it out there anyway.)

  8. Some MERP modules solve one dimension of the matrix-of-random-tables problem through color/texture. (e.g., Thieves of Tharbad, or Sea Lords of Gondor.) On the map, each building has a different color/texture, which tells you the result for what would have been the first random roll.

    Obviously, for grayscale books, you are more limited here than for color books. So you may need to carefully choose which axis of the matrix the color/pattern maps to.

    Alternatively, if there are factions grouped within close proximity to one another, you might use a color/pattern on the area surrounding/containing all those locations.

    And if grayscale works, but color would be better, this dovetails nicely into the free downloadable pdf idea.

  9. Thanks for the excellent advice, everyone! I think you'll all see why I asked for help when you look at Chapter 9 of Cyclopean Deeps.