1. Species that simply prefer living underground (either because they fear the sun like the drow or because they love the dark like the dwarves).

2. Magical construction techniques that make huge, underground constructions more plausible.

3. Magical creatures that either have an instinctual need to create underground complexes or which create them as an unintentional byproduct. (Where did all these twisting tunnels come from? Well, they started as purple worm trails. Then the goblins moved in.)

4. Catastrophes on the surface world that prompt people to flee underground are also a great explanation for underground complexes. (See Earthdawn. Or just an Age of Dragons.) Mix-and-match with the techniques above to explain how the huge cataclysm refuges were built. Then simply remove the danger and/or (better yet) introduce some new danger that came up from below and drove all the vault dwellers back onto the surface.

It’s also useful to establish a method for underground species to generate food. In my campaign world there’s fey moss, which serves as the basis for fungal gardens. Huge, artificial suns left behind in underdark chasms by the vault builders or the under-dwarves also work.

I don’t find it valuable to do full-scale urban planning or figure out exactly how many toilets the goblins need, but I do find that at least some degree verisimilitude makes for better games: If the goblins get their food from fungal gardens, then their food supply can be jeopardized by destroying those gardens. And that’s either the basis for an interesting scenario hook or it’s a strategic master-stroke from the players or it’s some other surprise that I hadn’t even thought of before the campaign started.

Fey moss itself is useless. It’s a black, scummy substance. If left unchecked it will cover almost any surface with a thick, tar-like substance.

Sunlight almost instantly destroys fey moss, causing it to burst into flame. Unadulterated fey moss is also extremely flammable.

But, like the plankton of the ocean, fey moss is the bedrock of the underdark’s ecosystem.  Ecosystems on the surface all ultimately draw their energy from the powerful rays of the sun – plants capture that energy; herbivores eat the plants; and carnivores eat the herbivores. But in the underdark the ecosystem ultimately derives its energy from fey moss (which, in turn, draws it directly from the magical ley lines).

Animals in the underdark either eat the fey moss directly or they eat the wide variety of fungal species which have adapted themselves to parasitically grow upon the fey moss. Civilized species establish vast fungal gardens to feed their populace.

Vornheim includes a set of stripped down guidelines for giving PCs a set of contacts in an urban setting: The PCs can hit up their contacts for information about a particular topic and there’s a table for randomly determining what their reaction to the question is. (Here’s an example of the system in practice.)

Technoir is built around a plot-mapping mechanic in which PCs are created with a set of contacts: When the PCs hit up one of their contacts, there’s a system for randomly determining what they know. (And here’s an example of that system in play.)

The Technoir approach is built around the assumption that the GM — taking into account the subject indicated, that subject’s position on the plot map, the contact’s relationship to the plot map, and the specific question that was asked of the contact — will provide an act of creative closure and figure out what the contact says. And, in general, that works just fine.

But I thought to myself: Wouldn’t it be useful and nifty if I had a Vornheim-style contacts table for Technoir? So that the rules of Technoir would produce the lead the contact was pointing them towards and then the Vornheim-like table would give some guidance on how they ended up pointing them at it?

The Vornheim contacts table includes some null value (“I don’t know anything about”) values, which don’t work well in Technoir. So I tweaked the table a bit and ended up with this:

Sometimes it takes years for your brain to puzzle things out.

For example, I wrote Revisiting Encounter Design way back in 2008. The basic thesis was that you should generally abandon the new wave fetish for My Perfect Encounters(TM) and embrace a more flexible method of encounter design that would emphasize faster-paced, strategic-based play.

The four major tenets I argued for looked like this:

(1) Design most 3E encounters around an EL 2 to 4 lower than the party’s level.

(2) Don’t be afraid of large mobs (10+ creatures) with a total EL equal to the PCs’ level. The common design wisdom is that these creatures are “too easy” for the PCs. This is true if you’re thinking in terms of the “common wisdom” that sprang up around misreading the DMG, but in practice these types of encounters work just fine if you’re looking for fast encounters and lots of them.

(3) Encounters with an EL equal to the PCs’ level should be used sparingly. They should be thought of as “major encounters” — the memorable set pieces of the adventure. It actually won’t take very long before the expectations of your players’ have been re-aligned and these encounters leave them thinking, “Wow! That was a tough encounter!”

(4) And that means you get even more bang for your buck when you roll out the very rare EL+2 or EL+4 encounter.

(The general philosophy of this advice, it should be noted, is widely applicable beyond D&D. In Feng Shui, for example, it means “keep a healthy supply of mooks flowing through your scenario.” In Shadowrun it means not letting a ‘run bog down into a single giant melee; keep the action on the hoof by making it possible for the PCs to cut their way rapidly through waves of security. And so forth.)

Most people seem to have grokked what I was selling. But there was a smaller group of people who insisted that I was wrong: That if they built an encounter with sixteen CR 2 creatures that the PCs would take a lot more damage than if they used a single CR 10 opponent. In fact, I still get fairly regular e-mails to this effect six years later. And I could never figure it out: Running the math on hypothetical scenarios regularly confirmed what years of play and hundreds of game sessions had taught me. It certainly wasn’t impossible for the mob of CR 2 creatures to out-perform the CR 10 creature, but in general the fighter was going to rapidly cleave through the mooks or the wizard’s fireball was going to rip them apart.

Were the people e-mailing me fudging dice rolls to toughen up the weaklings? Did their players just have no idea how to use mass damage or area control spells?

Nah.

It took six years, but then I was driving down a highway in Wisconsin the other day when epiphany finally hit me: These are reports of anecdote. And the problem with anecdote is that it selects for the exceptional and the unusual.

You don’t remember the 19 times that you used a CR 10 monster against a 10th level party and it took a few rounds to take it down while tearing out a few large chunks of hit points from the group. Instead, you remember the time that the party faced a single tough opponent and miraculously chopped his head off in the first round of combat.

Similarly, you don’t remember the umpteen times that your wizard casually fireballed a group of mook orcs and cleared ’em out without any hassle. What you’ll remember is that one time that a horde of kobolds left the PCs screaming and fleeing in terror.

It’s also likely that the actual numbers aren’t actually being looked at in these anecdotes: “Remember that time that the six orcs in area 4 were a lot tougher for the party to take out than the demon in area 10?” Sure. But were those actually EL equivalent encounters? Or were the orcs all CR 8 (for an EL 13 encounter) while the demon was CR 10?

None of this is a problem, of course, unless you start using the exceptionalism of your anecdotes as a guiding principle of scenario design. You want your scenario to be exceptional, of course, but you won’t achieve that if you’re expecting the statistically exceptional in every encounter.

(Of course this is another advantage of the encounter design method I advocate for: By increasing the number of encounters experienced, I increase the number of opportunities for the memorably exceptional moments to happen. Sometimes that will be the result of improbable math; sometimes it will be the result of clever and unexpected play. That’s the beauty of a non-deterministic medium.)

The first thing to say about Big Hero 6 is that it’s a ton of fun wrapped in a beautiful aesthetic surrounding well-earned emotional heartaches and catharsis. If you’re the type of geek who’s likely to be reading this website, then you’re probably going to adore this film.

With that being said, I was interested in the way that Big Hero 6 failed to be an ensemble movie: It comes very, very close (featuring a diversity of interesting characters in supporting roles), but ultimately misses the opportunity. (And that’s not necessarily a bad thing for the film, which instead succeeds very well at being a film about its protagonist.)

I found myself making an easy comparison to The Incredibles, which is in many ways a completely different movie but which also features the formation of a team of superheroes. But whereas Big Hero 6 just misses being an ensemble piece, The Incredibles is a very successful ensemble piece.

The key distinction here is that all four main characters in The Incredibles are fully developed while each possessing a full narrative arc (which is also connected to the overall narrative arc of the ensemble). This is contrasted to Big Hero 6 where all five members of the team are given unique personalities and great dialogue… but notably lack fully developed arcs.

The reason I bring this up is that I noticed that a key difference between the films is that The Incredibles featured multiple sequences in which the main characters are separated from each other, whereas Big Hero 6 basically did not.

So what I’m saying here is:

SPLIT THE PARTY.

Because it’s a really effective way to allow individual characters to develop identities separate from the group identity. (Which will, ironically, enrich the group identity.)

(More on splitting the party over here.)