Goblins and game theory

In my game world, hobgoblins and bugbears are not a separate subspecies, but goblins who have been “promoted.” A goblin tribe has the power to elevate its members to hobgoblinhood or bugbeardom. Let me talk about how that ability, plus elements drawn from game theory, might naturally produce goblin-only tribes; goblin tribes ruled by hobgoblins; bugbear gangs; and a few powerful hobgoblin empires – in other words, enemies that can challenge PCs at different levels. In fact, the same goblin tribe might rise with the PCs from level 1 to name level, remaining a stubborn threat throughout.

Ok, so the rule is, any dozen goblins can cast a ritual that turns a thirteenth goblin into a different goblinoid (hobgoblin or bugbear). Any one of the dozen goblins can sabotage the ritual without revealing its identity. Only goblins have this power, not the other goblinoids.

Now every goblin tribe is in a prisoner’s dilemma.

Let’s take a tribe of all goblins. Each would like the power of a bugbear or hobgoblin, but no one wants to give rulership to another. Even a goblin chief has trouble finding a dozen goblins loyal enough to elevate it: at least one (but probably all) of any dozen will secretly ruin the ritual.

Of course, all the goblins would benefit if they just agreed to promote each other to hobgoblins or bugbears (except the unlucky last 12 goblins who don’t have enough compatriots to cast the ritual on them). But a goblin would benefit even MORE if it was promoted and others weren’t.

Thus, a tribe of goblins are in a Nash equilibrium (part of an economic game-theory model proposed by John Nash, the Beautiful Mind guy). No goblin wants to change the status quo for the better, because whoever makes the first move (by promoting someone else) is likely to benefit the least.

This goblin mutual distrust means that players will encounter lots of goblin-only tribes, suitable for first-level characters to beat up on.

However, when things get tough for the goblins, this changes.

For short-term threats, like an owlbear wandering nearby, goblins might convert a few of their number into bugbears. Bugbears are tough and sneaky, but unlike goblins, they’re not community- and lair-minded. At first, they might accept tribute in exchange for fighting the tribes’ enemies; but they’ll soon get bored of bullying their weak cousins and wander off to form their own bugbear clique – perhaps even hunting goblins of their original tribe – or to seek their fortunes as minions of mad wizards. That’s why bugbears are fairly rare as part of goblin tribes, but are often found as wandering wilderness monsters or level 2 dungeon encounters.

For long-term threats to a goblin tribe, raising a hobgoblin to rule the tribe looks pretty good compared to being enslaved by orcs or slaughtered by humans. So when goblins are under serious attack, a hobgoblin chief arises.

A tribe with a single hobgoblin is not stable. Hobgoblins are teamwork-oriented, so the first hobgoblin will probably demand a second hobgoblin, and so on. However, hobgoblins like to have someone to bully, so they’re likely to stop once they’ve gotten a nice little hobgoblin war band surrounded by goblin slaves: a feudal system, essentially.

The equilibrium of this configuration means that there are lots of goblin tribes with elite hobgoblin nobles, especially in areas where the PCs have been slaughtering goblins. So on day 2 of the Caves of Chaos, when the PCs return to finish clearing the goblin lair, they’ll find that their opposition just got stronger and more disciplined.

This dynamic is stable until the tribe begins to meet with success. Once the hobgoblins have lots of non-goblin slaves, they look at their goblin minions, not as servants, but as potential comrades in arms. They’ll expand their army by converting all their goblins to hobgoblins, except for the smattering of goblins needed to cast the ritual and a few bugbears to act as scouts.

That’s where you get your classic Roman-style hobgoblin armies with dreams of conquest – a good match for mid-level characters. PCs returning to the Caves of Chaos on day 3 may find that all the surviving goblins are now hobgoblins, and the kobolds and orcs are now their footsoldiers.

Of course, this “leveling up” of the goblin tribe relies on the PCs never doing the logical thing and slaughtering the whole tribe. But game theory suggests a plausible solution there too. The Nash equilibrium inspired the military doctrine of mutually assured destruction.

Let’s say that there’s a fourth type of goblinoid, whispered about in human villages, feared even by goblins, but rarely encountered: a mindless Tasmanian devil of wanton destruction which kills everyone it encounters, including the goblin tribe to which it once belonged. It can be created through the same ritual that promotes goblins to any other goblinoid.

In tribute to the Nash equilibrium which inspired it, let’s say the goblins call this monster “the Gnasher.” For stats, I suggest using a flesh golem. There are three reasons for this: 1) with its high HP and immunity to nonmagic weapons, a flesh golem is capable of slaughtering goblins and low-level PCs indiscriminately; 2) it’s got a rampage mechanic which describes how I want the creature to act all the time; and 3) it’s alphabetically close to goblin, so you only have to flip a few pages to keep track of monster stats for your apocalyptic goblin battle.

Goblins are not suicidal. They know that if they cast a ritual to create a Gnasher, they’re likely to be its first victims. But goblins are also vindictive. If they’re cornered in their lair by PCs bent on slaughter, the last 13 goblins will join hands and chant, and then, next round, one of those goblins will turn into something the PCs might not be able to handle.

As a side effect of this, here’s another encounter the PCs might stumble into. While exploring a forest or in a cave system, the PCs find some moldering goblin corpses, and then, further on, among goblin huts and fortifications, a reeking slaughter, like a goblin battle with no survivors. Right here, the PCs should probably decide to go back the way they came. If they continue, they’ll see a lone creature, like a big, blood-stained, grotesquely muscular, misshapen goblin, walking in circles and yammering and growling to itself. And then the monster will see the PCs, and charge.

8 Responses to “Goblins and game theory”

  1. Rhenium says:

    Very nice post. One could use a thoul as the gnasher…

  2. greyhobbit13 says:


  3. Yora says:

    Here’s one method by which a powerful goblin leader can assure his ascension: Decimation.
    Get a dozen of the lowest ranking goblins of the tribe who have little support from tribe members who wield some influence and power, and order them to perform the ritual. If the ritual fails, one of them will be randomly killed and replaced by a new one. And this will repeat until the chief becomes a hobgoblin.
    Which would also be interesting in terms of game theory. At some point the chief won’t be able to afford killing more underlings or someone might want to assassinate him in fear of being the next one to join the group.

  4. 1d30 says:

    Wonderful idea, a lot like the party card game Resistance. Methods from that game (and Werewolf, Mafia, etc) can be used by the would-be hobgoblin to identify the spoiler-goblin in the tribe who keeps refusing to let the ritual succeed.

    Yora’s decimation idea is nice but would require that the leader already has political power, otherwise the other goblins would just kill him instead of even starting the ritual. How can he have supporters but not enough to complete the ritual? I think he would eventually get a success by swapping out ritual members until he got 12 others who were on his side.

    It would be funny to have a goblin tribe that’s upset because they keep trying to promote their chief but it always fails – because it’s the chief who is spoiling the ritual.

    You might wanna rephrase the rule as follows:

    Any 13 goblins can begin a ritual to promote one of those 13. They verbally chant the name of the goblin to be promoted, and then mentally cast their ballots “yes” or “no”. Even a single “no” will make the ritual fail. When promoting a Gnasher, a random goblin in the ritual will be promoted if all 13 vote “yes”.

    This means the ritual is explicitly a private vote on a public motion, rather than potentially having each goblin casting a private vote for a private promotion. If that were the case, six could vote for Gobba and seven for Gobbb (resulting in no promotion, of course).

    This also means that you can’t have 12 goblins get together and promote Gobbc to Gnasher just because they’re dicks. Gobbc needs to be in on the ritual with the understanding that he might be transformed or he might be slain by his fellow who gets transformed.

    (Now this is the Dwarf Fortress talking) 13 goblins could create a set of pens with trapdoor floors leading to the dungeon entrance. The 13 goblins get in their individual pens, hold hands or whatever through the bars, and do the ritual. Whichever one gets Gnasher’d is spotted right away and an assistant with a bank of 13 levers drops the Gnasher into the attack chute. The other 12 goblins are released from the trapdoors in the tops of the pens. This is expensive and labor-intensive to set up, but it means the whole tribe can produce Gnashers and drop them in the way of an invading party. With multiple tunnels and a system of valves to open the way, and a drop chute from the ceiling at the very end, and some kind of animal lure for the Gnasher to chase through the tunnel, the goblins could deposit a Gnasher virtually anywhere on the dungeon level in a few rounds.

    But I guess they probably wouldn’t think of that or have the craft skills to set it up.

  5. 1d30 says:

    I’m sorry I don’t understand what just happened but I think some bastard is drinking my milkshake. Please delete those video links if you can – I absolutely didn’t insert them.

  6. Gus L. says:

    The goblin mythology is a fun bit of vanilla D&D setting modification and monster ecology, but more interesting to me here is the idea of dynamic monsters that this post points to. Not a new idea but…

    In a sandbox setting it may be worthwhile to promote encounters (especially random encounter) that survive the PC party. E.G. 2d6 random encounter goblin woodsrunners become a small teibe with a log fort in that hex if over 1/2 survive an encounter with the PCs. A second encounter may increase the monster’s power more (bugbears join the tribe, a shaman with som thoul goblin mummies is added, or something). Of course monsters that actually kill PCs might get a name and extra HD and such as well.

    The power level and complexity of foes, or more specfically creatures the party interacts with, increases with thier time “on screen”. Not with PC power level, but with how important they become to the players.

  7. Vulture GM says:

    Inventive idea! I plan to steal this, but use the following modifications:
    1: six goblins needed per ascension ceremony.
    2: any one promoted a fourth time becomes a golem, but keeps the bugbear brute trait.

  8. Herb says:

    I am about to start a campaign and I will be working with this. In fact, I’d like this to be a secret that the players discover during play. While I will leave clues I will also plan some locations where this ritual is a planned tactic to use against an attack.

    One touch I’m going to add is the goblins doing the ritual do not know who will be upgraded. Thus each must be willing for one to be changed and be willing to be that one. If the players actually see the ritual happen, as I hope, it will be a random goblin they see upgrade.

Leave a Reply