What I like about Dungeon World, and what I do not

by robdalexander

Status: quite confident. I’ve continued to update this since posting it, partly from some notes I found from 2014 (when I was playing it regularly).

I’ve played a fair amount of Dungeon World — perhaps 45 sessions in all, about 8 as a player and the rest as a GM. You may reasonably doubt my memory of these, as only three of them were in the past two years. Nevertheless, I have views, and I shall state them.

Overall, I like some properties of DW, but I strongly dislike other ones. And I do not know how to make a game that has only the ones I like, or to what extent that is even possible. My primary goal here is to help myself understand DW, and my experiences with it, so that I can design games that I like better.

Some top-level clarifications based on feedback:

  • This is not a review. It’s a very idiosyncratic exploration of my subjective response to the game and the reasons for that. That said, if you’re evaluating DW before buying or running, it may be of some value (insofar as you are like me). If you’re designing a DW-like game for a broad audience, it may be of some value (insofar as many people in your audience are like me). But primarily this article is for me. If you want to understand me (as I do), it’s likely to be useful.
  • When I say “design”, I don’t just mean game design. There are aspects of the writing and art that don’t work for me, and I think they strongly colour my experience of “Dungeon World”.
  • It may help to know that my current game is Lamentations of the Flame Princess (LotFP), house-ruled and run as described elsewhere on this blog. If not stated in any specific case, that is probably the reference model I have in mind (especially when I’m describing where DW works well for me). This is not to say LotFP works for me, either — overall, I like it less than DW.

This post is long. Bring a torch.

Things that I like about Dungeon World

  • Compared to Lamentations of the Flame Princess, it has a much more explicit top-level structure — a game loop that runs continually. In brief:
    1. The GM describes the world state, and what’s just happened
    2. The GM asks the players, or one player in particular, “What do you do?”
    3. At least one player responds
      • If they described plausible action corresponding to move trigger, execute the move
        1. If the move fails (roll of 6-) and no special handling of that is given in the move text, the GM makes a hard move
      • If they describe a golden opportunity for the world or an NPC to fuck them up, the GM makes a hard move
      • If neither of the above, the GM makes a soft move
    4. Loop back to (1)
  • That top-level structure constantly reminds the GM of how they are supposed to run — they are to respond to player actions with GM moves that (a) are interesting right now (b) set up further need for action and (c) are consistent with the fiction as previously established in play.
    • In doing so, it reminds the GM that they’re not supposed to be following the common alternate practice (and which I mostly use in Immergleich) — to focus on simulating world logic and maintaining the integrity of the setting, including things that the GM previously committed to (e.g. in written prep) but that the players have not seen yet.
  • The top-level structure keeps the conversation tightly coupled to the fiction — to the natural language description of the imaginary world. There are few artificial constructs (e.g. rounds, turns, stakes, Let it Ride) that you need to keep in mind or spend precious words on. And staying in that one “space” (the fictional one, as distinct from an abstract rules framework that scaffolds it) makes a lot of reasoning task easier. It is easy to make the action of DW flow smoothly.
    • There are a few abstract constructs that matter (Forward, Ongoing, Hold) but they are not always present and generally seem manageable. (Although I’ve generally felt Hold was more intrusive than the others)
    • A downside of this —is that I find hard to zoom up to a higher level of abstraction. If we zoom away from the immediate and into the “over the next several days” none of the moves as written will ever trigger.
      • In contrast, Burning Wheel and The Shadow of Yesterday are extremely strong at this, since their resolution systems are much more generic — they care about ability and stakes, not about precise situations.
      • To be fair, most RPGs (especially D&D variants) are even worse at this — their rules mostly don’t apply, and are much harder to abstract.
  • Monsters stats are easily created on the fly — flick down a list, generate three numbers, one free-text instinct and a couple of free-text moves and you have a monster as complex as the rules expect. Everything is built on the assumption that that is all you have done.
    • Contrast any mainline D&D edition, where you can get buy with simplified stats but the rules assume a larger statblock exist (and, usually, there are no handy default values e.g. what is the default Con save for a creature like this).
    • The resulting stats do a weak job of modelling overall power level (see the reference below to “the 16hp dragon”) … , and yet they do some job, which is confusing. But I’m putting this, overall, in “things that I like” because it is so effective at supportive improv of monsters.
    • (To optimise the process, Jeremy Strandberg’s Monster Creation Cheat Sheet is much easier to use than the standard one)

Things that many others dislike about Dungeon World, but which don’t bother me

  • Lack of mechanical-tactical depth — the experience of DW is nowhere near the rules-based miniatures wargaming of 4E combat, nor near Torchbearer in how it extends such rules-mastery-focus to everything you do. By contast, in DW much more of your tactical effect depends on the GM’s subjective judgement, much less on the rules.
    • In practice, this doesn’t bother me, and it seems to play to the interests of the players I attract. They most don’t want to master rules; they want to respond to the fictional situation, play with it, and enact their idea of their character.

Things where I am conflicted — I see the benefit, but I the cost claws at my heart

  • The PBtA move format (“When you X…”) is a nice way of capturing new special-case rules
    • …but it’s hard to remember those triggers in play. We are not computers. (And even there, programming with triggers that can fire at any time is hard. Aspect-oriented programming died a hard death of programmer incomprehension.)
    • It feels like there are a huge number of moves that can be notionally in play and able to be triggered at any time. I see little hope of keeping track of all of them, and yet that is the expectation as written. Although there are clear general principles for when GM moves take place (see earlier), there are no such general rules for PC moves — they can trigger on any damn thing at all.
      • I think my main problem is that there are so many (countless) special cases that may trigger a player-facing move. I.e. countless special cases where “some rules engage”. In contrast, I am much happier with games like Burning Wheel or The Shadow of Yesterday which have a single universal principle for when you move from reasoning about fiction to using explicit rules.
      • A number of people have said to me “players are responsible for spotting when they’re triggering a player-facing move”. I’ve not found players reliable, however, in doing that. And I don’t want to offload that on them, anyway, especially when combined with the exhortation to say what their characters will do in the fiction, rather than to state moves.
      • It may be that we’re not actually missing many move triggers in play (at least compared to the number of rules we forget in other games), or that the game isn’t suffering for it. Maybe the only problem here is my own disquiet and disatisfaction because I think we’re missing triggers (and thus missing good opportunities for the rules to improve the game).
    • The presenters of the Misdirected Mark podcast express a similar complaint about PBtA games in general, although they are talking about confusion when learning a new game rather than playing one for a long period. It’s in episode 265, starting around 56:50
      • They contrast this experience with the Blades in the Dark one, which is very similar to Burning Wheel in that is has a single trigger for its main mechanic (the Action Roll)
    • The move model is, of course, easily ported to other games
  • The basic three-way action result (fail and hard GM move; yes but; yes) is fun, especially given that 7-9 results are the most common.
    • … but it cries out for more generic implementation in place of move-by-move special-casing
    • … and coming up with 7-9 results on the fly can be hard.
      • In contrast, stake-setting games like Burning Wheel and The Shadow of Yesterday make you decide this before the roll, when there’s less immediate time pressure. I remember to do this by using a ritual form — I set up every roll with “The Obstacle is X. If you succeed… If you fail…”.
    • … and in any case it’s not universal —  some moves instead use (fail and a hard move; partial success; full success) which is another form to master and remember.
  • Because of the narrow range of die results, and the fixed values for for interpretation (6-, 7-9, 10+) there is very little room for modifiers e.g. from equipment or spell effects.
    • Stacking modifiers are not always a good design decision. In going from 3.5e to 5e, D&D stripped back a lot of modifiers, and limited most of them to (advantage/normal/disadvantage).
  • The assymmetry of players and GM, where the GM follows completely different rules and never rolls dice, feels odd to me.
    • This has advantages, though, in that the GM’s rules are much simpler and less fiddly per-character, and taking away dice means there’s one less thing for them to think about. And GM time is the main bottleneck in GM’d rpgs.
  • The assymmetry of PCs and NPCs seems odd to me.
    • RPG rules seem most natural to me when all characters are modelled the same way. I blame RuneQuest, which was my formative model of what game rules should be like (even though I never actually ran or played it). It modelled PC and NPCs (including the most alien of monsters) with exactly the same parameters — attributes, derived stats, skills, hit locations. To this day, D&D giving monsters “hit dice” rather than class levels, and (in most editions) not giving them attributes, feels slightly weird and wrong to me. It’s like moving to a country where no-one ever wears socks.
    • I’m pretty happy to abstract NPCs, e.g. to assume that characters have average attributes unless specified, and indeed to assume that a generic human adult has the same standard properties as any other. But modelling NPCs completely differently never quite feels right.
    • This has major benefits, though, in making NPCs very mechanically simple to stat up and to adjudicate.

Things where I have only displeasure — I see only problems, not benefits, and they smell of death

  • The player move triggers always apply, and are not subject to direct GM judgment. The tone-management and fiction-integrity-maintenance role of the GM is dropped, and not directly replaced anywhere.
    • Contrast Burning Wheel, where the GM’s role is explicitly to decide what is interesting enough to warrant a roll (rather than just “say[ing] yes”) and to set tone (e.g. through fine-tuning the difficulty levels (“Obstacles”) of skill rolls).
  • There is the beginning of a difficulty level model, which spells out some key ideas about move applicability, but it only applies to one move — it’s in the expanded text for Hack & Slash. It’s therefore something extra to remember for that move, but adds nothing to other moves that might benefit (Volley is the most obvious one).
    • (In brief, an action that at a glance meets the Hack & Slash move trigger (“When you attack an enemy in melee”) may in fact be:
      • Simple murder
      • Unimpeded strike (do normal damage)
      • Run the move (roll 2d6+Str and do damage if you roll high)
      • Impossible (you can’t achieve anything here) )
    • The idea behind that (fictional positioning is has a major impact on whether moves are applicable ) is how you run the 16hp dragon, which you need to do to make many common fantasy things work in DW. But except for that one move the rules are very vague about how it should be applied.
    • Having similar special-cases expansions for all other moves would be even worse, as it would make all moves complicated.
  • The base game, as written, is predominantly about combat and adventure. It has minimal support for social, craft, political etc conflicts.
    • Contrast Burning Wheel, which is good at this
  • The damage system somehow always feels “soft” to me — it feels as if injuries are not injuries, just some vague idea of such.
    • Is this because damage is by largely by class, regardless of weapon or other context? Is it because there ? Is it because there’s a simple roll that can let you survive death? Is it because weapon tags are a mix of mechanical (e.g. precise) and purely fictional (e.g. messy)? I don’t know.
    • It’s possible this isn’t about the rules at all (see my comments on the art and writing, below)
  • There are no rules for PvP conflict. There are hacks and shoehorns to be found online, but none appear satisfying to me. In contrast, in many other games if two PCs are locked in a dispute I have ways of resolving it mechanically.
    • I could shoehorn such rules into DW, but they break the general pattern of how the rules work, which reduces my confidence that at any point I am applying them consistently, fairly, and in a way that will lead to good experiences longer term.
  • I greatly dislike the clean, friendly, cartoony art. It fees lifeless, sanitised, and active sucks pleasure out of play. I’d enjoy the DW book more if there was no art at all.

The original version of the following point was badly worded, and created a lot more heat than light. I’ve revised it. You can see the original, along with more detail, in my follow-up post.

  • As with the art, the writing is sanitised, soft-edged, anxious not to offend. It’s like the authors are very nice people who don’t want to offend or upset anyone, and so are watching their every step. I am not like that — I am not so nice, not so scrupulously respectful, not so keen that everyone has a good time.
    • Adam Koebel definitely seems like that. Watch him on Office Hours — have you ever seen a sweeter guy?
    • As a writer and GM, I’m not here to make you happy — I’m here to fuck with your neurology in a way that might stimulate new impressions. I don’t aim to be “nice” — I aim to be raw, vital, and unique. I aim to fire up, not to sooth; to generate, not to heal.1
    • As writers, I like the uncompromising “this is the product” of James Raggi (and many writers he publishes), the sharp edges of Zak Smith, and the the surprise and alarm and WTF of Patrick Stuart.
      • I wouldn’t necessarily use their work as-is, particularly Raggi’s. But it’s what I want to read.
      • If I could steal the powers of any writer, it would probably be Stuart2.
  • At heart, DW is a “love letter to D&D”3. Although I have used D&D a lot, I have no love for it.
    • My formative influences in gaming were Runequest 3rd (as rendered by Games Workshop in the UK, with the implied Fantasy Europe setting, not Glorantha), Dragon Warriors, Fighting Fantasy gamebooks, and to a large extent Warhammer Fantasy Roleplay 1st Ed. None of those are D&D; all of those are at least partly British.

What thoughts does that stir?

I am interested in comments on the above. As I said near the start of the post, my primary goal is to help myself understand DW, and my experiences with it, so that I can design similar games that I like better. I am particularly interested in your comments if:

  • You’ve experienced a specific problem above, and can tell me how to fix it
  • You’ve enjoyed one of the benefits above, and can tell me how to port it into other games
  • You’ve experienced anything I mention, and have new words to describe it

If you haven’t experienced the same things, if you haven’t felt the same way, I’m a lot less interested. But you can say words if you like.


1. Conjecturing beyond “what I like” and into “what is of intersubjective value” I see hard-edged, raw, sometimes-upsetting writing as potentially “real” — as containing value. I do not see soft-edged, carefully filtered, designed-not-to-offend writing as being capable of that. The universe is hard-edged, raw, and often offensive. Its truths do not care about how you feel.

I don’t mean that writing to be “dark” or “edgy” is good. That’s mixing up goals with side effects. Real, honest writing often ends up upsetting and “dark”, or pushes over the boundary of what’s socially acceptable. But those aren’t worthwhile goals in themselves.

For a discussion of an analogous error, see the aforementioned Patrick Stuart on mistaking the spoor for the beast itself.

2. Although I would probably misuse it.

3. IIRC that wording is from the creators of Torchbearer and/or 13th Age, not either of the DW authors.