Losing is fun: What we can learn from Dwarf Fortress

Liveblog notes from an IET Technology Coffee Morning presented by Daniel Allington, from the OU’s Faculty of Education and Language Studies.

‘Losing is fun’ comes from Dwarf Fortress‘s original documentation. Losing can be fun – for the player. But also for the creator? And what are people trying to achieve when they make a game, and what do players want?

Dwarf Fortress is a work of art in a way that’s uncommon. Beautiful to look at – e.g. Myst when it came out – as the typical standard for games as art. But modern art isn’t beautiful to look at – think pickled sharks, not beautiful background.

Addendum, Dec 2012: For all those of you finding this blog post trying to find out ‘How does toady/Tarn make money out of Dwarf Fortress?’, the answer is simple: People who like the game give him donations. That’s it.

Tarn Adams, the creator, is the key figure, talks a lot about his life including Scamps the cat. Dropped out from successful academic track. First game had small following, modified to make it more and more complicated. So Dwarf Fortress is a sequel, downloadable for free – that was the original intention. Usually ‘free to play’ games are ad based. Dwarf Fortress has none, only developer’s Paypal account, makes about $30k/y. Industry very interested in the game – offered $300k franchise deal, but turned down because he feared it might reduce donations. Also offered developer post at a major dev company, turned down because he wanted to work full time on Dwarf Fortress. Does very little to make it popular.

Minecraft, hugely successful, was developed by a fan of Dwarf Fortress. Magic The Gathering – trading card game – the creator is also interested in Dwarf Fortress.

That strategy wouldn’t look unusual if it was a painting, or a novel. We assume people develop games to be popular, or make money – but not art.

Original inspiration: Nethack! And other Roguelike games. Plot isn’t set, not fixed mazes – generated anew when you play. All based on ASCII characters – @ for humanoids. Move using vi editor keys – hjkl. It’s virtually unplayable – your character very likely to be randomly killed in the first part of the game. Then very boring mid-part of the game, before you finally get to complete it. Cult appeal.

Dwarf Fortress graphics very much inspired by that in visual style, and gameplay. It’s a 3D world, but you only see it in 2D slices. You have to hold the vision in your head. Gameplay all via key presses and navigating a series of menus. ‘The only reason Toady [Tarn] gave Dwarf Fortress an interface was so he could test it’. It’s just a side benefit that it means that other people can play too.

World creation is extremely detailed, and complex. It simulates geology, and history – 250-1000 years of history, civilisations and figures who thrive and fall, and so on. Legends that you can encounter. There are three languages, so things have names in those languages, based on the history. You get other people coming to visit – wandering monsters to humans. The megabeast that wanders in will have a whole history, wandering across the whole map. Interacting with a large history.

Looks simple, but is incredibly demanding in processing power. The detail is in terms of the simulation – not rendering polygons. It tracks everything each individual dwarf does, their history, and the influence on their psychology. Romantic and friendship relationships. If they get too unhappy, they get destructive – might smash something up, or go mad. Can get in to a tantrum spirals, where a dwarf getting annoyed triggers a whole cascade of tantrums.

Nobles might turn up, once things get larger, and start demanding things. All at a ‘completely insane level of detail’.

Fan-created tutorial videos. Two modes – fortress (where you build a fortress), and adventure (where you move across the world). The combat is at a similarly silly level of detail.

Tarn Adams wakes up around 4pm, starts coding, at some point goes for his one meal a day, then codes until he crashes out around 6-7am. But the software is still alpha. It’s been ‘being released’ since about 2004. Estimated time for beta is about 20 years. Bits of it are constantly breaking.

A fan has created a 3D graphic visualiser. Also someone’s created Dwarf Therapist – a nicer interface to manipulate what each dwarf is doing. Tarn has (pre)released a game with an almost unusable interface, people loved it, so some have created graphics, and more usable interfaces.

Not released as open source, because Tarn fears losing income, and control – wants just a single vision.


Being released in a terribly broken state, is terribly boring, gets everything wrong – but has a small number of fanatical fans. (As opposed to a large number of casual fans like Angry Birds.) Works for the creator. His quest for autonomy is driving all this. Not dealing with the bits of the game he doesn’t care about. Bourdieu says this is what’s most important for the art world: rejection of popular, commercial value; art for art’s sake. That’s what we see here. Is fun, on its own terms – losing is fun.

Gets respect from other games producers. Artists want to be appreciated by other artists. Getting some mainstream world recognition. Interviewed by games sites, but this summer by the New York Times. Had Dwarf Fortress as an exhibit in NY Museum of Modern Art.


The community is really interesting. The relationship between them and Tarn is really important. He often allows things unexpected to happen. E.g. created a Necromancer that could raise the dead. Made rule – as long as the thing was alive, has a head or hands, can be raised. Put a dead body in the world, butchered, Necromancer raises a walking skeleton, and a walking hollow skin. Decided to keep that; it came out of the rules and was interesting. The dwarfs keep getting eaten by carp. Carp are carnivorous and human-sized; discovered that they eat the dwarves. Undead rules – they don’t mind on land or under water. Apply that to zombie fish that can go on land – zombie sharks that come on to the beach, or skeleton sharks, that come on land.

J: Is very open in some parts of the process.

Yes. Takes ideas from forums and implements them. Pays brother small salary to write short stories based in the world. Analyse stories, pick out things that’d be fun to have in the games. All posted online.

S: What does conceiving of it as art buy you?

Partly an explanation of Tarn’s motivations. Pursuit of own vision, creative autonomy, acts in a similar way to modern artists. Rejected commercialisation.

Yishay: Why wouldn’t you see it as art? I find it intuitive to relate to it as art

Jon: The conversational collaboration – with the audience, players, fans, taking account of what they think – is unusual in art.

Depends what you mean. Bordieu says artists produce for each other; talk about each others’ work. In that sense they are having a conversation within a smaller circle. It’s only people who make the investment to engage who have an influence.

J: Very niche community. How many play?

Nobody really knows. Number of downloads is meaningless – most people won’t play it. Two main tutorial producers on YouTube, each has many hits for first tutorial, which then trails off. 40k views, trails off to 10k; other nearly 500k, trails off more slowly to 10k.  Give an idea of size.

Yishay: How does someone become a Dwarf Fortress player? Incredible investment just to play the game.

I’m a casual Dwarf Fortress player. Started when had a lot of time with nothing to do – wife and son were staying with her parents. Requires a lot of time investment.

J: Other player’s stories?

Worth looking in to. Get hints. When most recent version released, May, one person posted saying they live in a trailer but still going to donate.

Rebecca: Do you have to be actively involved or do it in the background?

Have to be quite involved – e.g. while I was talking to you, the traders left before I had time to buy things, so there’s no food and they’ll likely starve. Does require a level of attention. Something Tamagotchi about it.

J: Reminds me of The Sims.

Trying to keep the entities in the games happy. Like when people post blog stories about The Sims – e.g. someone trying to create homeless people in the game. Absurdities from the simulations, where it fails. Like Boatmurdered fortress – a lot of counter-intuitive behaviour. Human interaction with the Sims. Not quite like real people.

Doug: Emergent properties from rules.

Everyone likes it when it surprises you.

Someone: Stories, when things change, evolve; also the game developer’s ideas. Updates that change things, and do you want it.

One effect of new version is the game is less violent, it’s harder to beat the goblins, so focus more on the pacifist elements of the game. So if wanted more fighting, stick with earlier version.

Someone: Autonomy, and how that rubs off on the player.

You can play earlier versions. Possibly a nice introduction.

J: Example of a guy who left the door open so the goblins came in. Is it losing is fun, or the challenge.

Interesting thing is the stories, which tend to end in disaster. Doesn’t matter – like Greek tragedies. The guy did the door to make the story more interesting. The best way to defend your fortress is with traps, but a perfectly defended fortress isn’t very interesting.

The more popular tutorial, I don’t like, he shows you how to beat the game. It’s a walkthrough that almost ruins it. Eventually you dig all the way down to hell and the demons come out. But this guy blocks off hell with an exploit – demons can destroy any object on the same level as them, but he placed a hatch on the floor above (which they couldn’t destroy). So changed story – no longer a vicious end. Then caused millions of tons of rocks to collapse on hell which destroyed all the demons.

This work by Doug Clow is copyright but licenced under a Creative Commons BY Licence.
No further permission needed to reuse or remix (with attribution), but it’s nice to be notified if you do use it.

Author: dougclow

Data scientist, tutxor, project leader, researcher, analyst, teacher, developer, educational technologist, online learning expert, and manager. I particularly enjoy rapidly appraising new-to-me contexts, and mediating between highly technical specialisms and others, from ordinary users to senior management. After 20 years at the OU as an academic, I am now a self-employed consultant, building on my skills and experience in working with people, technology, data science, and artificial intelligence, in a wide range of contexts and industries.

6 thoughts on “Losing is fun: What we can learn from Dwarf Fortress”

  1. Have you played it much Doug? It’s insanely complex. I don’t know how much of that complexity is interface (which is horrible really — I don’t mind rogue-a-like but it is pretty obscure to find how to do some things) but there’s a lot of genuine complexity beneath the surface (if you’ll forgive the accidental pun). In the end I was put off as it was too much of an investment of time — which considering I play RPGs with 100 hour time scales is saying something. I still dip in from time to time. I like the fact that, like minecraft, fans have elaborate in-game creations.

    Many sim-like games (e.g. the sims and sim-city) become dull after not too many repetitions as the underlying mechanics become obvious — you can almost “see” the equations it is working to. For this one it’s not so clear — but I’m not sure how much of that is genuine complexity and how much is simply “abundance” — that is, many many simple things interoperating — the system is not interesting but there is so much of it it takes an age to work that out. on the other hand, the interaction of simple entities can make complexity emerge. (Magic the gathering the card game is fascinating like that — the simple rules on each card can interact in unforseen ways in such a way that the creators have a “bugfix” team to prevent “broken” combinations of cards).

    1. No, I’ve not played it much. I play a fair amount of Minecraft, which seems very much what you get if you tried to make Dwarf Fortress slightly more visually appealing and slightly more forgiving of casual play, and a lot simpler. It’s still old-school graphics, impossible for a novice to pick up in game, and pretty complex.

      I know what you mean about the obviousness of the underlying mechanics. Minecraft has really pretty simple underlying mechanics in many ways – certainly compared to Dwarf Fortress – and the fun I get is more about building stuff out of those simple components. Civ (probably the game I’ve played most ever) can get really pretty dull when the mechanics wear through.

      My sense of Dwarf Fortress is that it’s a combination of depth of detail of simulation and interaction of entities that gives it whatever it is that people find rewarding.

      Don’t think I’ll be getting in to DF any time soon. Not least because Minecraft came out of Beta at the weekend and I have a whole new world to explore right now 🙂

    1. That’s a clear (and actually quite astonishing) example of how the player can combine simple components in DF (eg. pressure sensors) to produce something complex (ie. a Turing-complete programmable computer). However, this can also be done in other games, including Minecraft. The true complexity of DF emerges from having several hundred AIs, each with a randomised initial state, operating independently or semi-independently of the player in a world simulated to an unprecedented level of natural and social detail. Never mind fluid mechanics: this is a game in the process of simulating an economic system, a justice system, a feudal hierarchy, migration, warfare, argriculture, ecosystems, visual art practices, and even literature (in the forthcoming version, it seems that historical figures spontaneously write biographies and autobiographies).

      The DF player community has put an enormous amount of collective effort into discovering the underlying rules of the simulation (partly through experimentation, and partly through examination of game files). However, you can see from Tarn Adams the developer’s blog that even he is frequently surprised by the interactions of those rules (scroll down the page here: http://www.bay12games.com/dwarves/)

      It’s object oriented programming gone mad, really – but that’s the beauty of it. It’s as if Tarn Adams took the functional programmer’s point that mutable state makes programs hard to reason about, and said: ‘But don’t you think that’s kinda cool?’

  2. Why is the grammar so strange in this article? Is english not your first language? The entire thing read like an outline for an essay and not an actual coherent organization of thoughts.

    Not to say that the core of the article isn’t good. I’m a huge dwarf fortress fan and there’s a lot that can be learned from the motto, “losing is fun”. Losing is about learning, and trying new things.

    Also, some of the things you said are just basically wrong. I don’t think you have a full understanding of the game.

    For people just getting in to the game, be sure to download the Lazy Newb Pack: http://www.bay12forums.com/smf/index.php?topic=59026.1095 although you might want to try an older, more stable version of the game once you get the hang of it. Personally I’ve been playing 28.181.40d

    Nihilizo needs alcohol to get through the working day.

  3. Thanks for blogging my talk, Doug – and I’m so glad you found it interesting. I hadn’t realised quite how enthusiastic I was about Dwarf Fortress until I started speaking!

Comments are closed.

%d bloggers like this: