Review of Dwarf Fortress

First I’d like to thank Leonard for taking the time to write two responses to my earlier post about Nethack. I mean to write a response to those responses, but right now I haven’t the brain, so instead I am going to talk about Dwarf Fortress, which an anonymous commenter on the previous post recommended.

Dwarf Fortress has a roguelike mode, but everyone says it’s not very good; the interesting part is dwarf mode, in which you direct a group of dwarves as they build the eponymous fortress within a mountain. I love city design games, and Dwarf Fortress does some things with the genre that I haven’t seen before. For instance, you’re building inside the mountain. You are also wanting to mine every last vein of ore and pocket of gemstones out of the mountain, because it’s a city of dwarves, and that’s what dwarves do. (It is fitting—albeit poorly implemented in the present version—that one of the ways you can destroy your fortress is by digging too deep into the mountain, where lie things better left buried.) You have to trade off greed against sensible internal organization. A giant cavern with pillars at the maximum spacing to avoid cave-ins is neither pleasant nor defensible.

However, the game is at present flawed on just about every level. (The next version of the game, due out Real Soon Now, appears to have a lot of these flaws fixed, but you review the game that you have, not the developer’s blog.) I started to write about the flaws, but I realized it was like criticizing Sirius Cybernetics Corporation products; I could write thousands of words about the superficial user-interface flaws and not be done, and there would still be the deep gameplay flaws! And besides, this isn’t Game Roundup, this is Zack Appeases the Voices in his Head by Mining Ideas for a Roguelike out of Whatever he can Find. So let’s incorporate by reference Leonard’s review of Dwarf Fortress, which says enough, and move on.

I do want to highlight one deep gameplay flaw, that goes right to the core of the design of this or any other resource-management game. You have to give the player enough to do, or it’s not a game, it’s a screensaver. But you also have to not give them too much to do, or the game becomes tedious and boring. DF is presently way over on the player has too much to do end of the spectrum. You have to micro-manage everything; the dwarves can’t be trusted to pick appropriate stone blocks to build buildings out of, nor to choose appropriate furniture for their bedrooms, nor even to understand that a request for bronze swords necessarily implies smelting down some copper and tin ore so you have bronze ingots to make the swords out of. I don’t hand-calculate the dependency list when I recompile programs, and I don’t see why I should have to do it in a game either.

So what ideas from DF would make sense in a roguelike context? In his first reply post, Leonard talked about some Japanese games that have an inner roguelike plus an outer game more in the CRPG walk-around-and-talk-to-people mold, and putting a plot tree there in order to avoid having to start it over again every time you die in the inner game (a la ADOM). The trouble with the Japanese games that do that is, it vitiates the start-over-from-scratch-on-every-death principle of roguelikes; the outer game is persistent over many character deaths and progressively makes the inner game easier. To some extent that’s unavoidable, or the outer game would be a pointless waste of time. It seems to me, however, that a middle way would be if the outer game were huge, persistent, and very slowly evolving. This is precisely the way it is in DF; both dwarf-mode and adventure-mode games have persistent effects on the huge world map, but only small ones.

DF has no overall plot, however. With an overall plot, I envision it working so that each inner game (exploration of one dungeon) only takes you one step forward in the overall plot; if you survive an inner game you start at an advantage in the next inner game, but no more than if you find a good bones file early in a dungeon. Contrariwise, if you die in an inner game it is a setback but not a huge setback; you pick up a new avatar and go on. Critically, progress in the overall plot needs to not run you out of things that a low-level character can do.

Another possible thing to borrow from DF is a tiny piece of the city-building game. In this context you would want to make the player do even less than the ideal state for DF itself; but suppose that, for instance, one thing you can be tasked to do is defend cities (not necessarily dwarf fortresses) from sieges, and while you can do this by getting out there with your @ and chopping up the invading letters, you can make it easier on yourself by advising the locals on good placement of walls, traps, catapults, etc.

And finally, DF lets you explore in adventure mode fortresses built and then overrun or abandoned in dwarf mode. The analogue in a game more focused on the roguelike component would be to use some sort of city simulator instead of the usual random map generator. One of the things that always bugged me about both CRPGs and off-the-shelf D&D modules was the way the monster-overrun cities you explored made no sense as, y’know, cities. (I’m looking at you, Pool of Radiance.) Of course, this is because a well-designed city is not a well-designed dungeon crawl, and vice versa; but it might be interesting to try to find a balance. I suspect it would work best with relatively short inner games.