Permadeath considered harmful

Typical video games take 10–100 hours to complete; when something bad happens, e.g., your character is killed, you can just reverse the bad thing by restoring a previously saved game. Permadeath means that this is not allowed: you have to play again from the beginning. This post argues that (the emphasis on) permadeath is the root of all problems in the traditional roguelike community.

So why is permadeath historically important in roguelikes? I would say this is because roguelikes have evolved from Role-Playing Games. It makes a bit of sense in the context of role playing— it helps immersion. However, roguelikes have largely evolved away from other CRPGs (computer role-playing games). CRPGs focus on telling a specific story. But telling the same story again and again would be boring, and making the story different every run would not be possible with the 90’s technology. So instead, the story in roguelikes is rudimentary, to not get in the way. Instead, roguelikes focus on the gameplay — this can be made different and fun every run. Actually, they feature a very specific kind of gameplay…

Now, roguelikes are actually better seen as closer to strategy games than CRPGs. I have realized this myself when I was asked in a poll about my favorite gaming genre, with options “strategy” and “RPG” (no “roguelike” option), and I picked strategy, because I tend to like non-roguelike strategy games while dislike non-roguelike CRPGs. Permadeath, randomness, and reduced story make much sense in strategy games, in fact the same could be said about games such as Civilization or X-COM (it is clear that permadeath is the intended way to play these games). I still prefer roguelikes though: the number of interesting decisions per second seems to be higher than in such strategy games. This is because roguelikes focus on a single character, eliminating the need for micromanagement and context switching, and thus allowing very fast gameplay.

This change of focus was also visible in the roguelike community; especially to note is the Seven Day Roguelike Challenge (7DRL): by nature, the 7DRL games cut some features, and they did show that you can still have fun roguelikes without the RPG elements. The Berlin Interpretation (2008), the most serious attempt at defining roguelikes, did not even mention that roguelikes are a subgenre of RPGs.

If you look at early definitions of roguelikes, permadeath is often not mentioned in them. See, for example, the discussions which established roguelike as a genre, early version of the roguelike FAQ, later version of the roguelike FAQ, NetHack guidebook section 2, or people considering Diablo (controversially) or Castle of the Winds (generally) to be roguelikes.

However, Berlin Interpretation included permadeath as a core roguelike feature. Does that mean that permadeath is really that important? No. Roguelike players have some intuitive feel what a roguelike is, but they cannot really explain it —instead, they compare it to some game that they feel is similar but not a roguelike, and mention the most visible difference. So the roguelikes were originally defined as “free games using character graphics”, a combination of two very visible red herrings which have nothing to do with gameplay. I consider permadeath to be another red herring, one of many red herrings included in the Berlin Interpretation. All these three features (being free, using ASCII graphics, and permadeath) are historically very important for roguelikes, but they are still red herrings.

I would argue that emphasizing permadeath was one of main reasons why roguelikes were a niche genre, despite their unique and amazing gameplay. Just like in other strategy subgenres (e.g. games like Civilization or X-COM), permadeath is great if you love the game so much that you want to master it. Roguelikes were non-commercial efforts, so people would learn about them only from existing roguelike fans. Who would naturally be serious players and thus talk about how great permadeath is. But listeners would only know typical CRPGs and not understand why this would even be fun, hence this amazing genre remained niche. This was better understood by commercial developers who have created games inspired by roguelikes. Permadeath was one of the first things that went away. Most of them made permadeath optional, or (like Diablo I) not even mentioned. And Diablo I got very popular.

Later, the focus on permadeath did hit roguelikes again. You see, they were actually not the only genre with permadeath. We forgot about arcade games, these were popular in the 80’s, but who played them anymore? Derek Yu (Spelunky) has revived this genre, with a great effect, by including the lessons learned from roguelikes, particularly the random generation. People realized that permadeath can actually be fun — the runs were short, so you do not actually lose anything of value because of permadeath. And since Derek Yu mentioned his roguelike inspirations, people looked what roguelikes were. And what did the Berlin Interpretation say? Random environment generation and Permadeath — oh yeah, these things are great. I play the game again and again, and the game is fun every time. Turn-based — WTF, why should I care about that? I want a game that I play again and again and is fun every time, why is turn-based even relevant? It would make more sense to classify roguelikes according to some specialization of permadeath (e.g., whether your failed runs unlock new powers or not). As a result, all the modern arcade games have started being called roguelikes, which greatly displeased the original roguelike community, who felt that these games did not have much in common with what they felt that roguelikes were. That is, their unique amazing gameplay.

So two big problems of the traditional roguelike genre seem to be related to emphasizing permadeath. But there is still more to say.

• You like challenging games? No, you just like the gameplay in some games so much that you want to challenge yourself. (At least this is true for me: I play roguelikes which are challenging, but there are other challenging games which I am not interested in; if I play these games, I tend to get bored after some time, and just cheat to see the rest of the story.) If you are a game designer, tell the players how the game was intended to be played, but let them decide themselves how much challenge they want. In roguelikes, this is done by making permadeath optional.
• I have seen people say “if you play a roguelike without permadeath, it is just another RPG”. I disagree with that. Roguelikes play very different than other CRPGs, due to their focus on interesting gameplay and replayability. After playing a CRPG, you are not likely to want to play again. In a roguelike, you may play it again… possibly, this time with permadeath. It took me some time until I learned to enjoy permadeath. I played DCSS only with permadeath, and the feeling of eventually winning it was great, unmatched by anything else in gaming, especially knowing that it was exactly the same game which half a year earlier I would likely lose on the first level. Today, I try new roguelikes with permadeath at first, if I think they are worth it, I win them this way, otherwise I either stop playing them or disable permadeath. The ones where I disable permadeath still have better gameplay than other CRPGs, I just prefer to master DCSS instead.
• I have seen people consider Diablo very different to roguelikes due to the lack of permadeath. But in fact, it is a very small difference! First, many prominent people in the roguelike community around 2005 considered Diablo to be a roguelike; this interpretation seems to become less popular later, due to the influence of the Berlin Interpretation and the arcade confusion. Second, Diablo was actually intended to be played permadeath; the developers were aware that a permadeath game would not sell, so they did not enforce it. So the only difference between Diablo and a typical commercial roguelike (regarding permadeath) is that typical commercial roguelikes do suggest the player that permadeath is is intended, while Diablo does not. Diablo shares many of the crucial properties of roguelikes which separate them from CRPGs and make permadeath fun. This is actually why Diablo was not considered a RPG and its genre was called “action RPG” instead—but really, the naming “action roguelike” would be more accurate and IMO better for the identity of roguelikes in the long run.
• The last example shows that permadeath in roguelikes describes the roguelike culture more than the games themselves. This contrasts permadeath in arcade games (including arcade games with roguelike elements), where it is the core design principle, making it fun even for somewhat more casual players (short runs). Some games with roguelike elements do have longer runs — they seem more similar to major traditional roguelikes, but also strangely less popular.
• Focus on permadeath might reduce the enjoyment. For example, I have seen people recommend reading spoilers before playing ADOM (so that you are not unexpectedly killed due to incorrect assumptions). Is that the intended way of playing ADOM? Probably not, given that ADOM is a game with “mystery” in its name, you are supposed to discover the mysteries on your own. Obviously permadeath is intended too, but if you are not motivated enough to handle both, which one should you drop? I have played my first games of ADOM without reading spoilers and without permadeath, and I would recommend this way of playing. Experimentation is a cool thing in roguelikes, but unfortunately, permadeath discourages experimentation with advanced characters.
• Traditional roguelike fans are often accused of gatekeeping. Usually for wrong reasons (saying “X is not a roguelike because it is not turn-based” is not gatekeeping, because gatekeeping is about denying access to the community, not about the meaning of words). But there is a grain of truth in that: “you are not really playing a roguelike if you disable permadeath” is gatekeeping. There are good things about gatekeeping (roguelike communities were very nice compared to other gaming communities — trolls usually do not target niche audiences; and easier to find friends with similar tastes), but most roguelike fans seem to want more people to play them.
• Genre taxonomy should be simple and ideally close to a tree of subgenres (strategy → turn-based grid-based tactics → roguelike). Contrary to other TBT games, in roguelikes you play a single character. That specifies the short-term gameplay, and you do not need to mention features orthogonal to that, such as being free, using ASCII characters, or permadeath. Roguelikes are fun because you can play these games as fast as you want (which is hard to do in strategy games where you control an army or a squad), fun with RPG elements, fun with permadeath, fun when it all takes place on a single map, but such features should not define the genre, they are just typical characteristics of a good roguelike. Actually it is more specific (Desktop Dungeons, Decker, and Kerkerkruip do not have that cool gameplay which is unique to roguelikes, so I do not consider them roguelikes, although they are very close) but that becomes hard to define.