The Zarfian Cruelty scale, revisited

I have written up some thoughts on the Zarfian Cruelty Scale, which has now been floating around for over twenty years. I first posted about it in a newsgroup which was (at the time) entirely focused on parser IF. It still gets invoked today, but I think that its assumptions haven’t kept up with the modern expansion of IF.


Interesting. I thought that from the beginning, the notion of cruel games are: “games where you are stuck or dead, or preventing from winning without knowing.” I learnt recently that in other genres that state is called “walking dead”.

anyway, that thing, I’m amused that the original definition has nothing to do for what I was thinking it was :slight_smile:

1 Like

Hm, I would have said that your phrasing matches my 1996 definition pretty well.


My general feeling is that medium-sized undo stacks have cut the legs out of the cruelty scale as a scale, because Merciful games and Nasty games don’t feel different when I can get myself back into a winnable state with undo. In both cases, I needn’t worry about getting locked out of victory. There’s now only one question: “can I be locked out of victory beyond what undo can fix?” Which is to say, “Is the game cruel?” This is a good question, but that’s no longer a scale.


Graham Nelson’s Player’s Bill of Rights is a great guide to help avoid creating cruel games. :man_teacher:


Said that, who knows there’s multiple undos in a row available? Only a handful of the hardcore players.

Even I, I simply forget about multiples undos.

So, IMHO I don’t think the cruelty scale should change deeply based on undo, also taking in mind not all interpreters behave the same.

1 Like

Nowadays, I think unless there is no saving available or some mechanism that thwarts it or undo, cruelty is more about structure and player notification. A non-cruel game should not let the player get into an unwinnable state and continue doomed play without letting them know. One move instadeath isn’t really cruel with undo, I’m talking letting the player progress to day 5 without a critical item only accessible on day one and no clue how to progress.


I’d say the cruelty scale still holds but game structures got even more complex where it’s always possible to win the game but find out you missed some optional objectives.

1 Like

I’m approaching this from the perspective of someone who spends a fair bit of time trawling the IFDB looking for new games to play. I think the most important question is, how does the Cruelty Scale (which IFDB calls the “Forgiveness Rating”) help a prospective player? Why would the player care, and what are the most common use cases for this information?

I suspect this is still largely accurate, even if a player doesn’t put it in those words. I also baselessly assert that calling it “Puzzle Forgiveness Rating” would more clearly indicate what players and authors mean by it 95% of the time.

Similarly, I think phrasing this instead as “designed for the puzzle world” resolves a lot of the tension here, as would the addition of an “NA” rating. I know this sort of discussion tends to lead to “Well just what is a puzzle, anyway?” arguments, but this doesn’t seem too controversial.

1 Like

The Zarfian Cruelty Scale isn’t necessarily about puzzle difficulty, it’s about wasting the player’s time - the game needs to at least inform the player if they are checkmated.

I wouldn’t think that’s cruelty though; optional branches not being followed is a feature with importance to interactivity and replayability! You’re right though, and that also applies to most game design: challenge without insurmountability.

This relates to zarf’s point about the player’s assumptions about the structure of the game. If the player expects multiple separate but equal branches - a structure often (not universally) identified with choice-based games - then it’s not “cruel” not to let them see everything on one play-through. If the player thinks there’s one ideal ending and anything different falls short of it, they may indeed find it “cruel” if they can’t reach it - especially given how obsessive text adventurers can be about getting a perfect score! (Thus Adventure is more or less Polite if you only want to finish the game, but Nasty to Cruel if you want all 350 points.)

By the way, I’m intentionally framing this as a question of the player’s expectations rather than the author’s intentions. They don’t have to be the same! I think that goes a long way towards explaining the wide ratings spread of certain IFComp entries.

1 Like

I understand what you’re saying, but I’m not sure I’d label optional content as “cruel” that isn’t required to finish or “win” the game. Is a game cruel if the player decides not to examine every item for every piece of text? Is 80 DAYS cruel because you can’t see the entire globe in a single run?

The scale does also presuppose the method of saving and undoing that you find in most existing parser systems (even though it’s theoretically separable from parsers), where you have to actively create save files and can make as many as you want–and where you can expect to be able to undo something like between one and eight turns. (Though that depends on the game and the interpreter! Used to be you could only ever undo once.)

There are a lot of Twines that let you undo as much as you want, though undoing a lot of times would be a pain–and most Twines don’t come with save systems at all. (The first Twine I thought of that was puzzle-heavy and extremely cruel, “Lost in Time” by Gerardo Adesso, does have a save system that’s like the parser system where you can save anytime and keep multiple slots.)

And of course there are lots of other choice systems, puzzly or not! For example, Lethophobia is a puzzly Storynexus game, which means there’s no undoing or reverting to save states. Which means it kind of has to be merciful. I don’t know if Hanon would consider his Final Girl to be a puzzle-based game.

Final Girl was definitely cruel, but designed random and replayable. (pours one out)

I really struggle with how to categorize my games on the Zarfian cruelty scale.

Tingalan is a “optimize your score” game like Sugarlawn, except it’s more of a “push your luck” game because the more you push the limits of the world, the more likely you are to die.

Six Silver Bullets requires the player to die, many times, to solve the game’s puzzles.

Skybreak largely presents content by way of an RNG

There are a couple games in this comp (including ones I really like!) where you can get the game almost completely right, but reach a point where

  1. you can’t complete the game without starting over, and
  2. Replaying provides no new content or insights until you catch up to where you were

This is what I consider “cruel”. Skybreaker and Choice of Games aren’t cruel in this sense because replaying gives you a substantially different experience.