Roguelike games aren’t narratives.  As heavily as they rely on procedural content generation, they cannot be.  And mostly, they don’t allow the fail-and-restore mechanic.  This essay is about how those two features of roguelike design complement each other.   The fact that these games are not narratives, in fact, makes fail-and-restore actively harmful.  I will explain why.

Usually, when we pick up a computer game that has anything like roleplaying elements, if it isn’t an MMO, what we are getting is a story.  When it is an MMO, it’s not even really a game because there’s no definite ending that is a win or loss, but that’s another topic.

When we are invited to manage or more importantly to identify with a particular character, we want that character to be a hero.  So  in these games the narrative of a story is already written, and the character is the hero in that story.  Our character’s part has been carefully considered, plotted, orchestrated, timed for dramatic impact, scripted, packaged, and delivered.

And, just for the record, whichever character you are being asked to manage and identify with, wins according to that narrative.  This is not absolutely necessary for a good story, but violations of the principle in narrative games are unheard of.  If Lord of The Rings came out as a standard CRPG, and the player were given the opportunity to play Boromir, there would be a version of the narrative in which Boromir does not die.  If the player were invited to play as Sauron, there would be a version of the narrative in which Sauron is not defeated.  If the player were invited to play as Gollum, there would be a version of the narrative in which Gollum survives.  This is an ironclad principle of the narrative game.  Whatever character the player is invited to identify  with, the narrative is a narrative in which that character achieves glory and fulfills his or her objectives.

So in narrative games, gameplay consists of fulfilling short-term tests of skill along the way; you don’t have to worry about long-term strategy or plot; that is all predetermined narrative elements, and the game program takes care of it for you.

In fact, the game takes care of those things so conscientiously that you are not allowed to have a better idea. If you find a better strategic idea – that is, if you realize that this character should be able to do something that would be a more effective or less wasteful way to reach the game’s overall goals than the way that the scriptwriters planned out for you – you won’t be allowed to do it.

You will find yourself stopped by an impassable waist-high wall, or an unclimbable mild slope, or an uncrossable river where a river makes no geographic sense, or a rope that is three feet too short and there are no longer ropes available in the game and you cannot tie knots.  Or you will be flatly unable to give an order to your loyal troopers other than the order which will lead to senseless waste of lives, ships and resources as dictated by the drama-building needs of the scripted narrative.    Or whatever.  The point is, the narrative is predetermined, and you are not allowed to affect it.  If you find that some idea better than the one the game’s writers had is actually achievable in the context of the game, it is not a “creative solution” or even a “good strategy.” Because it messes up the narrative, and these are narrative games, it is a “bug.”

And mostly that’s okay, because in that narrative, you are more or less guaranteed that your character wins.  That is, you know that if you just follow the script, no matter how many plot holes there are in it where your character could have done something more effective or sensible, no matter how badly hurt you get, no matter how poorly equipped you might become, no matter what odds you face, your character will eventually win.  There’s a contract between you and the game designers; you accept their plot holes and suspend disbelief, and in return they guarantee that eventually, if you get through the gameplay scenarios they’ve decorated their plot with, you will get to see your character win.  Because your gameplay is not allowed to have long-term effects on the narrative, the scope of mistakes you can make is limited to short-term in-scenario mistakes.  This is why you can always get through to the next part of the narrative by using the fail-and-restore mechanic; that is, if you fail, you just  restore from a recent (short-term) save file.

Because they rely heavily on procedurally generated elements and random determination of opponents and available equipment, roguelikes are a different game every time you play them.  You don’t  have to traverse the same maps and face the same opponents with the same equipment if you start over, and you don’t get the same choices to make again.  So a newly started game can be a fresh challenge, even if you’ve played many times before.  But for the same reason, roguelike games couldn’t enforce a predefined narrative if they wanted to.  So roguelike game designers are pretty careful to keep the random generation of challenges exciting, but within the range that a player pursuing good strategy could handle or escape.  The winning or loss of  roguelike games, like standard CRPG’s, depends on consistently good tactics in the generated short term scenarios.  But, unlike standard CRPG’s, it depends even more heavily on longer-term strategy decisions.

The most important difference between Narrative and Strategy is in what adversity means.  In a narrative, it’s nice for the hero to be down to a very few resources, facing horrifying odds, because that builds tension, making the turnaround and comeback on the way to victory all the sweeter.  So you’ve probably been conditioned by mainstream games to expect that survival means you have advanced the plot and are therefore winning, regardless of how ill-equipped, crippled, hurt, or too-late your character has become, because all that long-term stuff is managed by the scripted narrative.  But in a strategy game – say chess for example – if you reach midgame and your opponent is ahead by a queen and a knight, it doesn’t mean your eventual victory will be sweeter, it means that you screwed up and are losing.  In fact, if your opponent doesn’t specifically make a mistake, you will lose no matter what you do.  There is no narrative demanding that you win, or even demanding that you should get an opportunity to win from this position; there is only the cold hard fact of strategy and the consequences of you having messed up.

In a roguelike game there is no predetermined narrative arc.  Yes, it is possible at the beginning of the game for your character to be the winner, just as it is possible in chess.  But you are not guaranteed to be on any narrative path that leads to a win, because you don’t have a winning narrative that’s managed for you by the game.  You only have strategy, and you are responsible for your own strategic decisions.

Your character is not a predetermined hero; he (or she) is just some schmoe off the street who has the crazy idea to go take on a whole damn underworld that chews up and spits out a dozen people like him (or her) every night.  If you make a bad strategic decision the game doesn’t care.  Your character becomes one of the dozen, that night.  Maybe the next character you play will make a better decision.  Maybe not.  The game didn’t care the first time you made the mistake, and still won’t care if you make it again.

Get used to the idea that any particular character in a roguelike game has less life expectancy than a squirrel.  More importantly, get used to the idea that if you repeat losing strategies, it becomes overwhelmingly likely that you will also repeat losing results.  Unlike a narrative game, survival does not mean you are winning; if you’ve already made some unrecoverable combination of strategic errors, even (likely!) a combination you’ve made in previous games and survived for hours after making, your character will die, run out of time, or otherwise lose the game.  Survival only means he or she hasn’t died yet.

In roguelike games the situation is not usually as clear as it is in chess.  It is easy to recognize that you are down by a queen and a knight relative to your opponent. It is less easy to recognize the difference between strong and weak strategic positions in a roguelike game. Learning to recognize the strategic errors is, in fact, the most important part of learning how to win.  You may not realize that being down to your last two healing potions and no longer in possession of any source of fire resistance means you have poor odds of surviving a dragon fight your current location will not allow you to avoid.  That problem and its aftereffects, like the effects in chess of being down a knight and a queen, would remain no matter how many times you continued in games restored from a save made at that point.  Even if you do survive the dragon, you will be hurt a lot worse than you would be if you had made better strategic decisions earlier, so you’ll use up time and other game resources recovering from that, and another save will therefore tend to put you in an even worse situation relative to the game’s resource limits and threats still to come.

This is why most roguelikes don’t allow fail-and-restore. It doesn’t make the game easier the way it does in games that have narrative instead of strategy.  Instead, fail-and-restore tends to either amplify the effects of earlier strategic mistakes, making your situation ever worse, or help the player overcome strategic mistakes in the short term whose effects, if he continues to make those mistakes, will still prevent him from winning in the longer term.

Remember that learning to recognize strategic mistakes is crucial to learning how to win a roguelike game.  If fail-and-restore allows someone to overcome early instances of strategic mistakes, they’ll make later instances of the same mistakes whose cumulative effects  fail-and-restore can’t protect them from.  Given that they don’t actually make the game any easier, that makes them a net loss for both game design and player.  Fail-and-restore would mean wasting more time before you start a game you have a chance of winning.  More than that, its absence is crucial to helping you learn to recognize the difference between strong and weak strategic positions, and therefore learn to avoid strategic mistakes that are preventing you from winning.

Roguelike games are strategy rather than narrative games and tend to avoid having a fail-and-restore option.  Fail-and-restore, in fact, would actively weaken their game design, both by preventing the player from learning to recognize strategic errors and by encouraging the player to indefinitely waste more time on games that have already become impossible to win.  The more a game is strategic rather than narrative, the more fail-and-restore mechanics actively hinder players from winning.

Conversely, narrative games need fail-and-restore desperately; because a predetermined narrative would have to repeat in every new run of the game, and procedural content generation cannot be allowed to introduce any options that would change narrative elements or present the player with new strategic knowledge, there is nothing for the player to learn from repeating the early part of the game.  Without fail-and-restore, the players would get no closer to winning and besides that be bored by seeing again the same content they’d already seen.  Fail-and-restore allows them to progress and avoid boredom.  So the more a game is narrative rather than strategic, the more essential a fail-and-restore mechanic becomes.

And this is why best use of fail-and-restore mechanics in roguelike games (ie, DON’T) diverges from the best use of fail-and-restore in most other CRPG’s (ie, MUST).  Roguelikes are strategic and most other CRPG’s are narrative, and these different structures both benefit from opposite fail-and-restore mechanics.