Writing a "polite" game

Which response to a game-ruining decision do you prefer?
  • No decisions are game-ruining
  • Game over with an opportunity to undo
  • Gameplay continues, but player is notified
  • Other (please explain)

0 voters

background: The Zarfian Cruelty (or Forgiveness) Scale (eblong.com)

The game that I’m working on offers a few opportunities to ruin a playthrough. I’d like to handle these player mistakes in a polite (“You only need one save game, because if you do something fatally wrong, it’s blatantly obvious and you’ll know better than to save afterwards.”) way.

What are your preferences or thoughts on politeness? I’ve considered handling mistakes with a humorous game-ending message (suggesting an undo) immediately. An example of a game-ruining mistake would be using a single-use item on the wrong noun.

Though really, I suppose that approach is Merciful. Returning a strong message but permitting further play would be polite. In the case of the single-use item, warning the player before using it would be a polite thing to do. Asking the player to confirm their choice would be polite, too.

Anyway. Players, how do you prefer to experience these sorts of failures? Authors, what tactics have you employed to deal with busted games?


I’m old school; I’m fine with Cruel. Life is full of situations where you don’t have the right thing with you and don’t have a chance to go get it. Eliminating those situations in a text adventure seems very artificial to me.

In Zork II, you’re screwed if you

eat the green cake without taking the red cake.

Similarly, in Planetfall, you’re screwed if you

miniaturize without the laser.

Notifying the player of either of these things just wouldn’t have made any kind of sense.

Besides, deciding to notify the player of game-ruining mistakes can mean a lot of extra fiddly state-tracking. It might be hard to know for sure that the player has made the game-ruining mistake. Without getting too specific, I beta-tested a game where you would suddenly enter an endgame if you went to a certain room after solving the necessary puzzles. A particular object was crucial to the endgame, and if you didn’t have it, the game helpfully informed you you were missing something necessary and gave it to you. But I got that message even though I had left the object in the room!

I’ve never been against having to play through a game multiple times in order to solve everything. I play a text adventure for the joy of solving it. Simply reading through a story is for static fiction.


I like this approach. I don’t play cruel games anymore-- I just don’t have the oomph to find the right save file, figure out how far back I have to go, remember to save at every crucial moment (many of which you don’t know are crucial at the time), and keep doing it. I used to do it because that was just sort of the gig with IF, but now there are enough options that it has to be a really special game for me to suffer through the cruelty.

What you’re proposing here is quite reasonable. Unless I’m very into a game, a lot of the time if it kills me without warning, I’ll just quit. So I like automatic rewinds or the reassurance that I only need to save once. I think it comes down to trust. Are you, the author, playing fair or dirty? If you’re playing dirty, are you being honest with me about it? Or are you going to forgive me if I screw up?


My preference is generally for the game to end at the point that I’ve made the mistake, with UNDO prominently listed as a possibility which will take me back to before the game became unwinnable. Allowing further play past that point seems perverse, and if that action was so bad as to make the game unwinnable, presumably nothing that can happen afterwards would significantly impact the course of events so it’s possible to write an ending that’s more satisfying than just letting the player wander around forever without being able to progress.

If the situation that makes the game unwinnable can’t be fixed with a single-move undo - honestly I’d prefer if the game was amended so it did work that way (either changing things so one move is all you need, or allowing for stuff like that to just lock you out of an optimal, bonus ending, but making it so the player can still reach a minimally-satisfying one).

EDIT: I respect that some players like cruel games - more power to ‘em! But in the current ecosystem, where players generally expect games to be warmer and fuzzier - and they’re likely playing your game in a context where there are dozens more that are part of a Comp or festival that you’ve entered it in - cruelty is likely to lead to folks not playing your game at all, or just playing it with the walkthrough open in the next tab over. If that’s ok because your work is narrowly aimed at that particular old-school difficulty-favoring audience, of course there’s nothing wrong with that! But I’d definitely set my expectations for how the majority of folks are likely to respond rather low.


i get it, but Infocom games are really their own subgenre, aren’t they? I think people are willing to put up with more in an Infocom game than they are a modern one. People know what they’re in for when they fire up Zork I.

Really, though, I think Infocom games can often be nasty, tough, or even polite rather than cruel. Thinking of all the non-death failure states in Enchanter, for instance, most of them are detectable:

  • running out of food: hard to remedy, but obvious
  • misusing the kulcad spell: easy to remedy with a save, detectable
  • misusing the guncho spell: easy to remedy with a save, detectable
  • wasting pencil lead or eraser: easy to remedy with a save, obvious
  • getting your spellbook stolen: easy to remedy with a save, obvious
  • frotzing yourself: hard to remedy, not obvious (cruel)

Might Planetfall be Infocom’s cruelest game? If you’re already pretty sick and you left the bedistor, ray gun, etc in the first complex, you will likely need a very old save and have to repeat a lot of play. Then there’s the bent bar. Still, I think Zork I’s jeweled egg might be the meanest trick in the catalog.

But we talk about those cases because they are unusual, even for Infocom. Players who save regularly (that’s just how it was back then) usually have options.

I think this is the community consensus. Truthfully, I don’t have a lot of patience for zombification in contemporary games either. I have to love, and I mean love (I’m looking at you, Anchorhead) a game to replay major portions of it.

When playing comp/festival games, it simply isn’t feasible to do that kind of backtracking. Which, yes, is how people first encounter many titles, myself included.


I prefer games don’t have choices that “ruin” them. Failure should be as interesting as success. Or if I fail to a “game over” and there are mechanics I need to wrestle with, I want the game to be compelling enough that I have no angst about starting over. (e.g. Captain Verdeterre’s Plunder)

Narrative wise, I want either a bag of chips I can’t keep my hand out of, OR I want a meal I’m done with after great effort that is satisfactory and canon whether best or worst ending.


I chose “Game Over with the option to UNDO”

I would like to amend that with “Give me an awesomely written page-long death-scene before I type UNDO.”


Yes. Think of being devoured alive (with a gory description) by an Invisible Killer while roaming in ancient alien spaceships and then undoing, while still retaining a nice, new Achievement!

(Works only if you implemented another couple of solutions for the endgame where you would need that Invisible Killer, but it fled).

1 Like

When you say “interesting,” do you mean a compelling or entertaining dramatization of failure, or an equally implemented and treed out story branch? I do hope to write interesting game-ending messages, but I would only implement them to avoid implementing story branches that stray too far from the core narrative.

When I think about it, the least interesting failures tend to be the cruelest. Discovering, for instance, that you shouldn’t have solved the Royal Puzzle before visiting the aqueduct in Zork III is quite boring. There is no dramatic realization or insidious change in the world. Instead, the player eventually gives up.

That is the goal, I think. We’ll see!

That’s an important point. While winning is great, a lot of us also play these games to see all of the text. A well-written failure can be very satisfying!

So far as game endings go: I think it might be nice to have a couple of optional puzzles that add to the conclusion. There would be a set “victory” ending, but it would be possible to see more. There would be no bad endings in such a case,

just good, better, and best.


That’s what I’m saying. If you don’t feel like writing plot for choices you don’t want the player to make, why even allow those choices? That’s essentially false-agency.

It’s the author’s job to keep the story running smoothly. I’d much prefer a linear narrative with wide multiple lanes to the ending rather than a time-cave snarl of possibilities. The interesting thing to me is how the player gets to the end, not if. And that’s my personal preference; there’s nothing wrong with doing it another way.

I mean if the player can choose something that fails the core narrative, it should be it’s own interesting version of the story, or I’d prefer the author figure out a way to channel the player back on track. Simple death-ends from one choice are not interesting.

Instead of “and so you joined Darth Vader and became evil, the end…” give the player a side-branch where they quickly realize The Empire isn’t what it’s cracked up to be and let them take an escape pod back to the Resistance, lesson learned. (Even better if the story acknowledges in the denouement that they tried to go evil!) Even if the player indeed wants to remain full evil then maybe eventually give them an epilogue scroll, but they should be fully aware and not just “oh crap, I chose a thing the story didn’t want me to do nor account for…UNDO.”

This can be a place for parser snark - “Actually, you didn’t hit the red button that says SELF DESTRUCT because you know better than that. You’ve taken many safety classes that instruct you not to hit the red button. You’ve got to escape!” - or Rameses style character refusal - “Vader’s offer was tempting. But Starbiscuit knew deep in his heart if he betrayed the Resistance his name would forever be marked as traitorous, and his exploits would be the subject of many an off-color ballad sung in the cantina back on the planet of Seblooglu for generations to come. He couldn’t bear that legacy, even if he died, which was likely in this new career.”

Essentially, the author is in control. It’s one thing to give the player agency, it’s another to let them derail the story in a way you don’t want. Just because a story is “interactive” shouldn’t mean the player should be allowed every opportunity to run it aground.

And yes, that’s a style thing based on author intention and is a little more important in a choice-narrative. This doesn’t apply if you’re doing something farcical like Pick Up the Phone Booth and Die or a time-loop where the point is for the player to die and restart.

Of course we should all strive to avoid Zarfan-cruelty where the player can neglect to pick up the only ball-point pen in the game in the first scene that becomes critical to saving the world five hours later.

This goes along with my philosophy of “always be on the player’s side.” Even if the player thinks they have are solving a diabolical death trap and in huge danger, there’s nothing wrong with the author nudging things behind the scenes to keep the story going and giving the illusion of game-over danger instead of actually pulling the trigger.


I don’t have the patience these days for Cruel or Nasty games (on the Zarfian scale) that are designed to require replays or aggressive saves.

I also don’t have much sympathy for the “hey—that’s reality” argument. As someone I know said about literary fiction: “Reading about reality is like reading about my job on my day off.”

The problem I’ve run into with the Zarfian scale is what constitutes obviousness. I find myself torn with childproofing: Setting up artificial boundaries to prevent game-breaking actions.

What I’m struggling with is distinguishing between good-faith and not-so-good-faith actions—what’s obvious to some notion of an average player.

I suppose this puts me on the Tough rung of the scale. I’m okay with that, both as a player and an author.


Yes, both of these, exactly. I view IF (and all narrative) as a negotiated interaction between author and reader. Success is when both parties are working in good faith.


It’s interesting that you’d say that. To me, weightless choices (let’s say you do the good or bad thing, but a few turns later you arrive in the same space) don’t feel like agency to me. Those choices live or die by the strength of the writing, but the same might be true of a game-ending failstate. Since UNDO is a baked-in feature these days, the difference between a snarky INSTEAD check isn’t so different from a snarky END THE STORY mechanically. A writer ought to ask whether these experiences differ subjectively; one might be closer to the desired effect than the other.

I do agree with this. The main reason I am asking these questions is that I am trying to devise a gameplay loop that players will enjoy. On rare occasions, the player will acquire–lets call them components–as rewards for problem solving. They are precious, limited resources. I am trying to avoid the “gotcha” of letting the player waste the resource, yet still give the satisfaction of problem solving (and of acquiring precious resources!).

It would be interesting to have a playthrough where the player never uses ANY components (or uses them wrong), but I think it would be a big effort for a first game.

I think that this is what matters, regardless of the tactic taken. The author needs to deal with the player in good faith, and while players may not know what to expect of a game, they ought to have a sense of where the author is coming from.


I’m all about “simulating” agency and reducing the amount of branching and plot I need to do. This is my personal style, and it might drive some players nuts. The trick is not do do it during plot-critical choices. Here’s an early encounter in robotsexpartymurder:

Every single one of those leads to the same passage where the conversation continues, with perhaps a wee bit of text variation. My choice as the author was to make this character somewhat blithely dense and not pay attention, therefore not requiring a ton of conversation branching since I’m needing to introduce an important character and get across his whole deal. But it feels like there’s a ton of choices that can be made and the player isn’t just clicking to continue.

The point of all the choices is to let the player role-play their character whilst essentially just hitting “continue” on a rather linear conversation. It also allows me the opportunity to unload some exposition in the choices where it usually isn’t expected and not as obtrusive - the month is October, the player has potentially been to some dodgy parties, they’ve been here at least a year, there’s a reasonable amount of character amnesia with not being able to remember a party months ago, etc.

There are other baroque lists of choices that do subtly change some variables about the game. At certain times the player may not feel they have agency in these types of scenes, but there are gears turning backstage!

People who play a second time will realize “oh this goes kind of the same way each time” but I’m more concerned with their first play through. If they’re playing again, they like the game and I’ve got them.

(There are actual tricky ways to branch this conversation differently - specifically if the player brings in knowledge from a previous play-through…but that’s rather the whole gimmick of this game.)

1 Like

I can see real value in this approach to a game where the main character is elastic. The pre-Bethesda Fallout games give players lots of options for defining the PC’s personality, and I’m crazy about them. It’s a matter of intent!

As you may have anticipated, my beef is with choices that are presented as impactful but turn out to have no effect on gameplay or plot. It erodes trust between me and the author.

I’ve made a risky choice of having a strongly defined character. Risky because it may not be a character people want to play!

There will be a handful of moral choices that enhance the base ending. That may or may not encourage a second playthrough. I suppose I would need to make them very interesting.


I get it completely, but wouldn’t it be also true that a choice that sends the player off-the core narrative enough that the author isn’t interested in continuing it - a “game over” screen no matter how elaborately written - also really has no effect on gameplay or plot? The player didn’t change the story, they just ended it. Unless the player decides that’s their canon ending and is happy to not continue, I guess. It seems if you write a cool enough mid-game ending the player accepts, they won’t see all the work you put into Act 3.

Nothing wrong with that at all. In fact, a strongly-defined character with their own worldview offers you more leeway; it makes sense your strong character wouldn’t make a choice that branches so far off the the narrative you are writing that the game ends.

Basically I think we understand each other. I’m not saying non-canonical mid-game endings - especially if justified - are the worst thing. I suppose it is common for players in some genres (like VNs) to grind every ending if they’re significant enough. But that’s why VNs usually let you skip seen text to make a different choice…which is essentially equivalent to restoring or undo-ing.

The thing I find pointless is giving the player an opportunity to end the game for a non-plot-critical reason.

It’s like in old Sierra games that let you walk off a cliff with a stupid cartoon slide whistle and a splat, then popped up a “Welp, you’ve gone and died again! Too bad for you! [Restore][Restart][Undo]” dialog.

It’s like…why even let the player do that? Who thinks that’s fun? That to me is what erodes that trust with the author and breaks mimesis since the story stops cold for doing something the author allowed you to do. That’s not a choice at all. That’s a speed-bump that implies “You thought you had a choice here, but you really don’t. Quit or go back and make the right choice.”

1 Like

See, for parser games in particular, I feel the opposite. Part of the illusion of a parser game is that you can do anything you want and the world will respond appropriately. It is, of course, an illusion, but letting me do something stupid like DRINK POISON or COLLAPSE TUNNEL rather than saying “you shouldn’t do that” does make it that little bit stronger.

Being able to get things wrong also makes it feel like I have more agency in the story. If I need to collapse the tunnel from a distance to stop the rampaging alien dinosaurs from coming after me or drink the poison in front of the automatic medical robot to activate its emergency mode, being able to do the wrong thing and have the game respond to it (with a response of “you die”) reinforces the idea that I can interact with those items in various ways, and it’s up to me and my cleverness to figure out how to use that, rather than only being able to do the one specific author-intended thing that will progress the story. It’s again somewhat of an illusion—dying in a tunnel collapse is presumably not actually going to advance the story the author is trying to tell—but it’s an illusion that feels more fun to play.


I think there’s probably a distinction between narrative choice and puzzle-solving choice. Ideally, it is a fine distinction: I strongly dislike “brain teasers” that have no connection to the story or world of the game. Examples include the peg jump and Towers of Bozbar in Zork Zero. They could easily be stuck in any other game, and they would make no more or less sense there.

Puzzles like this can be very forgiving. Since their relationship to the game world is so flimsy, having a handy “reset” mechanism is not going to destroy an illusion of consequential decision-making.

Puzzles that have strong linkage to the game world operate differently. Facile reset buttons do not always make sense in the world of the game. Choices may be more fateful because the player stakes are usually higher than those raised up by an artificially imposed gate. In Enchanter, you only have one chance to survive the sacrifice, and this accurately reflects the nature of the scene.

Having recently replayed the Quest for Glory games: yes, those were bad.

Gag deaths are a different thing. Besides implementing too many of them, Sierra’s greatest crimes were that the deaths were usually not funny, and that the linkage to the game world was often very tenuous (lots of pop culture gags, etc). I’ve never seen anyone complaining about being able to AIMFIZ MERETZKY in Sorcerer, for instance.

This is key for me. I associate in-game agency with stakes. Failure states are one way to establish the stakes.


You are totally right, and I do agree with that, my worldview is choice which tends to be more story-focused rather than object-experimentation focused.

That is part of the reason I gravitated to choice. The author is fully in control of pacing and story flow - the player can’t really do anything I didn’t plan for (yes, of course you can throw the silverware into the ocean, player, but WHY? - yes I know, BECAUSE YOU CAN…and now I’ve got to make sure there’s a store where you can buy more silverware…)

Parser implementations excel at detailed touchy-feely movie-set/museum environments where you want to spend 100 turns in each room - read everything, poke and prod and pick up and drop and stand on the chairs and search under the couches and paw through waste bins and wave that magic wand you found four rooms back - never mind the Stormtroopers are on the way, you want to find all those paperclips in the desk drawers and read the history of each one!

I say that with love, I really like well-done parser environments. And having worked in both, I try to make much more interactive choice narratives that aren’t just [open the right door] [open the left door].

And testing is of course much harder for parser. If I so much as mention a door mat, the player is of course going to try to look under it, pick it up, throw it, eat it, set it on fire…and that has nothing to do with my story about door-to-door vacuum salesmen!

It can be frustrating writing a parser game because you’ve gotta account for people basically trolling your environment. It feels like a narrative fail if the player is so bored that they’re trying to light your scenery on fire instead of advancing the plot, or if the player engages with the plot correctly and doesn’t spend time exploring the richly detailed environment. Which is why (IMHO) parser excels in a scenario where the story has already happened and the player is exploring the aftermath of it.

That’s why I thought Wry from Spring Thing was so clever: the scenario of being tasked with waiting in one room…and seeing what trouble occurs when you get bored and start experimenting with the environment as parser players will.


Regarding the idea of catering to a broken game state by allowing the player alternative paths/actions to restore the situation: don’t bother. I used to spend a lot of time on elaborate alternative solutions, but people won’t notice, anyway. Instead they’ll state publicly that they managed to get the game into an unwinnable state and how could the author do this to them in this day and age. Now, you might think that I should be chuckling about their ineptitude and my own ingenuity, which they were not perceptive enough to anticipate, but that’s not why I write my games.
In conclusion, I think authors (and especially those of parser games) should surreptitiously railroad their players (and especially those players who like to think they can do anything), because they don’t deserve better. If you come up with a scene where the player is in for a hiding and getting all their hoarded treasures robbed off them, reroute every action they try before, during and after said hiding to yet more pain. I say whip 'em; they’ll learn to take it like a man.

Exactly! This goes beyond agency, stakes are part of what defines the story.

Sure, but those examples tend to result in an immediate failure mode. How should a parser game deal with actions that leave the game in an unwinnable but otherwise playable state?

I don’t think there’s a single answer to the question. I’m coming to see that those decisions may be at the heart of an IF design aesthetic.

1 Like