My dream game type as an author

I suppose we all want to write our own dream game, but of course if our standards are high then it will be very hard to satisfy that. I wonder (a) if the general principles I try to follow are similar to those of others, and (b) whether anyone has any recommendations for good games that also seem to aim at these principles? This is what I’m going for:

  • Pure text-based, i.e. nothing that wouldn’t work read aloud.
  • Flexible parsing with the world simulated in detail so that players have a great deal of control and choice (e.g. I won’t have any nouns in my text that the player can’t refer to in meaningful ways), but without this leading to two often associated problems: a steep learning curve; and potential to generate lots of text that is boring and plot-irrelevant.
  • A steep learning curve should be defeated by the game being easy, at least initially, and doing hand-holding and prompting in a way that is very supportive yet unobtrusive.
  • Not generating boring plot-irrelevant text while making everything feel referable to and real is just a difficult key skill that we are all developing, I suppose (tips welcome).

These are principles that I try to follow. The first point is pretty clear, but I think I could recommend three readings to help with the other points:

  1. For flexible parsing without the associated problems, I recommend The imitable process of Ryan Veeder, an excellent essay by an author who has made extraordinary and well-implemented games without too much clutter (his most recent works are
    Several Other Tales from Castle Balderstone
    and A Rope of Chalk , but if you want a simple old-fashioned puzzle quest you might like
    Crocodracula: The Beginning
    . His biggest games are Taco Fiction and Robin & Orchid ).

  2. For hand-holding and a gentle learning curve, I can’t recommend this essay by Lynnea Glasser enough. It concerns her game Coloratura , which won IFComp and was one of the best puzzle games of the 2010’s.

  3. For not generating boring, plot-irrelevant text, I’d recommend this quote by Adam Cadre:

A.C.: The player should always have a pretty compelling reason to type something other than QUIT. The fact that something came out of your head may by itself make it fascinating to you, but you’ll have to put in some other source of interest and motivation before players will find it so. And just giving the player a goal isn’t enough. The experience of achieving that goal should be rewarding. Every diegetic piece of text should have something to recommend it: a joke, a nice bit of imagery, a character insight, something.

You can see some examples of this in play in his games Photopia and Lock & Key .


Great essay, thanks so much for pointing it out. Strongly recommended by me now also. Looking forward to reading more of this stuff next week!


Work on my work-in-progress has been slow lately (on the computer that is, it’s still going on in my head). These are some of the principles I started out with:

  • Big world: The player should have a sense of real traveling, covering distance when walking around the map. I’m doing this by giving each movement command a short description (terrain, effort, sliding or stumbling,…) Also, The borders of the map should flow naturally from the geography, giving a sense of a greater world outside the game limits.

  • Detailed NPCs: If the PC encounters someone, I want that someone to be a person, not a cardboard cutout. Fleshed out conversations are a way to do this, but I find it harder to make less important NPCs feel real without deeply implementing them. (Description has to be spot on, as do the two or three non-commital responses I want to give them.)

  • I think I have a few good mechanical puzzles, but I’m still figuring out how to create a good NPC-based social puzzle.


I tend to do this too. It’s particularly important in a sequence I have where the player is supposed to be running away from pursuers. Feeling you on the NPC issue also! I’m scared of NPCs for exactly this reason.

I was thinking more about the Lynnea Glasser essay. I really like the techniques she suggests for allowing puzzles to get easier with time. On the other hand, this goes against something I have also included which is allowing the player to fail and die. I’m conflicted about this. I remember how annoying I found things like deadly grues back in the day. On the other hand, I’m finding it challenging to know how to provide a sense of danger in an action sequence without the possibility of real death. For example, I have a turn where because of what I hope are really obvious cues, the player must jump or climb or hop or go upwards any old way, or else perish. I’m hoping the generosity of the parsing and the abundance of cues mean the vast majority of players will twig and thus get a sense of achievement from avoiding the danger; if not they must replay the sequence. I hope that (similarly to how people will replay a platformer level until they don’t die) players might want to replay, but I’m not sure. Playtesting will tell, I suppose.

1 Like

I feel like this comes down pretty directly to “how compelling was the thing they were trying to do before they died?”


Yes, I think that’s fair. If you’re writing a game in which the PC’s life is in peril, it can undermine the sense of drama if the player realises the PC can’t actually die. Provided you make it very clear what the danger is and you give them enough turns to figure out how to avert the danger, I don’t think many players will have a problem with it. If it’s a very long game, perhaps remind the player in the ABOUT text that they need to save frequently.


It also depends on how much the players have to replay after they die. As long as you provide an UNDO command, I think you can get away with quite abrupt player deaths. Without an UNDO command, I’d be very wary of ever killing the player at all unless it was very close to the beginning of the game.


Depending on the authoring system, an easy method for the player to get back to where the problem action happened (either an auto-save option that can be restored that autosaves just before the bad action, or a parser command/choice to go to the trouble spot) is a viable substitute for (or augment to) rollback. Basically, recovering from a potentially game-ending mistake should be as effortless and graceful as possible. Only a limited subset of people want to play games ironman-style these days.


I’m finding it challenging to know how to provide a sense of danger in an action sequence without the possibility of real death.

For an example of a game that creates excitement and tension, I recommend Absence of Law by @mathbrush. The whole game is great, but there’s a particular part (minimal spoilers, it involves fighting robots on a train) that feels like an action sequence from a movie and still plays well as a turn-based text adventure puzzle. I’d credit it to equal parts good writing and good implementation.