Concept: IF version of "Sleep is Death"

So, all this discussion about Cleverbot has got me thiking about “On the Other Side” (a nifty curio where you play the computer/parser in a text adventure), and that in turn about Sleep is Death.

And I’ve started idly wondering about an IF version of “Sleep is Death”, to wit, a 2-player IF story told in real-time. A player typing the commands, a writer writing responses and making the commands happen, and maybe yanking the rug from under the player’s feet.

So I started thinking about it. The most tedious chore is map-keeping and keep having to write the default responses, but what if the program could remember the default responses and keep track of the map?..

Sort of inspired by Inform 7, I started imagining a system where the “designer” would be able to, in his turn, select from some options: say default message, move to a new room, write response to command, write event, add to inventory… little more than this, designed to be quickly selected. The program would take care of map-keeping, so as to avoid keep having to re-type room descriptions: it would only prompt the designer about changing anything before completing the move, and maybe the option to change anything (permanently or one-off) in the room description. Objects wouldn’t really have to be implemented - if I write scenery and the player interacts with it, I need only write a response. Then the program could ask me simply, "make this the default response to “eat chair”? Make this the default response to “eat "? Make this the default response to " chair”? Then, evrey time the player wrote a command that matched a previously-defined default, the program would show the designer “Default msg: . Display?”, and if not simply write a new one, with the option to “Replace Default msg?”

There could even be a separate window where, for those simple one-response commands, were kept the command and the response. That way, if the player tried to “eat chairs”, the designer would have only to copy/paste the same text he’d written for “eat chair” (if, of course, it’s not already been set to a default “eat *” that happens to cover the plural as well as the singular).

At most, some items would need to be in inventory. So the program would keep track of inventory.

Theoretically, with something this bare-bones, an IF version of “Sleep is Death” could be achieved. A two-way dynamic story/game/experience. An extra mile would be the ability to then output it all into source code for a system, but that’s prpbably overkill.

I dunno, I was just wondering. Does this spark anyone’s interest? Not that I could program it, you understand, just checking to see if anyone else would be interested by this.

EDIT - Overkill number 2 would be the player-side option of dynamic map, if so desired. If the designer could manipulate the map in a similar way to ADRIFT’s system, it could create awesome effects. Overkill number 3 would be the possibility of displaying images and sounds, but that’s tricker to do in a real-time system because we’d either be limited to things already on the net, or would have to take the time to actually upload them.

EDIT 2 - A scoring system would be optional but probably wouldn’t really fit the concept. Save/load, on the other hand, might - as long as it’s only to save and quit, and to load previous session, like a roguelike. “Undo” would probably be right out as well, but again, conceptually it probably wouldn’t be necessary - I don’t envision great elaborate cruel puzzles being done in this way. Though if necessary, I suppose it might be possible for the player to “Restart” back to what was defined as the beginning; if the program remembered everything that the designer had written, the designer wouldn’t even have to repeat himself.

But this is probably overkill number 4…

That would be One Cool Thing. Not that I could program it, either, but it sounds real cool to my mind – or ears or taste or whatever.

Ok, this is frustrating.

Not the lack of response - something else.

Namely, I thought “Ok, people are rather fed up with ‘Here’s a great idea! I can’t execute it, but please someone do’ threads. But I really can’t do this. Unless… unless I try my hand at a one-computer proof of concept in GLulx?..”

And I can envision it all. I was thinking past hurdles such as “What if the game is suddenly using non-cardinal directions? How do I use that using the same engine I was using for regular directions, while retaining the full force of Inform’s map-navigating?” (answer: have an option as “Set as Direction”, and then inform the game that “starboard” is supposed to mean “east”, then translate every further instance of “starboard” internally into “east” before the “moving player” apparatus kicks into action)…

…and then I realised I can do this, I really can, but brother, I really just don’t have the time. It’s a commitment I can’t take on without shaving time off things that are much more important to me.

And in the meantime I’m getting frustrated because my head is ticking away at all the problems, finding ways to solve them (a table of default messages, using Inform’s “word/word2/word3” to allow for synonims; placeholder rooms and inventory items to be filled in and mapped as needed; the possibilty of setting default messages by specifying exactly what the command would be, then in the future searching the player’s command for matches - thus allowing, say, “look very closely at” as an impromptu verb that may be surprisingly useful in a game - giving the maximum result for the least effort), knowing I won’t have the time.

This is infuriating. It’s vexing. And, in a word, it severely blows.

I have to ask, have you ever RPed on IRC or a mud?

“Roleplayed”? No, not my cup of tea. But this would be as different from that as “Sleep is Death” is different from a MMORPG…

I don’t see it that way, since this looks like textual online Parsley, and that’s not a bad description of IRC and mud RP when there’s a game master involved (which is a lot different than a MMO).

Never having RPed in IRC or Mud I won’t argue, but I will say that a textual, online, 2-people Parsley isn’t a bad thing. Especially considering the amount of freedom there would be in the format.

When you say parsley, you mean Parsely, right?

To me, this looks like a cross between D&D and an IRC, too.

While from the software side it’s an interesting concept, I wonder at the playability: Wouldn’t the use of default responses be rather boring for the player, as opposed to a GM who can come up with witty and fresh responses everytime? Who gives extra hints when he feels the player is stuck? Who can quickly and readily understand synonyms or deduce the intention behind a convoluted question or a typo? I also think that the lag between typing a question and receiving the written answer would all in all slow down the game very much.

The promising aspects are of course the map, the inventory and the use of text blocks, which eg allows the GM to repeat textual descriptions correctly, especially in case when a precise description is needed. But add to this a combat and scoring system, and you have a tool for conventional RP GM’s, but then of course the question is, why would this need to be played over the web, rather than face-to-face?



I think the idea is the designer would build a bare-bones world model before play, with some initial scenario and things and their descriptions , plus some custom commands they’d likely provoke. (Presumably the authoring system would provide some built-in functionality, ala other IF systems.)

Bet the real benefit here relative to standard IF would be the ability to respond the off-the-wall, yet still plausible, commands that the author hadn’t anticipated beforehand, as described in the OP. Syzygy, the author could inject personal responses at any time – or they could just click the “auto-default” checkbox for certain things/places/actions and continue what they were working on.

And the real benefit here relative to a simple chat client is the automatic world-modeling. The parser can automatically handle responses to a lot of unexceptional actions – get, drop, wear, examine, go, any custom commands the author has specified and checked “auto-default” – freeing the author to build content in other parts of the world that the player is heading towards.

The default responses would only be a shortcut - everything else would still be possible. I mean, realistically, if the player types “examine me” every five turns, or some such, it would be nice for the designer not to always have to type the same thing. Unless they WANT to type something new. It would just be a convenience, everything else you mention would still be there.

I would also have thought that. But then I thought about the lag in “Sleep is Death” where more complex things have to be done (and have seen complaints to THAT). And I thought about “Cleverbot”, where there is a bit of a lag that bothered me at first but just not anymore. And I thought of conversations I’ve had in MSN, sometimes waiting for long chunks of text. Add in an automatic map-keeping so the designer doesn’t have to keep re-writing the same room descriptions, so that moving around is very quick; and add in some default responses; and finally add in the fact that for many things, it’s only a matter of the designer writing a couple of sentences back. I’m confident it wouldn’t be that painful.

If you live in a country like mine, because you CAN’T play it face to face except in small communities you have nothing else in common with. It’s hard enough for me to talk about IF to my friends, let alone something that takes as long to prepare and play as an average GM session. Not to mention that I myself just don’t have the TIME… I once tried to play System Shock 2 multiplayer. I played it for about 15 minutes. I loved those 15 minutes, but I didn’t have TIME for more, and realised I wouldn’t have in the foreseeable future.

Plus, some people just like to tell stories (I’m going from experience here, from my girlfriend). Programming a story into an IF is cool, but like writing it down, it lacks an audience, and some people thrive on the audience. This would be a quick way to tell a story - and what’s more, to have someone else participate it in. Two people telling a story, letting it grow.

That’s how “Sleep is Death” works, certainly, but it’s not quite how I imagined it. In “SiD” that’s necessary, you’re dealing with graphics and such. In here, I really thought it would all be “create as you go”, apart from what would already be created. The barer the bones are, the more plausible this system would be to even created…

…but since I already said I wouldn’t be able to create it myself it’s probably best to let the idea grow, here, outside my head, where maybe it can mature some more.

Yep. That would be SWEET! Bring down the illusion of talking to a computer, you’re actually communicating with a PERSON, capable of understanding your input and using it to shape the world! Or yank it from under your feet! This taps directly into one of the main attractionf of IF ever.

I think what you’re describing isn’t “Sleep Is Death IF.” SID is just a graphical environment for a traditional table-top (paper and pencil) RPGs, like D&D or GURPS. Indeed, the most common critique of SID is: “why would I buy and play SID, when I could just play a two-player RPG, on IRC or Skype or even in person?” Without graphics, that criticism would be fatal for SID.

I think what you’re describing is maybe something a little closer to Neverwinter Nights, where the Dungeon Master can design a full map with NPCs, and can also possess NPCs and modify the map on the fly.

But but but I still don’t get it. What’s the point of a NWN IF, a partially automated Parsely game? You can play a non-automated Parsely game right now; the tedium of “having to write the default responses” is overcome by using the spoken word, which is a lot faster than typing. Or, even better, you can play a free-form table-top RPG, like Dogs in the Vineyard, Amber, or even D&D, where you aren’t constrained by what a parser can understand.

I don’t see how you could make Parsely more fun by automating just part of it. You can make it more scalable (though not more fun) by automating all of it–by building a conventional IF game that hundreds of people can play asynchronously. But if you only automate part of a Parsely game, you’re incurring the cost of a human DM without any benefit in scalability or fun.

More generally: for years people have tried to build D&D with a fully automated Dungeon Master. Nothing comes close. You can make a great CRPG with no human DM, but it’s totally unlike a real DM. So, for years, people have tinkered with partially automating the Dungeon Master, providing software that the DM can use during play to automate what they do by hand.

I claim that there is no fertile ground in partial automation; it’s a barren desert. If you have a bad DM, you can’t make them any good with software automation. If you have a good DM, you can’t make them any better with software automation. If you have some bad DM software, it requires a good DM to make the software effective, at which point, the good DM should discard the software and just run a conventional game.

Software works when it does things a human DM can’t do by him/herself: graphics/sound; managing large numbers of players (MMO). You can also get a market success by eliminating the human DM entirely and embracing the constraints of what a computer can do in order to reach more players. But partial automation? Dead end.

Yeah, I can see this working as a sort of directing style for an RPG with IF-like elements (a focus on the evocative and informative power of objects; slow, exploratory pacing; interaction between characters done mostly by proxy) and, if you designed it right, it might be fun. But I really, really wouldn’t put a fake parser in there. (I actually did a pen-and-paper IF-like adventure game when I was young and didn’t know TADS existed yet. Didn’t work. Players who were happy to accept that computer interaction would be kind of formulaic and dumb wouldn’t accept that from a human GM).

I’ve played a pretty great deal of RPGs in an IRC-like environment (i.e. ifMUD) and it’s… different, but by no means worse. You think differently in text. Some of my very best RPG experiences have been in text with no face-to-face whatsoever. So I don’t buy that in-person RPGs with no typing required are an automatic advantage. That’s going to depend on the strengths and tastes of your players.

Anyway, some partial automation is absolutely useful there – to wit, dealing with dice rolls, which are slow and boring to figure out in meatspace. (I’ve not done complicated statwanky RPGs that way, but in theory I can see how automation might be even more useful in that situation.)

On the other hand, I could see it working in a way similar to interactive theatre, where automated code deals with props and sets, and actors/GM orchestrators deal with the high-level stuff: but then you’re looking at something much more like a MUD-as-performance.

I’ve also heard of people who use live-IF as a way of alphatesting games without having to stop the test session whenever the player runs into trouble: this is more like an augmented IF reading, with the player shouting out commands and the author entering them into the game (and responding either from game output or ad-lib, as appropriate). Can’t speak to how well it actually works, but I think some people have found it valuable.

I agree that rolling six dice and computing the sum is faster with software, but IMO rolling one or two dice at a time is more fun by hand. Software dice lack the (fake) feeling of control you can get from rolling your own dice. IRC games often use a dicebot just to rule out cheating, but that’s just one of the ways it’s not as fun.

Ironically, that’s often where the automation falls down the most, because the statwankery tends to toss in a lot of circumstantial bonuses and one-off advantages that are more tedious to explain to a computer than they are to compute by hand.

For example, a +4 to Strength yields +2 to damage on one-handed weapons, but +3 to damage on two-handed weapons. Explaining this to a computer requires an interface to differentiate between one-handed and two-handed weapons. But sometimes you have to hold your bastard sword in one hand while a rogue climbs down a rope you hold in your other hand. But then the cleric casts a Prayer spell while the bard sings a to Inspire Courage. At that point, the rules get pretty complicated, but explaining the situation to a computer is even more complicated.

But, if they’re all just text, why use code for the props and sets? The DM could just tell you what happens.

Yeah, I don’t really find the physical act of rolling dice inherently exciting, but obviously YMMV. (I’m pretty sure that the function of dice-rolling is different for different players. Some people really enjoy doing the maths.)

Mrm, I’m thinking about precisely the sort of RPG that I really don’t enjoy playing, so my judgement may be off here. I’m thinking about a situation where the GM just tickyboxes all situational conditions that apply, the code spits out a bonus, and nobody has to consult any tables. But possibly I’m overestimating how tidily procedural high-statwank RPG rules are.

But, if they’re all just text, why use code for the props and sets? The DM could just tell you what happens.
Well, normal GMs sometimes have handouts. Maps, in-game documents, that kind of thing. There are some kinds of information which are not well-delivered by off-the-cuff speech. (Of course, a lot of RPGs are specifically designed to not rely on that kind of information, just as IF is generally designed to avoid requiring information that’s awkward to deliver in pure text). Pre-written text is absolutely not the same thing as extemporaneous speech (or extemp writing, if you’re in an IRC-like environment). Different strengths, different weaknesses.

And practically speaking, in a moderately large group the GM can’t pay attention to everybody at once; having the basic-ass stuff covered would allow some PCs to have things to poke at while the GM’s attention was focused on other PCs.

I basically agree that this would involve a disproportionate amount of prep-work, unless you were planning on running the same module many times with different groups (or passing it around between GMs, or making it more like an MMO), which, once again, is really not something I’ve ever been very keen on. So my argument is really more that something awesome along these lines would be possible, but it’d almost certainly be a labour-of-love hopeful monster rather than a bold new direction for games.

This project, which may be dead(?), seems to have at least some affinities with the concept Peter put forward: … tten-world

I link to the Kickstarter page because the project website is down. The project was funded nearly a year ago, and I think it went into beta testing a couple of months later.


FWIW I just checked the Written Word website and it was up, but I got a site down message on one of about ten pages. So they may just be having a bit of hosting trouble. The blog hasn’t updated for a few months, though (it sounds like it was in alpha, not beta).

Looks interesting!

Yeah, I was actually a backer way back when, but they seem to have gone MIA. Kind of annoying.

The best way to do this would be asynchronus play like in Draw Something or Words With Friends. The trick is making it fun for both people so one is not just a DM.

Wait…didn’t this already sort of exist as a drawing game where there was pre made art and event assets like rain and fire ? It played out like a collaborative picture book.

Perhaps if the “writer” also had rules like “here are three words that must be used in the next turn…” or “the other player will see this clip art so involve it some way…”

Edit: duh. Sleep is death is the game I was thinking of and mentioned.
.I did not remember that as the title. I was thinking it was an obscure novel about a game player perhaps.