Designing multiplayer puzzles

Let it be said: multiplayer puzzles are hard. Back at my first post I discuss how a puzzle where two players push buttons simultaneously is changed into a natural action.

However, sadly, such a setup is often presented as a puzzle. Surely, multiplayer puzzles can be more interesting than "everyone perform action X in multiple locations". (To be technical, I'll call them symmetrical puzzles.)

The Legend of Zelda: Four Swords Adventure presents some excellent multiplayer puzzle design. A brief overview: the style is 2D action-adventure, like Zelda in the SNES days; there are four characters or "Links", each a different color; each can be controlled by a different player. I find three major principles:

1. Asymmetrical puzzles.
This is similar to the simplistic puzzle of requiring all players to find places and push a button, but in this case different players need to perform different actions (example: one player holds open a section of wall while another shoots through it with an arrow).

2. Uniqueness.
The Links each can carry only one item, so the game has player-selected temporary uniqueness. One Link may be carrying the hammer and another may be carrying the feather; with both required to pass a certain puzzle, multiple players are required.

Note that uniqueness can be intrinsic (that is, permanent uniqueness), in the same way different characters in an RPG have different "powers" that make up a team. In Four Swords this is done in a fairly simplistic manner by matching certain devices with certain colors (so only the player of the right color can use them).

3. Dependency.
Uniqueness can create a condition where two Links are cojoined in proximity; to be specific, sometimes one Link needs to carry another. (For example, the Link with the feather may need to carry the Link with the hammer to get to where the rock should be destroyed.) These moments create a greater sense of teamwork (since if one Link falls, the other does as well).

Missing history

Quiz time: what game is being referred to in these quotes?

In essence, the animals would do to each other anything that they could do to or with you. So we would constantly have animals interacting in ways that had never been progammed or envisioned.

Also . . . you could interact with the animals in ways we'd never thought of. So people would constantly be writing to us telling us they'd done things that we never thought of, and didn't realize the game was capable of.

STALKER: Shadow of Chernobyl, perhaps? Or possibly The Elder Scrolls IV: Oblivion?

Nuh-uh. These quotes are from Veronika Megler in an interview on the site L'avventura è l'avventura.

They are referring to The Hobbit, first published in 1982.

It wasn't a poor seller (over half a million copies sold in Europe). However, it seems to have fallen out of the gamer consciousness.

Certainly, if one pokes about at the various histories, Beam Software (the developer) and Melbourne House (the publisher) make it in. However, most details I've seen (on those companies specifically and in general) tend to be on the companies themselves, rather than innovations in game mechanics. There's a lack of material on the actual content of games, so a student looking for a particular element needs to start from scratch; there's an intimidating number of works to plow through if someone is searching for a mechanic rather than a plot theme.

I find a real need for the sort of history work done with art and music history, with details about content that go past "in the old days, there were more mazes than there are now" so a future scholar can pick out that obscure game from 1980s that advances his or her point.

Telling ambiguity

So, as Sean Barrett points out in my last post, I botched considering the possibilities of TELL in the ASK/TELL system to break out of conversation as a mere series of questions.

Adding TELL makes conversation a series of nouns, but it's an improvement. However, there's a deeper problem with the verb, and it's the ambiguity of, say,

>TELL LESTER ABOUT THE KNIFE
What are you saying to Lester about the knife, exactly? That it exists? That it was found as evidence in the bedroom? That his fingerprints are on it? That you claim his fingerprints are on it, even though they aren't?

or

>TELL HENRY ABOUT HENRY
Is it supposed to be a compliment or an insult?

or

>TELL FRANK ABOUT THE HOUSE
Is this about how much money it cost, or how good it looks, or the location?

With ASK, assuming the player meant to say "tell me everything you know about topic X" doesn't cause many problems. The limit of information given is defined by the NPC. TELL is trickier to parse; since the information is inside the PC it should theoretically be the player specifying where to begin and stop. The player briefly loses control of the PC (moreso than with ASK, at least).

Sometimes context makes it obvious what the PC should say, but if the author means to allow a great deal of TELLling, there's going to be nonobvious cases as well.

Conversational cutscenes

A long-held charge against conversation menus is that they are modified cutscenes of sorts -- infodumps where the player will pick every choice given to avoid missing anything. Take the following, for example, from Andrew Plotkin's review of The Longest Journey:

And that means each conversation is a cut scene. It is not interactive. I can't put it more clearly than that. You sit back and listen to the pre-scripted dialogue, occasionally clicking a menu option to hear the next paragraph. You could skip some options, or leave early, of course -- but why bother? You're just going to come back later and listen to the rest. It's the shallowest kind of interactivity.

However, I contend that the ASK/TELL system can contain the same problem -- only even worse.

Consider the average static NPC with information. The standard behavior with ASK is to try every reasonable term that comes to mind, with the difference that a.) everything the PC says is a question, so there's not a great deal of dialogue variety from that end and b.) half the responses are "I don't know anything about that."

Now, one might argue ASK/TELL is superior in it requires more ratiocination (and it does) but it also contains the two flaws I just mentioned. It is a tradeoff. In a mystery, I'd say the thinking process of ASK/TELL is more important than the possibility of hitting unknown topics; in a game where the characterization of the PC is vital, the story benefits from expanded dialogue options (so Guybrush Threepwood from the Monkey Island games always has a selection of jokes to choose from).

I'd say the problem in general is not with the system chosen, but the static infodumping NPC in general.

There are alternatives:

* Galatea is interesting in it has a "mood maze" where the same question may receive different responses depending on what occurred before.

* Other NPCs react relatively dynamically to events and will respond differently to the same question based simply on situation.

* Dialogue options can be mutually exclusive, and can be actions that affect the plot -- the choice of gilded words versus an insult, the choice to lie rather than tell the truth.

I'll make a deeper examination of these tactics in the future. Any of them can reduce the cutscene effect, with both dialogue menus and ASK/TELL systems.

---

As a postscript, let me also add: even with an static NPC's dialogue menu, the experience isn't quite the same as a cutscene. A cutscene essentially makes the game a hybrid -- whether with a novel as in text games (like The Legend Lives!) or with movies as in graphical works (like the Xenosaga series). The mere act of choosing what to say next serves to break the information in chunks, and maintains an illusion of player control of the PC (however rudimentary).

Fatalism

Zankage isn't the answer to this one.
-- Carl Muckenhoupt, The Gostak

Samuel T. Denton's Endgame, the winner of the C32 Contest from last year, had a particular bit that impressed me. I'm going to reveal a mild spoiler, so you may want to go play the game first (it is short, don't worry).

Now, did you try using the ring while wearing it?

Did you try putting the ring back on and using it again?

10 times?

There's quite a lot of IF out there that would simply say 'No, that'd be too dangerous' or 'Not while it's on your hand!' and forget about it. Others might let you try the ring-on-hand once and then disallow any future tries.

Endgame lets you be foolish, 10 times in a row.

Authors these days are very protective of their players. Many a cliff I have attempted to jump off only to be chided with the message "That's too dangerous." Look, I know it's dangerous. Can't you let me do it anyway, and maybe give an amusing death message on the way?

This protective tendency seems to have grown from the fatality-at-every-turn days of old. A mere fingerslip in King's Quest 1 could send the player to their doom.

Later, games became more forgiving and gave adequate warning when death was about to occur. This is a good thing. However, what if the player does something that is obviously fatalistic? Shouldn't their wish be fulfilled?

Similar logic can be applied to what one might call venting actions. Most IF players know, innately, that violence is (nearly always) not the solution to a particular problem. However, being able to hit the door one is frustrated with is a form of catharsis, even if it results in player death via stubbed toe.

Similar things can be said about licking the floor, or jumping up and down. If the response is amusing enough, or at least doesn't merely chide the player, it's encouragement that Yes, everything is ok. My game is well coded and anticipates actions well. The puzzle you are stuck on has a reasonable solution. You are not stuck merely because the verb you tried is not exactly the one I had in mind even though it is equivalent.

So, authors: let us zank once in a while. It'll make us feel better. Maybe we'll last, oh, 60 seconds more before we reach for the hints, at least.

Soup cans

Cliff Johnson's game The Fool's Errand contains puzzles -- word searches, jigsaw puzzles, and so forth -- that are separate from any narrative.

7th Guest is game infamous for a puzzle involving soup cans; enough so that the term "soup cans" has entered the language of IF theory. However, I have never found this specific criticism entirely fair, because in a way, 7th Guest is exactly like The Fool's Errand. The puzzles only very slightly connect with the narrative. It is a "puzzle game" rather than an "adventure game".

Still, there's a tangible difference when playing the two in sequence, and I believe it's this: the mechanisms of the 7th Guest puzzles occupy tangible positions in the world space. It's pretending that the soup cans really exist, whereas in The Fool's Errand the word searches occupy a sort of metaphorical space.

I would also say J. Robinson Wheeler's Colours (which as he describes, was written a "little bit in the manner of Cliff Johnson's works") suffers the same problem from a slightly different angle. There's no narrative to be contrived (and therefore no mimesis to break) but the world space is the same as the puzzle space. As soon as the world space is occupied, there seems to be an impulse on the part of players to explain why the world space is configured as it is.

However, I don't see any reason why a work of IF couldn't mimic the Cliff Johnson effect precisely and come out well. Navigation would be like normal IF; the puzzles (triggered by, say, touching certain objects) would be presented in a different interface from the rest of the work. Touching the photograph might get an acrostic, while touching the doorknob might activate a jigsaw puzzle.

Categories of interactive fiction

Find a pack of cards. Take a card at random. Replace, shuffle well, draw again. If you get the Queen of Spades twice in a row, you are born dead. Go to 0.
-- Kim Newman, Life's Lottery

I have been noodling about with a visual model of the various sorts of interactive fiction, which I have here:

Visual model

Some examples of each:

"IF Games" (world state stored, high freedom of input): Zork, Curses, Pick Up the Phone Booth and Die

"Gamebooks" (world state stored, low freedom of input): the Lone Wolf series, the Fighting Fantasy series

"Hypertext" (world state not stored, low freedom of input): Patchwork Girl

"CYOA [Choose-Your-Own-Adventure]" (world state not stored, low freedom of input): the series by the same name, Life's Lottery

"Chatterbot" (world state not stored, high freedom of input): ELIZA

Caveat #1: The term "gamebooks" is sometimes generalized to include choose-your-own-adventure style books. I have separated it here to indicate that there is some sort of world storage state, generally referring to a player inventory and RPG-like statistics.

Caveat #2: I am using "chatterbot" here to refer only to "dumb" chatterbots with no memory beyond the previous command.

Caveat #3: These categories aren't exactly parallel. ELIZA has a great deal more freedom of input than any IF game; gamebooks tend not to store nearly as much detail as an IF game. It may help to think of the corners as directions rather than absolute points.

Some relatively interesting effects can be attained by adding more variables. For example, here's what happens if you care about whether or not the user is a character in the story or not:

Visual Model #2

The left square represents "the user is a character in the story" and the right square represents that "the user is not a character in the story".

The hypertext and CYOA categorize get separated, so that's rather pleasant.

"Expert model" refers to things like a taxonomy system that answers simple questions and routes down a branching tree (what color are the feathers on the bird? does it fly? etc.) Unfortunately, expert models tend to be in truth limited in choice (even though they have free entry) so the position is filled somewhat uncomfortably.

"Expert system with world model" refers to a system like SHRDLU which resembles a chatterbot but keeps track of world state. Nick Montfort cites it in his book Twisty Little Passages as an early form of IF, and by this model his statement is accurate (although it still does not get lumped together with "games", which makes logical sense).

"Advanced hypertext" is a word I invented because I was unable to come up with anything that met the criteria. It would require a hypertext that stores information about the world state; for example, perhaps if the user picks links related to death scenes across the entire hypertext become darker. There are some Japanese games that appear to come close like Radical Dreamers but I don't know the genre in detail enough to make an authoritative statement here.

"Interactive poetry" is another word I invented, and really only matches (at the moment) with Andrew Plotkin's work The Space Under the Window. (The title of this weblog also hints at a work in progress of my own which falls into this category.)

The blank spaces and invented words may seem a bit of a stretch, but it's really the blank spaces on the map that interest me most. Different variables like "RPG elements" and "graphics" have brought up things of interest. I'll cover these results in the future.

Natural action

My use of "structural elements" to name things that are actions but not really puzzles was a bit of a hack. I meant that those elements were working as structure, rather than something to "stump" the player, but since normal puzzles define structure as well, it's not the best term. (I also put in a bit of bias with my first two examples for having the structure require navigating somewhere else first; however, the CPR scene in Photopia doesn't follow this logic.)

I'm going to stick with natural actions for the moment, defining them as "actions that advance a narrative without requiring insight or logic."

I've also contemplated the problem I posed regarding the "easy" puzzles of Photopia: what makes them different from using a key on a lock? I think the hinge here is the puzzles involve fantasy elements, invented items with possibly no context in the player's experience. They're relatively typical Western fantasy elements, so readers with a nodding attachment to those tropes should have no issue, but it's quite plausable for a player to know nothing of Western fantasy.

Looking over my previous examples, then, I'd say natural actions either a.) Give direct instruction to the player, b.) are repetitions of a puzzle in identical contexts, so after the first time it is quite natural what to do, or c.) are connected enough to a player's background and context that no thought is required.

C seems like a relative case. Take the possibility someone has never seen a lock and key before. Using a key on the lock then *would* be a puzzle (never minding the fact they probably wouldn't know the word UNLOCK either). If this is taken as fact, then the "easy" puzzles in Photopia really *are* natural actions for some players, so for those people they are not puzzles at all (which explains why when the game was released there was confusion as to whether it was puzzleless or not).

Now, to the experimental IF question: could an (enjoyable) game be written entirely with natural action? There are, of course, actual examples (like Stephen Bond's Rameses) but they seem to enforce this through player inaction; pure navigation, or conversation, or a combination of both. Would it be possible to maintain natural action while allowing the player to be in the middle of, say, a spy thriller?

Structural elements vs. puzzles

In James Bond: Everything or Nothing there's a 2-player cooperative mission mode. Throughout the mission there are doors with two buttons; both buttons must be pressed at the same time for the corresponding door to open.

In some contexts this might be called a "puzzle". However, because the layout is consistent -- the buttons are always to the left and right of the door -- it's not a puzzle at all, but a structural element of the level. The doors force both players to be in the same location at once.

In interactive fiction, the equivalent sort of thing can be found with locks and keys. There may be a door with a corresponding key, and some difficulty involved in finding that key, but once the key is located unlocking the door represents a trivial puzzle. The actual matching of key to door is acts as a structural element (forcing the key to be found first) rather than a puzzle.

In Photopia, an early scene involves resuscitating a person. The player is given directions on what to do and a NPC intercedes if the player is simply passive. The events represent player action but do not represent a puzzle. The commands the player enters are part of the structural elements, but not a puzzle.

Later on (with the storytelling scenes), there are some "easy" puzzles. These are not merely structural -- the player is not told what to do, and some people have gotten stuck. The actions are strongly implied, but the leap to blatant obviousness isn't there.

The question I have is, at what point does something turn from structural element into puzzle? Clearly in Photopia's case the direct commands from an NPC managed it, but what about the more subtle ways? The structural aspect of the lock and key seems to derive from convention -- if there was only one IF game that used keys and locks, would that constitute a puzzle because of the lack of repetition? Intuitively, I'd say no -- I can't imagine a player of Dreamhold being stuck on the first locked door once they have a key -- but I have difficulty pinning down the exact reason why.

The interest to an experimental IF writer here would be to explore what limits there are in "puzzle-less" IF. Which sort of actions are allowed when avoiding puzzles, and which represent some sort of difficulty and are therefore off-limits?