Transcript of May 10, 2014 ifMUD Discussion on CYOA Structures

baf exclaims, “So!”
Zach says, “Choices!”
baf says, “The topic is supposedly ‘Structures in Choice-based Games’. Presumably ‘choice-based’ means CYOA-style.”
Zach says, “Right”
baf says, “And ‘structure’ means the properties of the graph of choices.”
baf says, “Like, how much it branches and whether branches rejoin and so forth.”
Zach says, “I was reading an article about the CYOA book series, and apparently there was an ideal number of endings that they shot for.”
zarf says, “the notion of branching is (of course) not so meaningful once you add state”
Zach asks, “Are Twine games considered CYOA?”
baf says, “I guess the obvious thing to point out about structure is that in a naively-constructed branching text, where each node leads to n distinct and unique other nodes, the number of nodes grows exponentially and rapidly becomes more than any author (or polynomial-time algorithm) can handle. So there has to be something preventing that from happening.”
klimas says (to Zach), “I think they’re definitely related.”
zarf says, “apparently we now get CYOA games that have iterative exploration and all that stuff that we think of as old-school IF”
eu says (to zarf), “Well, you can always split vertices into per-state vertices.”
zarf says (to eu), “sure, but that’s waffling”
baf says, “Twine games can be CYOA, and often are, but Twine is capable of more.”
aaronius says, “The useful distinction I’ve always found is whether the list of choices is easily enumerable on the screen, or if there is space in there for non-obvious actions. Although there are plenty of CYOA/Twine games that challenge that boundary.”
Zach says, “Yeah, with hyper-link games adding easter eggs or implementing hard to discover responses seems more challenging.”
katherine says, “but possible!”
katherine says, “hidden links, etc”
klimas says, “the classic example of a CYOA easter egg is a book called I think Inside UFO 54-40
klimas says, “the only way to get the ‘best’ ending in it was to disregard the choices and read through the book.”
aaronius says, “or security through obscurity. The non-obvious ending in howling dogs.”
eu asks (of klimas), “Was that the one mentioned in the recommended reading?”
Gunther says, “the ultimate choice”
klimas says (to eu), “possibly? I didn’t do all my homework :)”
eu goes to check.
eu says (to klimas), “Belatedly, yep, that’s the one mentioned at .”
Zach says, “It would be interesting to have certain choices appear only when links are clicked in a specific order.”
Zach says, “To reward careful players or discourage lawn-mowering”
Busta says, “That reminds me of the Grow puzzle games.”
eu says (to Zach), “Yeah.”
baf says, “There are certainly works that vary the choices according to the state produced by earlier choices.”
eu says, “Bee, IIRC, is fairly transparent about it.”
klimas says, “One example I can think of for this is a series of gamebooks about Sherlock Holmes, where there are branches like, ‘If you found clue E, turn to page 89′”
Zach says (to baf), “I think Choice of Kung-fu is the most complicated example of that I’ve played”
aaronius asks, “Zach, I haven’t read it, but how does Choice of Kung-fu use that technique?”
aaronius asks, “Different from other CoG titles?”
baf says, “Some of the trickier ones, like Trapped in Time from the last Comp, give you secret choices that aren’t even visible the first time through. Like, ‘The next time you see this character, you can choose to add 5 to the page number to reveal this clue’ sort of thing.”
baf says, “Of course computers make this sort of thing much simpler”
Zach says, “There are a number of choices, like picking your arch-enemy or your lover, that seem to stick with the player the entire game and result in a completely different experience.”
katherine says (to baf), “which is really clever if you can get the numbers to always add up”
Gunther says, “”Save the Date” uses this extensively”
Zach says, “Though I haven’t replayed it enough to know how differently the playthroughs are.”
aaronius asks, “But is it actually different, or just varying the name of the lover/enemy etc?”
baf says, “Hm. What Zach is describing sounds kind of like the Walking Dead model: avoiding the exponential explosion by having the story proceed along the same line regardless of player choices, but have the choices persist and produce small variations throughout that line.”
Zach says, “Hmm, I can’t remember.”
klimas says, “I like the idea of branching structures where the reader isn’t aware of being in a branch — stuff like what Zach is talking about, where the changes are pervasive.”
ChrisC says, “it’s entirely possible to change a story in large and small ways in response to certain variables being different”
Zach says, “But playing the game I *felt* like my choices really mattered.”
aaronius says, “Yeah, CoG calls that “delayed branching”… things continue the same regardless but there are opportunities later for the game to remember old choices and respond to them.”
ChrisC says, “From as small as changing a name, to different branching being possible.”
baf will remember that term
aaronius says, “And this also ties into the idea of cumulative stat changes… after enough small choices it changes a big state and opens up a different path. (Like, after you’ve done five evil things.)”
zarf says, “‘delayed branch’ sounds synonymous with ‘state’ when you get down to it”
klimas agrees with zarf — but it might be a more intuitive term for authors.
baf exclaims, “Simulationism!”
zarf says, “but the interesting distinction is between that cumulative stuff, which tracks what you’ve worked at consistently thorugh the game, versus single flags that come back later”
eu says (to zarf), “I think the distinction, if any, would be that “delayed branch” implies no obvious effect when the choice is made.”
ChrisC says, “Sure, but that is collapsing an integer value down to a boolean.”
aaronius says, “It’s a little more specific, I think… it’s an intentional design philosophy solution to the exponential branching problem.”
eu says, “Or that.”
ChrisC says, “I think it’s more interesting if things change gradually in response to gradual changes.”
Busta says, “delayed branching is hard to get right. It’s easy to frustrate the player by an arbitrary choice they made a long time prior.”
hubbard asks, “is the branch delayed or is it conditional? does it occur no mater what — in a delayed or deferred manner? or does it possible never arrive?”
baf says, “I guess my experience with hidden cumulative stuff is that it tends to come off as a little clumsy. Suddenly you’ve got enough ‘Character X dislikes you’ tokens for it to make a tangible difference: the result is that Character X’s attitude changes drastically without the player knowing why.”
ChrisC says, “Indicating this to the player is a difficult problem, yes.”
baf says, “Better to give the player a more dramatic way to earn hate.”
zarf says, “I would approach this more from intent rather than precise mechanics. Hidden long-term branches make sense if the game is about setting up a long-term plan that will all come together at the end”
djfletch says, “it seems to me that there’s a division where a game is either expecting players to play lots of times from the start, versus trying to give you a good fraction of the experience even if you only play once”
zarf says, “(on like the tenth play-through)”
zarf says (to djfletch), “right”
dfabulich says, “I did want to remark that the “By The Numbers” blog post (part of the recommended reading) is by far our best kept secret at CoG, because, for whatever reason, people really seem not to get it even after we explain it.”
Zach says (to zarf), “Huh, that reminds me of Lock and Key
eu says, “Lock and Key, in some sense, is a CYOA where you make (almost) all of the choices at once and then watch them play out.”
eu says, “(By which I meant that I agree with Zach.)”
ChrisC asks, “Hmm, how long is Lock and Key?”
eu says, “Not long, but it requires several plays through to get the winning ending.”
aaronius says, “I think it can also apply to personal stories. For instance Heroes Rise does a lot of figuring out what kind of superhero you’re going to be (morally speaking) and then towards the end of the game uses lots of small choices (converted into stats) to summarize your behavior into things like “You believe might makes right” or whatever.”
aaronius says, “Blue Lacuna did a lot of that too, but it was pretty opaque to players, I think.”
baf says, “Hm. Suddenly I want to make a choice-driven game with a Homestuck-like plot structure, with seemingly insignificant choices from early on suddenly reconnecting the story with horrific consequences long after you’ve forgotten about them.”
Busta says, “Oh man. That would frustrate me to no end.”
ChrisC says to aaronius, “I think Short’s approach of sprinking a ton of modifiers throughout helps convey those values, ala Alabaster.”
Zach says, “Yeah, the nice thing about Lock & Key is that it makes replays convenient. I’m not sure if that would work for story-heavy games.”
Zach says (to baf), “I would love that, but I could see it driving some people crazy.”
aaronius says, “Yeah, I think “showing the math” is critical. If something judges me “evil” at the end of the game, I want it to cite specific examples of my past behavior, rather than there being some invisible moment when a number passed a threshold.”
Zach says, “For optimal fun, combine that with permadeath.”
klimas says, “Fallen London puts all of the stats in the foreground — which to me leads to a kind of grind-y experience, but that may just be their individual treatment of it.”
Zach says, “At the end of Fallout, I liked that the game showed you the specific consequences of each side quest.”
katherine says, “a lot of games have this kind of stats-gathering behind the scenes with a debug command to reveal it”
ChrisC says, “Right, I don’t like the idea of displaying numbers explicitly in a textual medium.”
Zach says, “So that the ending can be a mixture of good and bad decisions.”
ChrisC says, “How it’s implemented internally shouldn’t be exposed to the player.”
eu says, “Yeah, you want the in-story reason for choice’s consequences, not the out-of-story reason.”
eu says, “*choices'”
Busta says, “It depends. If you take graphical games into account, people love seeing the numbers for rpgs and other stat-based games.”
klimas says, “Perhaps numbers aren’t best, but I feel like there should be some feedback to the player along the way. ‘The dark side has begun to pull you in.’ vs. ‘You have 3 dark side points.'”
aaronius says, “And RPGs are big on story, too.”
Busta says, “Displaying the numbers is a way to “gamify” your game, for lack of a better word, I think”
eu says, “Hm.”
djfletch says, “”The Play” didn’t tell you exactly that a certain stat value caused an outcome, but it did tell you the stats and there were a limited number of values they could have. So it was feasible to explore the space.”
ChrisC says, “Busta sure, but A) that’s an RPG B) that’s kind of orthogonal to playing a game based on narrative C) it’s kind of reductionist and not necessarily a behavior we’d want to encourage in players.”
katherine says, “the example that came to mind is Alter Ego
katherine says, “iirc the original version had the stats at the end of the game, and people didn’t like them, so they were shunted behind the scenes”
pollux asks (of ChrisC), “What do you mean by ‘reductionist’, exactly?”
klimas says (to katherine), “Alter Ego is a really good example to think about– lots of stats, lots of choices”
aaronius says, “I think djfletch’s point is key: it matters less whether the info is revealed through number or text, than that the player understands information is being revealed.”
Busta says (to ChrisC), “I don’t think there’s a right and wrong way to make a game, nor a right and wrong way for a player to play a game. Some players enjoy seeing the numbers, others don’t, and there’s room enough for both kinds of games.”
baf says, “(I was just looking for CYOA graph pictures at maga’s blog but none of the images are loading for me.)”
aaronius says, “If they don’t, consequences of actions seem arbitrary and the author may as well not have made any.”
djfletch says (to baf), “I had that problem before. Go to his new blog instead, they are on there.”
djfletch says, “
ChrisC says (to pollux), “The wrong term. I mean it can get to the extent of engaging with a game only to the extent of fiddling with a spreadsheet, purely numbers moving around, which RPGs by virtue of being so number-heavy and traditionally number-transparent can lend themselves to.”
ChrisC says (to Busta), “Right, sorry I didn’t mean to imply that.”
Busta exclaims (at ChrisC), “no worries!”
eu says, “Maybe the argument should be that the player shouldn’t end up playing the numbers game to the exclusion of experiencing the story.”
Busta says (to eu), “Nah. Some people really enjoy min/maxing games.”
katherine says, “there will always be people who will try to spade it out”
katherine says, “given a certain audience”
pollux says, “It probably depends most on whether the story is actually any good.”
ChrisC says (to eu), “Well, not even that, necessarily. Some games warrant it, too; take the kingdom/starship/country simulation genre.”
eu says, “Hm, well maybe I should clarify.”
pollux says (to ChrisC), “I think the point there was that, when it is the point of the story, it is not to the exclusion of the story.”
Busta says, “There’s a good article on player psycographics and why people play games. Lemme see if I can dig it up.”
eu thinks how to explain his thoughts.
baf says, “Hm. Looking at the Choice of Games doc, Im struck by: ‘There’s one tempting alternative that I will discuss briefly: instead of using stats, later chapters could refer back to particular choices from earlier chapters. Winning at chess in Chapter 8 might require the player to have chosen to practice chess back in Chapter 3, rather than depending on a high Cunning score.'”
baf says, “The conclusion there is ‘We do use that technique sometimes, but rarely, because if you use it too much, your story quickly becomes too complex to write.'”
djfletch says, “hmm, that sounds like it would be simpler to write, if anything”
baf says, “This strikes me because the ‘particular choices from earlier chapters’ model is exactly what we do at Telltale. We tried a bit of numerical stuff in Walking Dead season 1, and rejected *that* as too hard to author.”
Zach says, “That makes”
aaronius asks, “Fascinating baf– do you recall what was seen as hard about it? Authors ceding too much control or something?”
ChrisC asks (of baf), “Hmm, I think the reasoning is that if some choices are responded to explicitly rather than indirectly, it’s obscure to the player: why only some?”
Zach says, “And seems like it would be easier to work into the story.”
baf says, “I suspect that the difference is that ChoiceScript games tend to branch more, or have more varied branches.”
klimas says (to baf), “that’s what I’ve been pondering. Usually when you design branching structures, you’re drawing boxes and arrows — but I wonder if there are better ways to visualize or otherwise think about it, especially with stats in the mix.”
ChrisC says (to baf), “Whereas if they ALL matter, then you have to take all prior choices into account and you’re back at the combinatorial explosion.”
* rocketnia has joined the channel.
ChrisC says (to baf), “What I’ve found in playing CS games is they tend to branch *less*, at least the official COG ones.”
zarf says, “there is no good way to draw ‘branching’ structures with state”
zarf says, “they all suck. I wound up writing a dependency language to figure that stuff out”
ChrisC says (to zarf), “Hmm. I could see some ways mapping variables onto x- and y-axes. But combining that with a graph structure of choices… I have no idea how to even approach that.”
eu says (to ChrisC), “If you have few enough variables, perhaps.”
ChrisC says (to eu), “Yes. But I am more interested in variable-rich stories. :)”
eu nods.
Busta says, “you’d probably need a different axis for each state”
pollux says (to ChrisC), “I don’t see what’s wrong with a normal tree structure with conditional branches.”
baf says, “Numerical stats break the direct cause-effect connection.”
ChrisC says (to Busta), “Right, and when you surpass 3 dimensions you can’t even display it all at once on an interactive screen.”
baf says, “If you’re having trouble managing a complex set of causes and effects, that seems like the right choice.”
Busta says (to ChrisC), “correct”
eu says (to pollux), “I think the issue is the ability to understand the conditions at a glance.”
djfletch says, “you could have a little display in each box showing the minimum and maximum values that each stat could have at that point. But even that wouldn’t allow for “they can have 5 of this or 3 of this but not both”.”
baf says, “But if you want to fit the entire set of possible stories in your head so you can make sure it makes sense, numerical stats get in the way.”
pollux asks (of djfletch), “”x=5 ^ y=3″?”
klimas says, “I suppose Chris Crawford would say the fact that you can’t easily chart out a stats-based narrative is a virtue, not a problem.”
ChrisC says (to klimas), “It’s not the inchartability as a goal. It’s more of a side effect.”
Busta says, “it’s similar to developing a physics world model in a parser game. There will be a lot of things you only discover through play.”
dfabulich says, “whew, OK, I’m here.”
ChrisC says (to Busta), “Yes, a physical simulation is similar in some ways. As would be any sufficiently advanced system with a lot of moving parts and relationships between components.”
Busta says, “actually, that’s only with regards to printed text if the text varies by state. The paths should still be finite. Unless someone makes procedurally generated CYOA nodes.”
dfabulich says, “Specifically, in the “By the Numbers” post, I was contrasting the “hidden accumulation” model with a model in which there are a LOT of boolean variables”
dfabulich says, “For example, in one game an author submitted to us, there are literally hundreds of boolean variables tracking all manner of stuff throughout the game”
Busta says, “I’ll have to try By the Numbers.”
dfabulich says to Busta, “it’s just
ChrisC asks (of dfabulich), “Right, and enough boolean variables is essentially equivalent to stateless branching, right?”
klimas says, “(have to run! looking forward to reading the rest on recap)”
dfabulich says (to ChrisC), “Yes, exactly”
katherine says, “I still need to read that”
eu says, “Bye, klimas.”
pollux says, “Of course, the two models are mathematically equivalent.”
dfabulich says, “The combinatoric explosion of CYOA can only be managed by forgetting at least something of the choices the player made early on”
ChrisC asks (of dfabulich), “Has anyone written a CS game that bases results on more than choices? Like extensive use of number input boxes, or analyzing a name typed into a text box?”
dfabulich says, “In our model, the earlier choices effectively “vote” for what will happen later on in the story.”
ChrisC exclaims (at Klimas), “Bye!”
dfabulich says (to ChrisC), “I’m not sure if I’d call it “analyzing” but there are a few games that have put in puzzles where you have to type the right answer. The Race was nominated for an XYZZY in 2011 for one such puzzle”
dfabulich says, “ChoiceScript doesn’t have good facilities for coding with strings… it has no substring function, no arrays, no loop structure, etc.”
baf says, “I’ll note a contrast to the structure described in By the Numbers and the structure of print gamebooks: Both have lots of rejoining of paths, but By the Numbers presents it as a sort of great rejoining of all paths at the chapter boundaries, whereas gamebooks tend to just have lots of minor rejoinings.”
baf says, “But perhaps this is just a product of length.”
baf says, “A typical CYOA book being about the length of one chapter of a large ChoiceScript game.”
dfabulich says (to baf), “Yeah, I think Walking Dead‘s primary difference is that it’s actually a lot *smaller* than even a typical CYOA book. You can get the whole flowchart on a pretty tiny screen.”
katherine asks (of baf), “a combinatorial thing, you mean?”
dfabulich says, “Whereas I think it would require multiple pages of flowchart to chart out even our smallest game, Choice of the Dragon”
ChrisC says (to baf), “It’s worth keeping in mind also that this is just one method of organizing a game that CoG favors; the purpose is to help authors plan a game with manageable reactivity so they can finish it, while providing enough so that it’s satisfying to play.”
ChrisC says (to baf), “It’s theoretically possible to make a huge sprawling game in CS or otherwise that doesn’t merge its branches every chapter.”
dfabulich says (to ChrisC), “I agree. It’s like The Formula used for screenplays. You can get a dependably good product out of it”
baf says (to katherine), “I mean in terms of the number of nodes in a complete story from beginning to end.”
baf says, “Considered independently of the number of nodes in total.”
katherine says, “ah”
ChrisC says (to baf), “Ah, the length of a single playthrough.”
baf says, “Right.”
dfabulich says, “Our goals for authors is to give us at least 60K words of source code that generate 20K words in a single playthrough, as measured by our randomized test tool”
dfabulich says, “Typically delivering 20K words per playthrough forces the total word count up higher, closer to 80K or 100K”
baf says, “Hm. Putting it that way suggests a simple ‘how much does it branch’ metric: you want it to branch by a factor of at least 3 (and typically get a factor of 4 or 5).”
baf says, “Although measuring it that way would not distinguish between a story with lots of branches and one with just one choice at the very beginning followed by three different static novellas. So never mind.”
eu says (to baf), “I think it’s still a useful measure.”
eu says, “Just one that need to be considered in the context of others.”
ChrisC says (to baf), “Good point.”
ChrisC says, “Hmm, I had been thinking in terms of “breadth” and “depth”, but that’s not quite enough as baf points out.”
Busta says, “You could add number of choices to breadth and depth.”
eu says, “Another obvious metric is number of distinct paths.”
dfabulich says, “Yeah, we separately ask authors to avoid having more than 400 words between choices”
ChrisC says, “Busta Maybe, but if your choices are 1) open left door [turn to page 55] 2) open right door [turn to page 55] and choices have no real effect, that’s not saying much.”
Busta says (to dfabulich), “It depends on your audience, which probably works for most. I would love a story with long amounts reading in between choices.”
katherine asks, “surely that would work better when you don’t have to see the “turn to”s though?”
katherine says, “unless someone is lawnmowering, but they get what they deserve then”
katherine says, “(I’ve seen a few games where that kind of thing is played for comedy at lawnmower-ers’ expense)”
dfabulich says, “I want to raise a question that I ordinarily wouldn’t bring up, but this is theory club, so what the heck: What’s the relationship between stateful choice-based games and “true interactive storytelling?””
ChrisC says (to katherine), “It’s less apparent if it’s hidden, yes. But the false choices are still false and inflate the ‘number of choices’ measure.”
Busta says (to ChrisC), “Number of choices on its own isn’t a good metric, but combined with breadth (amount of distinct paths) and depth (length of each path), I think it works.”
baf says, “To my mind, stateful choice-based game is true interactive storytelling.”
baf says, “(I only get persnicketty when it isn’t stateful.)”
dfabulich says, “Which is to say, I think some would agree that stateless CYOA is interactive enough to call it “interactive storytelling,” but others would disagree. I think stateful/dynamic choice-based games are, but I think folks like Chris Crawford would disagree that it’s “true” interactive storytelling”
katherine says, “yes they are true scotsmen etc”
ChrisC exclaims (at I), “don’t think you can ask that without first defining “true interactive storytelling”!”
zarf says, “non-stateful choice-based-game is ‘true interactive storytelling’ if you’re smart about presentation”
dfabulich asks (of zarf), “what sort of cleverness do you mean?”
zarf says, “(Angry Birds is ‘true interactive storytelling’ if you’re a pedant)”
zarf says, “well, I go back to Meanwhile a lot, which has some state but in clever ways”
zarf says, “the obvious bits — the combination locks — are not the crucial ones”
ChrisC says, “I’m not even sure that ‘stateful’ is a sufficient qualifier: if all the states are booleans, say, that’s identical to statelessness, as Fabulich says.”
dfabulich says, “The Meanwhile iOS app has actual game state, IIRC.”
eu says (to ChrisC), “Yeah, all programs on finite computers are equivalent to FSMs.”
zarf says, “the clever parts are where the ‘state’ is what the player knows about the story, which builds up over time”
dfabulich says (to ChrisC), “Yeah, I struggled a lot to come up with good nomenclature. I think my personal struggles bored even the theoryphiles on, which kinda surprised me. :-)”
eu says, “The difference is in which way it’s easier to think about.”
zarf says, “and of course readers do that in any sort of fiction”
dfabulich says, “
dfabulich says, “I ultimately satisfied myself by distinguishing between “static pages” and “dynamic pages,” where the page itself changes based on earlier choices.”
zarf says, “so in theory even an ordinary novel could approach this sort of interactivity, the ‘holy shit everything I thought I knew about this story is wrong’ experience, which is deeply related”
zarf says, “to IF”
zarf says, “and there I go off into la-la land”
ChrisC says (to dfabulich), “That’s a good one. I think that sums up things well, actually.”
dfabulich says, “I would include Fighting Fantasy paper gamebooks in the “dynamic pages” camp, in the sense that the page says, “if you have found the key to this door, then turn to 67″ so the page may be literally static but the interpretation of the page is directly responsive to the player’s past choices”
baf says, “Trapped in Time from the last comp is a pretty obvious example of ‘stateful due to state carried in player’s head'”
baf says, “(or on player’s notepad or whatever)”
ChrisC says (to dfabulich), “And there’s gamebooks with even more stats, like health and so on and you roll dice to fight enemy planes or monsters.”
Busta says (to dfabulich), “your definitions work for me.”
dfabulich says (to baf), “Trapped in Time was also an excellent use of its medium, in that it used something paper gamebooks have and computer-based choice-based games often lack: you’d get to skim past alternate branches while navigating to your destination.”
katherine says, “oh huh, I always saw that as a drawback of paper gamebooks”
katherine says, “forced spoilers, etc”
zarf says, “oh, it can be used well”
zarf says, “again, Meanwhile
dfabulich says, “The player flips over to 45, but on the way, you’re like, “wait, what’s this thing about Hitler? Dinosaurs? Whaaat?!””
Alex says (to dfabulich), “Awwwk! Word on the street is that dinosaurs are or were huge reptilian beasts who ruled Earth long before humankind. They all became extinct years ago, but that hasn’t stopped them from starring in all sorts of movies, comic books, and TV shows.”
katherine says, “(granted it’s been a while since I read/played that, for all I remember it may have done something with it)”
dfabulich says (to zarf), “Indeed, Meanwhile makes excellent use of that”
katherine says, “invisiclues did that a lot”
baf says, “Trapped in Time used it to its advantage. When you’d browse past passages about velociraptors or ancient Rome, it enhanced the sense of being trapped in your time loop and unable to access things you should be able to access”
Gunther says, “well, *some* gamebooks had things like “you have found XYZ, add 39 to the page number you’re on to see if you can use it there””
katherine says, “different medium, same idea”
dfabulich says, “And, of course, UFO 54-40, where the easter egg is a double-facing page illustration in the middle of the book, where you’re constantly flipping past it”
Gunther says, “as has been mentioned, I believe”
baf says, “And yeah. Meanwhile did the same thing”
djfletch says, “so a disabled-but-visible choice is a different technique from a completely hidden choice”
baf says, “With illustrations, it’s even harder to avoid looking at it than with text.”
baf says, “Speaking of loops”
eu says, “Huh. It would be interesting to see that in computer-based CYOA. An animated page-flipping or some such.”
dfabulich says (to eu), “I’d say Trapped in Time is what does it, by being implemented in PDF, so you’re scrolling past the spoilers”
katherine says, “a few children’s books adaptations did that iirc”
katherine says, “although again that is based on memory from way back”
katherine says, “as in “children’s as in I owned them way back” memory”
ChrisC says, “Yeah, that is definitely a result of the book-as-object that gets lost in translation to digital.”
dfabulich says, “The nice thing about doing it in PDF is that you can be sure that the whole thing is completely static. “Nothing up my sleeve””
eu says, “(You can embed JavaScript in PDFs. But yeah.)”
dfabulich says, “I had a question for maga, but I guess maga’s not here, so maybe y’all have an opinion. Maga seems to have identified on his blog a type of CYOA structure which he calls a “time cave,” of which “Cave of Time” is the paradigmatic example. But I’m unclear on how exactly he’s defining the structure”
baf says, “It seems to me that a lot of choice-driven IF is also plot-driven, and thus avoids looping back to specific nodes within a play-through. But I’ve seen dungeon-crawl gamebooks where there are sections spent just exploring the corridors, and which specifically let you wander around arbitrarily revisiting nodes.”
ChrisC says (to dfabulich), “ is his note on that one.”
baf says, “And I’ve seen some Twine stuff that does this too, but it’s not as prevalent as in parser-driven IF.”
ChrisC says (to dfabulich), “” This is what I think of as the classic CYOA structure: a choice almost every page, heavy branching, strong identification with a second-person AFGNCAAP, almost no merging of branches. The diagram resists being drawn as a vertical flowchart: it wants to be a radial sea-creature. The Cave of Time doesn’t really have a unifying story: it’s a number of different stories that happen to share a general style and some early content. (The cover of Aaron Reed’s Creating Interactive Fiction with Inform 7 is a particularly nice example of this kind of branching.) “”
dfabulich says (to baf), “Yeah, I think that’s a very important structural difference. When you loop back, you get the feeling of exploration; without it, you have constant forward-moving plot”
baf googles up the cover to Reed’s book, even though it’s right there on the shelf like three feet away
dfabulich says, “I know a lot of IF lovers love exploration, but I’m personally drawn to choice-based games because I personally find exploration totally boring.”
zarf says (to dfabulich), “yeah, that is key”
zarf says, “although, to complicate life even more, Emily’s Versu stuff tries to appropriate exploratory scenes (long dinner conversations) for plot purposes”
ChrisC says (to baf), “Yes, and I think that was one of Crawfords goals in Storytron: to have author set up a lot of subsystems (verbal combat, throwing a party, gossiping) that could be reused with different participants involved and result in wildly different proceedings and outcomes as a result.”
ChrisC says (to baf), “Or even the same people, if they were feeling different that day.”
Busta says, “there’s different kind of exploration. Physical exploration is one. Character exploration, narrative exploration, etc…”
zarf says, “but I mean, structurally, the ability to try a bunch of options within the same context, like dfab was describing”
dfabulich says (to zarf), “Blood and Laurels doesn’t feel like it has totally exploratory scenes, though I can see why somebody would call them that”
baf says, “(I remember when I wrote The Gostak, I explicitly thought of the one conversational NPC as the game’s ‘maze’.)”
zarf says, “rather than being pulled forward steadily”
dfabulich says, “In B&L you tend to have choices like: Talk about X, Talk about Y, Talk about Z, or wrap up the conversation”
zarf says, “flirt with host, eat grapes, argue about poetry…”
dfabulich says, “But choosing a conversation topic wasn’t just exploratory; it felt like it was updating relationship scores. Wrapping up the conversation wasn’t just “bored now, let’s get on with the plot,” it was making a decision not to tweak my score any further”
Busta says, “I so wish I had snagged B&L. I missed it by like minutes.”
zarf says, “the point of hte ‘wrap up’ choice is that you don’t have to wrap up until you felt it was time”
zarf says, “hm”
zarf says, “this may just be personal approach; I don’t think in terms of relationship scores much”
baf says, “That isn’t at all how I felt playing B&L.”
baf says, “But I only played it once (with the San Fran IF group), so perhaps I was still in exploratory mode and would have shifted to stat-optimization on replay.”
dfabulich says, “I just realized that I should set up a Google Hangout “Let’s Play B&L“”
zarf says, “that would be nifty”
Busta says, “oh hells yeah”
zarf says, “pointing the camera at the screen for the folks that never got to buy it”
dfabulich says (to baf), “I’m not sure the word “score” was at the top of my mind, but I was definitely thinking, “I could make this character like me or hate me, hmm””
baf says, “Emily’s going to be in the bay area soon, fwiw”
dfabulich says, “What do people wish someone would do with CYOA structure that they haven’t seen much/before? [i.e. What should Practice Club work on? :-)]”
Busta says, “I tend to like my CYOA looooong.”
Busta says, “So not so good with Practice Club haha.”
eu says, “I don’t know if it falls into the scope of Practice Club, but I would be interested in seeing ways of extending these graph-based analyses to IF we don’t normally apply them to, like traditional parser-based IF.”
baf says, “Hm. Symmetries. Hidden patterns in the choice structure that you only notice when you graph it out.”
Busta says (to eu), “I think parser games really are CYOA games, with implied choices instead of explicit ones.”
eu says, “Zarf’s model checker that he built for HL falls somewhat in that direction.”
baf says, “I’m not sure I’d actually appreciate that if I played it, but I think it would be cool.”
eu says (to Busta), “Sure, but they tend to have an abundance of extra state that’s not plot relevant. Like is the foobar directly in my inventory or in my rucksack.”
dfabulich says, “I expressed a wish about this on Emily’s blog, but I want somebody to make an entire game in the style of the “Path of Love” sequence from Blue Lacuna
eu asks, “Er, ?”
katherine says, “oh hmmm”
baf asks, “That’s when you refuse to go on the adventure at first and it summarizes years?”
dfabulich says, “If you choose to try to stay and don’t paint the painting, you enter a sort of “montage mode””
ChrisC asks, “Maybe recursion built into the story structure, like Howling Dogs but the main hub reflects actual choices you’ve made in the past and keep returning to it to see new changes?”
dfabulich says, “Specifically, the interaction is: the game sets up a situation, then lets you type in a response. “Your daughter cooks chicken, and you think it tastes: ____””
dfabulich says, “Aaron put in some obvious guesses in the source: “good, bad, ashes, delicious” and searches for those words in your reply”
dfabulich says, “But if you say something that he didn’t think of, the game doesn’t beep and say, “sorry, I didn’t understand you” … it tries to fake it, like Eliza the AI therapist”
baf says, “I remember my own responses took a turn for the dark as I realized that this was going to keep playing out no matter what I typed”
ChrisC says (to dfabulich), “The thing about building like that is it requires a big dictionary of related words… If you aren’t going to go full-on Scribblenauts it’s a ton of work for uncertain gain from the player’s POV”
dfabulich says, “The montage-summary framing makes this easier because if he didn’t understand you, he can kinda say, “Mm-hmm, I see, well, three years later…” but more cleverly”
ChrisC asks (of Wait), “how does it fake it?”
dfabulich says, “The source is online, let me try to dig it up. It’s not sophisticated. (Neither is Eliza.) It just says something non-committal and changes the subject”
Maeja says, “That sounds like what I tried in my first draft of An Earth Turning Slowly, but I couldn’t pull total freeform input off in any convincing way.”
dfabulich exclaims, “oh, man, is down!”
Busta says, “The problem with the faking is it takes away the meaning. If I answered with ‘like the sweet nectar of the gods’ or ‘like a dead cat’, it doesn’t matter.”
ChrisC says (to dfabulich), “There’s been a lot of work done in making much better chatbots than Eliza in the last 50 years. There could be some interesting ideas using e.g. Markov chaining.”
dfabulich says, “
dfabulich says, “if pcmd matches the regular expression “(all right|alright|allright|ok|hurt|help|alive)”:”
eu says, “So layers of regexp attempts.”
dfabulich says, “Exactly”
dfabulich says, “Which is exactly how Eliza works, too.”
ChrisC says, “Yep.”
dfabulich says, “It looked pretty painful to code this in I7 :-)”
dfabulich says, “I bet that there’d be a nifty DSL to build Eliza-like chat bots for stageful CYOA”
dfabulich says, “stateful (curse you OSX autocomplete)”
zarf says, “so, we approach the two-hour mark”
dfabulich asks, “Is that the usual wrap up time?”
ChrisC asks (of zarf), “Is this the usual run time?”
zarf says, “I don’t remember, but if we don’t wrap up the transcript is never going to end :)”
Busta says, “I think now is a good as time as any.”
ChrisC says, “I think this is a good stopping point.”
Zach says, “I don’t want my hard drive to run out of space.”
zarf says, “Zach seems to be idle, so let me say ‘thank you for coming!'”
Zach says, “I don’t want my hard drive to run out of space.”
zarf says, “feel free to continue chatter in the channel”
zarf says, “aha”
eu exclaims, “Thanks, all!”
dfabulich exclaims, “This is the club that nev-er eeeends! It just goes on and on my frieeeeends!”
Maeja exclaims, “Thanks, everyone! It was interesting to watch!”
zarf says, “when the transcript goes up I will add another PracticeClub call to it”
ChrisC exclaims, “Thanks for running and recording it!”
Busta exclaims, “Thank you!”
Zach says, “And please come again anytime, theory club or not.”

14 thoughts on “Transcript of May 10, 2014 ifMUD Discussion on CYOA Structures”

  1. And to follow up on a couple of things, since I wasn’t there:

    I would have liked to have brought up 20 Strokes, from the recommended reading. It’s a seemingly very linear CYOA-esque Twine piece in which new possibilities appear if you wait long enough (in real time) on a particular screen, making time another axis of choice and providing a new way of concealing/revealing possibilities during play. I’m not sure whether I thought it was completely successful, since I didn’t discover that feature on the first playthrough and I think it would be possible for a lot of readers to miss it completely — and I’m always doubtful about techniques that allow readers playing in good faith to completely miss one of the major points of your work. On the other hand, it was interesting, at least to me, that it converted hesitation and doubt on the reader’s part into part of the narrative fabric of the game.

    More generally, I think allowing “wait” as a truly meaningful action (either in realtime or as a choice) opens up a lot of dramatic potential. It helps persuade the player that the NPCs have their own agendas, since they will go on and do more if the player is inactive. (I seem to recall once having a conversation with Jon Ingold in which he told me that true interactive drama was impossible without realtime, though it’s been a few years, so I may be warping his view here…). In any case, this was certainly one of the aims with Versu, to create a system where it was often possible and interesting to wait for the NPCs to do something, for a dramatic moment to open up. Come to that, Glass is an earlier experiment in making the player identify moments of dramatic intensity.

    I’m intrigued and kind of pleased that the people who liked exploration thought Blood and Laurels had it, and the person who doesn’t like exploration thought it didn’t. :)

    The issue of “do you count incidents and create numerical stats, or do you record specific events?”, teased in the middle of the above argument, is huge, and in my usual way I come down kind of in the middle. It’s satisfying to me as a player if I can look back at specific moments of drama where I did something big that had repercussions later, and the presence of this quality is one of the reasons I enjoyed Choice of the Deathless more than many another CoG game. But the stats-based approach allows for player planning and pursuing a particular agenda: “I’m going to play this game as someone with cutthroat ambition” is something the player can decide and then execute (especially if the game has said in advance what the key stats are going to be), whereas specific narrative moments of hard decisions are more difficult for players to build into a plan.

    One way to incorporate both effects is to have stats-counting events tracking what you’re doing, but to have the system make a note of which things you did that contributed to those stats, to be used later in some sort of call-back event in the story.

    Another thing I like (for dealing with that “wait, when did I tip over the line into being evil?” issue) is a double-down strategy: as the player approaches an important stat threshold, have the story specifically call it out somehow. Alabaster does this in several endings, with the graphics as an additional clue to what’s going on — the player gets a pretty clear signal that what they’re doing is akin to shoving a box towards the edge of a cliff, and if they keep it up, there will be a catastrophic outcome. But there’s also an opportunity to turn back at that point. If we want, that (choosing not to go all the way on a path they’ve started down) can also be framed narratively the character changing their mind about a particular value or aspect of their character and thus also be a strong story moment. Doing that procedurally does require a more dynamic engine than the average CYOA system, though.

  2. Re: That Homestuck-type game baf was talking about: Brendan Patrick Hennessy made a game like that, called The Thing About Dungeons. The disastrous consequences were mostly only connected to one or two at the very beginning, but otherwise it’s pretty close to what you’re talking about.

    1. Oh! And I guess it’s much too short for you to ever forget about making those choices, but I think this kind of thing could be sustained over games with longer playthroughs.

    1. This would be sufficiently annoying for me to set up as an ongoing thing that I’m not going to commit to it unless there are multiple people who want it and don’t have any way of turning HTML to PDF themselves. (For mobile reading, instapaper?)

      1. No problem, HTML is fine. You are doing such a wonderful service to the IF community, including novices like me; I would hate myself if I annoyed you!

      2. Ha, no, not that kind of annoying — just that it would be extra steps of work to make, and then I’d need to host it somewhere other than wordpress, and etc., and it would add maybe 30 minutes of fiddling to the transcript upload, which is already somewhat time-consuming to do because I edit fairly heavily to put in game links and so on.

        So I don’t want to get into a situation where the process becomes enough work that I can’t do it in a timely fashion after meetings.

  3. A tiny nitpick: as I saw already in earlier transcripts, the script that boldfaces names puts everything up to the LAST occurrence of ‘says’ in boldface (grep ‘Eliza’ in this transcript for an example), instead of the correct FIRST occurrence.

    1. In fact there is a Twine game that feels pretty strongly related to those, Michael Brough’s Candy Quest 3 . Personally, I really enjoyed both Candy Box and A Dark Room, though I’m somewhat more frustrated with the Kittens Game, which I am inexplicably still playing despite saying “This is really boring” to myself on a regular basis.

      Brough’s game is the only one of those I experienced as primarily narrative, though. There are, yes, narrative events in Candy Box, but I experienced them primarily as surprises. Surprise, the nature of the gameplay has changed! Surprise, it’s changed again! Which happened just often enough to stay engaging. The same applies to A Dark Room, while Kittens Game has been generally pretty static in its presentation and possibility space. The surprise-based pleasure of discovering that something that seems extremely simple is in fact not simple at all was a good trick, but I’m not sure whether it could be repeated in a satisfying series of similar games. It feels to me as though the genre may already be nearly played out. (I could be wrong!)

      Another factor was the nature of the choices being made, which were generally highly systemic and resource-management-related. There were very few choices that I made because I thought that they would produce interesting story results, so my expectations and sense of agency weren’t really in the narrative plane, but entirely about how many lollipops (or whatever) I was going to manage to produce, and what I would buy when.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: