Rick Smith's Comments

Question 1: What do each of the two sides mean by ’storytelling’? Sub-topic: What the heck are "Laws of Drama"

An example used at the conference by Chris was that going to the bathroom (to use the facilities) is not dramatically interesting. This point got a surprising amount of debate with people giving examples of characters going to the washrooms in movies. (The only two examples were the president finding out about the start of WW3 on the can at the in DR. Strangelove and a thug forgetting his gun & being surprised on the can in Pulp Fiction.)

Both were intended to be somewhat humorous, and something of a shock because they are unusual. I think that Chris’ point stands, in the sum of all movies there are very few examples of people ’going’ but in real life a much greater amount of time is spent doing this sort of low level animal maintenance.

Eating is also (or would be also) dramatically uninteresting except that a huge amount of human socializing is done at the table. Thus hundreds of movies show people meeting at restaraunts, drinking alcohol, visiting over foods, eating together on dates, etc. altho relatively little screen time is spent watching people chew and swallow.

I am sure that you can find some movie that shows humans doing anything it is possible for them to do. You could argue that in a given drama in this PARTICULAR case it would be interesting or dramatic to show this action happening. Wouldn’t it be nice if all of these rare events could be captured by our drama engine.

Well yes it would. But in reality, it is so difficult to focus on the dramatic stuff that worrying about how to simulate going to the can seems misdirected.

I feel that the first level of the dramatic laws are a subtractive filter, yes people have to get from location 1 to location 2, but unless there is some interaction on the way, we should skip that. (Starship Titanic did this well, by the third or fourth time I had gone up the elevator I was pretty tired of the joke (cheaper elevator music for the 2nd and 3rd class passengers) but with a key stroke you could fast forward thru the boring bits. We want to find way to fast forward thru (or better yet skip altogether) the parts that are dramatically sterile.

The next stage would be building the engine is such a way that dramatic events natuarally arise from the game play. A huge amount of drama comes from communication between people so I think that this must be simulated in some way. In The Sims, communication was VERY abstract (you can stroke people, say nasty things, etc.) Despite this rudenmentry communication system, the game has been quite successful in large part because of this try at establishing relationships. (Also many people are bored to tears with the normal game offerings and jumped at the chance to see something different than whole sale violence in their computer games.)

Note that I am not against violence or ’action’ in computer games. However in great literature (which is filled with violence) the violence is placed in a social context. It is also (usually) rare. Lastly literature typically spends time exploring the results of the violence on the survivors.

The violence in computer games is socially unmotivated, sustained and there is little emotional reaction in the survivors (usually because there are no survivors...)

Our drama engine should thus be able to handle violence but I believe that the social and emotional results of the violence should be handled by the engine.

The last step in creating a drama engine will be to allow the stories author to be able to create themes that the playing out of the characters actions will elucidate. Great literature is great because the fictional actions of these imaginary people throw light on truths that we do belive in. If an interactive fiction dramatic engine does not allow this it will forever be overshadowed by books and movies.

I don’t believe that a ’plot - engine’ is required or wanted. If I am free to make choices with my avatar, I should have the freedom to direct the plot. If I have an interactive fiction based on "Hamlet" then I want to be able to explore a story space that considers the effects of pissing off an murderous relative. I mean, what would happen if Hamlet was the first action hero and took out the bad guys by Act 2. (On the other hand, perhaps I would be a complete git and get everyone killed.)

In summary, all of the requirements I’ve listed above are very difficult to program so I don’t expect to see them reflected in current games anytime soon. I expect that a special purpose drama engine will have to be constructed.

What do each of the two sides mean by interactive storytelling? (Sub-topic: Emergent Behaviour as story telling.)
I’ve worked in the game industry for more than a decade and I do not feel interactive storytelling should be viewed as emergent behavour from existing games. An example from my own experience:

Total Annihilation was a game I helped design, and I remained somewhat active with the fan based web sites for 3 years after the game came out. Many of these sites would post stories of battles that players experienced.

These battle reports would only be of interest to someone who was a fan of the game to be sure. But the stories that were most fun to read were the ones that departed the most from a strict retelling of the battle, and concentrated on the personal conflict of the players. Indeed some were a ctual SF stories, loosely based on any actual battle(s).

However, to be interesting, the events in the game had to be interpreted after the fact by a story teller. Given the very limited material, the better story tellers had to characterize and elaborate the conflict in a variety of ways. These storytellers also would drop 99% of what happened and pick a very few events that were dramatically interesting.

Now I feel that a good story teller can take any action game and find something to tell about a interesting play thru of it. However all the story telling is being done by the writer.

What is more, the fiction bought today (books, TV and movies for example) does not concern itself with the major game genres. How many first person shooter novella’s have you read recently? Real fiction uses characters, plot, settings, etc. to elucidate themes. It uses foreshadowing, rising tension and a host of other literary structures to create drama. Last I believe that humans have an instinctive ability to process stories. They can take a events happening to others (fictional characters included) and generalize the experiences to their own world view.

I believe that playing a traditional game and saying that stories ’emerge’ from it is far too easy. Interactive stories should have plot, characters, dialog and the personal growth that we expect to find in any professional fiction.

A point that I did learn at Ph3, was ’the man in the moon effect’ for stories. Just as our minds project a face onto blobs not at all face like, our minds will project story like characteristics onto some very unstory like raw material. This suggests that some parts of the story SHOULD BE best left to the player’s imagination. (What parts of interactive fiction should be left to the player’s imagination would make a good question to debate at a future Phrontisterion.)

However I feel that as a minimum, what an interactive fiction engine must encompass is:

- Characters that can interact meaningfully. - Characters that can change based on story events. - Rising drama to focus interest and drive plot. - The author being able to layer themes and ’deeper truths’ in the story.

An Interactive fiction with the above characteristics would be worth of the name, in my opinion.

What would be the characteristics of a product that best realizes the dreams of each party? (Sub topic: What do you mean, branching stories don’t work?)
A number of my students, when discussing Interactive Fiction (IF), often suggest branching stories. However the group this year dismissed this possiblity almost at once. I thought I would point out some pitfalls of this type of story telling for the readers of this report. Note that this is a quick summary of some articles on Chris’ site and so this essay won’t be of interest to you if you have already read Chris’ ideas on story trees.

Branching stories in IF are best demonstrated by the Japanese comic book games. In North America they are probably best known as the ’pick your own story books’.

In all of these products, the story trees generated were very sparse. The game or book did not provide nearly enough branches.

Why have existing products created so few branches? In the last pick your own story booklet I read, there were about 120 pages of text, with each page (approximately) being a node of non-interactive text followed by 1 to 4 choices. On the pages where there was only one ’choice’, there was no choice at all. This was just a way for the author to continue a single node past his one page limit size. In many choices you died almost at once leading to sparse branches. Very often I wanted my character to do or try options that the author did not provide, so I was forced into picking branches that I did not want to go to. I have played roleplaying game suppliments where the nodes were short paragraphs and the booklet had a couple thousand nodes rather than 120.

The cost of typing text for a book is far less than building a game. In the comic book games, there are typically long branches with few choices, followed by a increase of number of branches right near the end of the game. (This is to minimize the length of all of the different branches just as the story is getting bushy.) Basically since the cost of making content is so high, the designers are forced to create the content along the most interesting branches and ignore the less productive branches. However this inevitably leads to sparse branching. (Note that the cost of creating a node in a comic book game is much less than creating content in the computer games we are used to.)

Let me give you an example: I grabbed the very first book of short stories near me (the excellent Rockets, Redheads & Revolution by James P. Hogan) and picked the shortest story in that book. (The story was "Last Ditch" which is 7 and 2/3 pages long.) I then went thru this VERY simple story and counted how many choices there were in it.

I counted each segment of dialog as 2 choices (continue the dialog or end it and what to say). If one person replied over several lines, I generally counted the entire block of dialog a just those two choices. As a science fiction story, there were several paragraphs of explanatory material that have NO dramatic choices whatsoever, so this story is even shorter than the 7 & 2/3 pages would suggest.

This story has limited dramatic appeal, its enjoyment comes from a surprise ending. (This is often the case with very short stories since they have so little time to develop drama or characterizations.)

I found that this _very_ short story, with very limited drama had the characters make 104 choices.

Now as Chris has pointed out in his essays on this website, at the Minimum, there must be at least two possible branches or it isn’t a choice. Therefore this dramatically uninteresting short story is 104 choices in length. How many endings do we need if there are 104 choices where there are only 2 possible choices at each branch? There are 2 to the hundredth & fourth power number of end branches. (This story ended unexpectedly. It could be argued that MOST of the possible branches should have ran for much longer stories.)

Two raised to the 104th power is: 20,282,409,603,651,670,423,947,251,286,016 different ending nodes for this story. To create the ending nodes and intermediate nodes would take about twice as many as this number.

Now it should be understood that this is a VERY conservative estimate. I have counted each time a character speaks as only two choices, but obviously that person has the choice of thousands or millions of different things that they can say and how they wish to say it. When a character calls a meeting, there are trillions of variations of who to include or not include. However rather than millions of branches coming from each branch point, I’ve limited it to the minimum choice of two.

The way that the branching stories quickly explode into astronomical numbers explains why in branching stories, the branches of the story line are limited to a microscopic subset of possible stories.

Chris discusses a couple of possible ways of trimming these huge story trees (using "fold back" and killing some branches). But if either or both of these strategies are used enough to make a difference, the stories become very artificial in feel.

Chris has come up with another method (story webs) which use feed back and state variables to cause a much smaller set of nodes generate billions of branches. However most of these branches of dramatically sterile so a huge part of his story telling engine is trying to cut away the uninteresting branches.

In summary, the problem of exploding number of branches hangs over every attempt to make IF using a simple branching strategy. The attempts to get around this problem create a class of stories that so sharply constrain the player’s decision making that we end up with artifically simple stories with sharply limited choices.

Since we knew this, the branching stories strategy was dismissed almost at once by the panel.