In games, there has always been an opposition between narrative and interactivity, in which the story is told, for the most part, when the game is not being played and vice versa. In addition, there is the problem of narrative linearity, in that a lot of effort is put into the creation of a story and, therefore, the gameplay has to be crafted so as to not damage the invariants of the story. In the most extreme cases, this can cause gameplay penalties to the player, as illustrated by the following story:

Quote: Aaron Miller, at dinner right after the GDC, paraphrased "So there was this one level in Red Faction for the PS2 where you were supposed to play it stealthily, but I didn't want to. So I went back and did everything necessary to go through the level with all my guns, and I'd have NPCs saying things to me like 'Why, hello, Mr. Parker. You look really good in that suit today!' And, of course, I found that funny. But what wasn't funny was that there was a cutscene and level transition, and it was clear that the cutscene was prerendered as my character appeared in a suit, which was fine, I guess. But when I started the next level, I had been stripped of all my weapons! And that's not cool - when story linearity penalizes the player.

The challenge to you is not to write a "non-linear" story. No, the challenge is to author a premise and plot points that can be combined out of sequence to yield a unique "story" that is effectively authored (or composited, if you will) by the player's play. Obviously there will need to be choke points to sort of "prune" the "story tree," but these are also good points to evaluate the interactions of the plot points and the sequence in which they are applied, and to show the player consequences and thus deepen immersion. Ignore for now all the technical requirements of actually implementing such a system. Just assume that it exists or will be made. Thanks, and good luck!

I had devised a sort of game, some time ago. The basic idea was all stated in the intended name: "Rise Through The Ranks".

The basic idea was to create a multi-layered game. You start in an imaginary army in the beginning of an imaginary conflict as a simple private. From that point, you play in a FPS. You have different directives for your first missions, and secondary objectives. The number of secondary objectives and your degree of achievement in the missions is what triggers or not your promotion in the ranks. After being promoted, you don't play a simple FPS, but a sort of RainbowSix-like. You lead a squad, and keep on trying to achieve your primary and secondary objectives. Next promotion is field commander, and your interface becomes a sort of cross between Commandos and a RTS, or more accurately a sort of EuropaUniversalis. You should still be able to choose to go down a notch and take a more active part in the conflict if needed, by taking a squad, and moving to a secondary objective on your own, while still having to manage the whole of the battlefield. You could end up being promoted to General and having to play one scenario, probably the last, as the ultimate battle, and play a sort of Civilization or Ardennes.
The overall results of your mission determines what kind of reward you get. You could get medals if you saved your officer, which would lead to upping the difficulty of later levels for FPS and RainbowSix like levels, or getting fieldpromoted if you did not save him, changing your gameplay. Players could then decide for their career evolution plans...

And of course, you would have to design the cutscenes in order to show the real outcome of your play, that is design it with each member of your squad having a value set to either 0 or 1, 0 being killed and therefore absent from screen, 1 being alive and on screen. An even better design would be to design each member in different state of health, and show them the way that suits best the way they ended the mission. It would require only some tweaking in the disposition of the animations, but a lot of animation design and art.

A big problem would be to code four different engines and AI in order to enable the possibility to rise through the ranks.
A bigger problem would be to make your achievements and failures actually have an effect on the storyline proposed. (If you successfully destroy that convoy, then it is not used a ressources by your enemy in the next scenario, and if you don't, then the difficulty is uped a notch. Same applies for each side and each secondary objective. you end up having a real tree-like scenario, with different levels of playability for each part of it. A real nightmare for the designers, a real treat for the gamer...)
The biggest problem would be to actually sell it, since it would repel all persons NOT strongly addicted to each genre, while only appealing to those that ARE strongly addicted to ALL.. Difficult... Tricky, at best...



Yours faithfully,
Nicolas FOURNIALS

Plot content accumulation

Quote: Obviously there will need to be choke points to sort of "prune" the "story tree,"

This is not necessarily true.

An underlying principle behind making a set events to form a story in any sequence is the 'accumulation of plot content'.

Imagine an RTS where there are nine locations for you to attack. You can attack them in another order. When you are about to defeat one locations, the troops in that location retreats to the rest of the locations. As you continue your attack, you will experience more and more resistance. The final battle, no matter what sequence you attacked, is always the hardest.

This method is not about simply making the next location you attack harder. This method is not about simply having independent tasks that the player can complete in any order.

There is a sense of history maintained in this method. The tension in the later stages is built up from the events leading to them. It is the content of the plot that is being accumulated from one event to another. And this is why the overall story will make sense and intensify no matter what sequence the events take place.

Quote: Ignore for now all the technical requirements of actually implementing such a system. Just assume that it exists or will be made.

There is actually not much technical difficulty when you see a story this way. All you really need to think about is for each event, what content is going to be carried on to the next, and how the content will intensify the following events.



[Edited by - Estok on March 18, 2005 8:25:41 AM]

Quote: Original post by Estok
Imagine an RTS where there are nine locations for you to attack. You can attack them in another order. When you are about to defeat one locations, the troops in that location retreats to the rest of the locations. As you continue your attack, you will experience more and more resistance. The final battle, no matter what sequence you attacked, is always the hardest.

This method is not about simply making the next location you attack harder. This method is not about simply having independent tasks that the player can complete in any order.

That's gameplay, not narrative. Narrative is inherently linear, and you have an exponential number of combinations for every decision point with narrative consequence in the gameplay. Sequence of elimination of targets within a short interval doesn't satisfy that requirement (it has no narrative consequence), so it neither proves nor disproves anything.

Consider the following: you're playing some sort of armed vigilate/maverick, and you come across a character that you can either capture and take with you or kill off. All subsequent narrative sequences until that character is eliminated from the story arc have to take into account whether or not your captured the character and brought him along. Further, if you didn't bring him along, a good narrative will question the morality of your decision to kill him at multiple points. Narrative consequence.

Transforming Narrations
The RTS example is an analogy. The accumulation of the retreated forces on a battlefield is an analogy to the accumulation of plot content 'field of the plots'.

Strategic locations <=> Essential plot elements
Troops <=> Plot content
Troop density <=> Plot intensity
Accumulation of Troop <=> Accumulation of Plot content = narrative consequence
Roads on the battlefield <=> Relations between the essential plot elements
Final battle <=> Climax of a plotline

I was not talking about gameplay in this thread. I was indeed talking about the narrative. If you represent the plot elements of narrative using the elements in an RTS, you will see that in an RTS there is already a solution to your original question. How does the intensity build up in the RTS that I described? What is it that allows the intensity to build up not matter what order you attack?

These two questions are the analogus to your original question: How can we allow the plot of a story to intensity no matter what order the player chooses to encounter the essential plot elements?

You are correct that there is an exponential number of scenarios, in fact it seems that there will be 9! overall plots for the RTS analogy. Would you clarify what you mean in this:

Quote: Narrative is inherently linear, and you have an exponential number of combinations for every decision point with narrative consequence in the gameplay. Sequence of elimination of targets within a short interval doesn't satisfy that requirement (it has no narrative consequence), so it neither proves nor disproves anything.

"Narrative is inherently linear, and you have an exponential number of combinations for every decision point with narrative consequence in the gameplay."

This statement needs clarification. There are two agents making decisions in the RTS analogy. The player makes decision on where to attack, and the GameAI makes a decision on what troops to accumulate. In the analogy, there is no complication due to the exponential order of combinations from the perspective of the GameAI. Everytime a location is about to be captured, the GameAI gather whatever resources there are left and retreat. Therefore, in fact, the combination is way more than 9!, because depending on how you attack, the GameAI might end up retreat with different composition of units. But no matter what happens, the next location that you will attack is strengthened by the retreated troops. I don't understand what you are trying to say in this statement.

"Sequence of elimination of targets within a short interval doesn't satisfy that requirement (it has no narrative consequence), so it neither proves nor disproves anything."

In terms of the RTS as an analogy, the elimination of the 9 locations is the entire story. It is not a short interval. I think I didn't make it clear that the RTS example is an analogy. I use a lot of analogies because often times people only focus in terms of the domain of the question without realizing that the same question is already answered in another domain. Mapping a narration using the content of an RTS is like taking the Laplace transform of the narration. If you see the associations you will see how the questions were already answered.

Can you state clearly what you think I was trying to 'prove or disprove', and what I was using to 'prove'?

Quote: Consider the following: you're playing some sort of armed vigilate/maverick, and you come across a character that you can either capture and take with you or kill off. All subsequent narrative sequences until that character is eliminated from the story arc have to take into account whether or not your captured the character and brought him along. Further, if you didn't bring him along, a good narrative will question the morality of your decision to kill him at multiple points. Narrative consequence.

Plot content accumulation already include this. I actually erased an example in my original post because I thought what I posted was enough. This was the example that I erased:

Example - Thirteen Tails
In this game, you play a top rank warden who had been personally ordered by the emperor to pursue 13 terrorists/outcasts known as the Thirteen Tails. You can choose to encounter them in any order. The following list some ways where the plot content is accumulated at the end of each encounter:

- You are unable to defeat the Tail. It flees to another one (The next time you will have to fight two (or more) at the same time)
- You defeat a Tail, but an alert signal is sent to the other Tails.
- You defeat a Tail, but it leaves behind a mystery/question that can be answered by one of the remaining Tails.
- You defeat a Tail, but it holds the solution to a mystery that you will encounter later
- You defeat a Tail, but one of the remaining Tails will reveal the true identity of it, which may make you regret your earlier decisions
- The Tail joins you to pursue the rest
- You join the Thirteen Tails
(The standard storyline itself is about the grey moral values of the Thirteen Tails.)

Each Tail has a set of plot contents that can be carried on to the remaining set. Depending on how many and which Tails you have defeated, the current Tail will pass on the content that will ensure that a climax exists.

I was talking about what you were expecting to be talked about. You can see that the inverse transformed contents are in fact narrative consequences, not just gameplay consequences.

Quote: Original post by Estok
The RTS example is an analogy.

...

If you represent the plot elements of narrative using the elements in an RTS, you will see that in an RTS there is already a solution to your original question. How does the intensity build up in the RTS that I described? What is it that allows the intensity to build up not matter what order you attack?

*Checks original post*

Nope, that wasn't my original question. Satisfying the dramatic arc is not a current consideration. Merely ensuring narrative coherence is. That's why your RTS analogy doesn't make any sense to me. Besides, so we have an accumulation of troops who must then be defeated through force in an RTS; what's the narrative equivalent of "defeat through force"? We get a large pile of plot elements that accumulate. What then? How do we fashion a narrative based on player interactions? What about chronology? What's the selection order?

I need a lot more specificity from you, as I still have no idea what you're saying.

Quote: Would you clarify what you mean in this:

Quote: Narrative is inherently linear, and you have an exponential number of combinations for every decision point with narrative consequence in the gameplay. Sequence of elimination of targets within a short interval doesn't satisfy that requirement (it has no narrative consequence), so it neither proves nor disproves anything.

"Narrative is inherently linear, and you have an exponential number of combinations for every decision point with narrative consequence in the gameplay."

This statement needs clarification.

You are Bob. You are presented with a choice, with two possible outcomes. You take outcome A1. You advance, and are then presented with another choice, again with two outcomes. You take outcome B2. At this point there are four narrative continuities/combinations: A1A2, A1B2, B1A2, B1B2. Add another choice, again with two outcomes, and the total number of combinations comes to eight. Exponential growth.

The key clause is "decision point with narrative consequence." Not all decisions have a larger impact. You could take a drink of water now or take one later, big deal. You could kill or sedate the sentry. Either way, he knows something was there and communicates the information to his fellow sentries and superior officer. No narrative consequence. However, if you avoided him using stealth, then that opposition between engagement and avoidance has narrative consequence, because in the latter there would be no propagation of data to other sentries and/or the superior.

Hope that clears things up.

Quote: I don't understand what you are trying to say in this statement.

"Sequence of elimination of targets within a short interval doesn't satisfy that requirement (it has no narrative consequence), so it neither proves nor disproves anything."

Simple: continuing with the scenarios outlined above, if the choice to sedate or kill (S1) is presented between larger choices, then sedate/kill becomes inconsequential, because it has no effect on narrative or approach: either way, the other sentries become aware of a hostile and armed/dangerous presence. In contrast, if the option to avoid is available for a latter sentry (S2), then the narrative can discard S1 because both outcomes have the same effect but can not discard S2. S2 has narrative consequence.

Here's the core problem:

Quote: In terms of the RTS as an analogy, the elimination of the 9 locations is the entire story. It is not a short interval.

See, that's where we disagree. In our classical gaming considerations of story/narrative vs interaction, the elimination of the nine locations is pure interaction, but the interaction should have narrative consequence. Narrative consequence is the mechanism for enabling player interaction - gameplay - to affect the story. Unifying them effectively eliminates the story, as there is no recounting or presentation to do anymore.

The story is larger than the elimination of the nine locations. The story may be how one unit survived a tour of duty intact, or how the brave sacrifice of one company turned the tide of the larger war. The story has to have emotional impact. I see what you're trying to say about reinforcing latter locations, but that just means more targets to (try to) kill. The real narrative impact in that same scenario would be the fact that the latter locations knew you were coming and had a good idea what to expect (narrative), forcing you to alter your tactics (gameplay). Perhaps they even sought to ambush you and severely reduced your ranks (narrative), which forced you to request reinforcements (gameplay) - but your spotty record means HQ is inclined to assign you to a desk job instead (narrative), so you cash in your chips and call in a big favor (gameplay)... Narrative consequence. It leads to a dynamic interaction between linear storytelling and gameplay, hopefully.

Quote: I was talking about what you were expecting to be talked about. You can see that the inverse transformed contents are in fact narrative consequences, not just gameplay consequences.

Indeed I do. Reading through your post again, I realized that I had taken your "elimination of nine locations" as a single, large combat encounter instead of nine separate encounters interspersed with all the narrative consequence of the gameplay outcome. Hopefully my elaborations from the preceding paragraph are compatible with what you were saying. I'll reread your post to see if there's anything of significance that I've missed.

Plot Content Accumulation - From the beginning

The concepts in the posts are rather misaligned. Maybe the RTS was too strongly associated with gameplay, but I was indeed using that example to speak of narrative consequence.

This is another example of plot content accumulation. Let's call this example 'CAT'. In CAT, you play the cat that is supposed to catch a mouse. There are three places that the mouse can be. Three is a very small number. Suppose we talk out all the other variable events that can happen in those three places, and just look at the sequence the cat can visit, then there are simply 3 choose 2 = 6 outcome sequences. This is a very small number that can be readily implemented using a plot tree.

Locations: Kitchen, Study, LivingRoom

P-KSL: The mouse stole cheese from the kitchen, you decided to see where it is going to bring it, so you decided to follow it. You followed it to the study, where the mouse was about to tear a page out of a book. There is no way you will let the mouse to that to the precious books of your owner, you started chasing it. You chased the mouse into the living room where there was a birthday party, and you made a big mess out of it.

P-KLS: The mouse stole cheese from the kitchen, you chased it to the living room, and going to make a mess out of it, the owner caught you while the mouse escaped to the study. After that you were angry and when you went to the study, you made sure that the blood of the mouse would stain on everybook in the study.

P-SKL: It was noisy in the living room due to the birthday party, so you hid in the study. You saw the mouse flipping through the pages of a book, you decided to stalk it. It went to the kitchen to steal some cheese, and you thought, "aha, we have a 459 here!" You started chasing the mouse and the two of you ended up in the party making a big mess.

P-SLK: It was noisy in the living room so you hid in the study. You saw the mouse flipping through the pages of a book. There is no way you would let the mouse touch the precious books of your owner, so you decided to pursue. The two of you ran into the living and were about to make a mess. Your owner caught you and made you sit in a chair. You vowed to make the pieces of the mouse spread across the entire kitchen.

P-LKS: You were at the living room when you saw the mouse looking at the cake on the table. You were conviced that the mouse was going to bring havoc to the cake, but instead of taking a bite out of it, the mouse went to the kitchen. You followed the mouse to the kitchen where the mouse stole a piece of cheese. You started chasing the mouse and ended up in the study. You could almost catch the mouse, but your owner grabbed you on the back of your neck.

P-LSK: You were at the living room when you saw the mouse staring at the cake but not taking a bite out of it. You followed the mouse to the study where the mouse was flipping through a book. Your owner had been searching for you and went to the study, where she saw the mouse and started attacking it. The mouse really hadn't done anything so you ended up protecting the mouse and let it escape. The owner caught you and made you sit in a chair with a funny hat and flashed bright interrogating lights at you for your treason. At the end of the day you were exhausted and went to the kitchen, where the mouse came out from the dark, with a tiny candle-lit cheese cake for you.

The above plotlines show what a plot tree can do with static story elements. There are three locations with time-variant but fixed events (if you divide the story in three time slots, the NPCs always try to do the same thing at a certain location during a certain time slot), three characters where two of them have static goals and personalities. Because the plot tree is explicitly constructed, the story arc is guaranteed to exist. However, it should be obvious that as the number of story elements go up, an implementation using a plot tree will become prohibitive because the complexity of the plot tree is in the order of O(n!).

The question was how can we implement the varying sequences without engaging the O(n!) complexity. How do we guarantee that the story will make sense, and that the existence of the dramatic arc? (The argument I was presenting was indeed a little stronger than what you envisioned.) The solution comes from an observation of the gameplay progress of an RTS. With all the variables inherent in a game of RTS, the complexity of the 'story', if put in a plot tree, is even larger than O(n!). Why is the 'story' in an RTS able to be coherent, and intensify? What is it that allows the story arc to construct itself?

If the intensity of the RTS game is represented by the density of troops, then we can see that the intensity increases because the troops accumulate by themselves as the campaign ('story') goes on. What is the equivalence of 'troops' in a story?

The equivalence of 'troops' in a story is what I called the 'plot content'. What exactly is 'plot content'?

The CAT example has six plotlines. These six plotlines are based on one main mystery:

M1: What is the mouse trying to do?

The plotlines with respect to M1 can be described as follow:

M1: The story will begin with the mouse doing some suspicious things, at some point, the cat can declare, "that is it, I have enough evidence!", and the pursue will begin. The subsequent plots are intensified by additional emotions, such as:

E1: You are trying to catch the mouse for your owner, but at the same time making her angry for making a mess out of it. The plot intensifies by the question, "Does the ends justify the means?"

E2: The mouse took advantage of your relationship with the ower to make you get caught by your owner in the living room. Your rage goes up. The plot intensifies by adding the element of revenge into a duty.

E3: In P-LSK, you never began pursue, because you believe that the mouse is innocent. In this case, the plot is intensified by introducing regret and doubt, "Did I do the right thing? What was I supposed to do?"

The sources of these emotion build-up are the plot contents. The plot contents are the plot elements that must exist for the plotline to growth toward certain climax. For CAT, the plot contents are:

PC-E1: The intensifying chaos, the involvement of the owner, the messes being made (To build the case of 'does ends justify the means?')

PC-E2: Being captured and punished by the owner (to build the motivation of revenge)

PC-E3: Choosing not to pursue the mouse, choosing to protect the innocent, the imprisonment of the pursuer, the accusation of treason. ("Did I do the right thing?")

The plot contents are delivered through various methods, such as foreshadowing. Notice that the number of catagories of plot content is 3, not 6. Imaging that there are three buckets labeled PC-E1, PC-E2, and PCE3. The buckets represent the plot contents already delivered to the player. Now, imagine that there are three lids labeled 'kitchen', 'study', and 'living room'. The lids represent the ending location of the game.

As the game progresses, the GameAI will select an appropriate plot content from the set to put in the corresponding bucket. The job of the GameAI is to fill up one of the buckets by the time the game ends, and make sure that a lid can be put on the bucket being filled up. In the CAT example, the PC-E3 bucket can only accept the 'kitchen' lid. A general strategy for the GameAI will need to is to throw more than plot content into the buckets at the end of each section to guarantee that one of the buckets can be filled with a lid on.

This topic is of a similar in nature to the topic on portraying story and so I'll borrow my revenge example form there and use to illustrate an concept I'll call Accumulation of Event Resolutions(AER).

The essential concept of AER is that rather then a tree approach to the story a node approach is used, where each node is triggered by accumulation of certain event resolutions. There may be several nodes at the same point each with a different set of requirements but the only node that is trigger is the one the player has accumulated. Let me illustrate this:

Preparation
The Main Character(MC) is now a young man in a run down and dilapidated training ground in the middle of the forest. As the player examines various objects in the training ground flashback sequences are triggered involving the MC, the female protagonist Celeste, and the object. The player the proceeds to play out the flash back sequence and associated interaction with Celeste that take place. An example of a flashback would be one that is triggered by the player examining a wood sword. The flashback consists of the MC sparring poorly with Celeste, accompanied by a dialog interaction with her during the match.

How these flashback sequences play out determines the MC starting state, their affection level with Celeste, and whether she is waiting for you at the exit to the area or whether you head off on your journey alone.

A five node resolution would be present at the end of this scene when the player goes to leave the training area. AER is used to determine which of the five nodes is triggered based on how the various flashbacks the player initiated where resolved. The actual order these flashback where triggered in doesn't matter as the scene converges at a single point regardless. The five nodes in this case would be:
1)Celeste left long ago, but the MC doesn't care all that matters is revenge.
2)Celeste left long ago, and the MC has a linger sense of loss and regret that they drove her away.
3)Celeste is still there but choose not to follow the MC on their quest.
4)Celeste and MC journey out together as comrades in arms.
5)Celeste and MC journey out together as lovers.

In this way AER can be used to direct the story based on how the player resolves a set of events and not necessarily on the specific resolution of individual events.

The player breaks into an army base, sneaks passed the first guard, then the second, and finally the third and enters the main building. You don’t need a tree with 8 possible out comes for this sequence sine you can use AER and have only as many nodes as are meaningful. There could be only two nodes, entered undetected, or alarm was raised. A third node could be added if the player is allowed to kill guards undetected in which case the alarm is raised in X minutes based on the number of guards killed.

Two clarifications:

1) The Plot Content Accumulation is not a method about plot trees. The plot tree in the CAT example was used to show how Plot Content Accumulation can be use to solve the complexity problem from using a plot tree. In other words, Both PCA and AER do not use plot trees.

2) AER is a simplified version of PCA where the order of events are fixed. The mechanisms mentioned in AER are actually used in PCA. The resulting structure of AER is commonly presented as a directed graph (not a tree). For the example provided:

      P1        P1 = PlotPoint 1  .---+---.     C1 = Celeste left long ago, ...  | | | | |     C2 = Celeste left long ago, and ...  C1C2C3C4C5    C3 = Celeste is still there but choose not to ...  | | | | |     C4 = Celeste and MC journey out together as comrades in arms.  '---+---'     C5 = Celeste and MC journey out together as lovers.      P2        P2 = PlotPoint 2    .-'-.       S1 = Sneak pass first Guard   S1   F1      F1 = Fight first Guard    '-.-'           .-'-.       S2 = Sneak pass second Guard   S2   F2      F2 = Fight second Guard    '-.-'    .-'-.       S3 = Sneak pass third Guard   S3   F3      F3 = Fight third Guard    '-.-'      P3        P3 = PlotPoint 3

It is a common implementation. If you count the number of resulting plotlines, there are 5x2x2x2 = 40. You can put the above graph in a DAT model (where DAT stands for decision, assessment, tally):

      P1           |      A1      A1 = The GameAI gather the data from the sequence of flashbacks      |            to determine the relationship between MC and Celeste      P2           Assessment Points are points where the effects are not       |            Immediate. The effects are delayed until a Tally point.      D1      D1 = The player makes a decision on how to deal with       ||           First Guard      D2      D2 = ...      ||      D3      D3 = ...      ||      P3

What you can see here, is that TechnoGoth has not addressed how the assessments in A1 is going to be used (at a Tally point). If the assessment at A1 has a significant impact on the plotline, TechnoGoth has simply delayed the effect of the assessments by pushing them out of the scope. The question is how much do the assessments really affect the plot?

This is a possible scenario in DAT model:

     P1      |     A1      |     P2      |     D1         S1 = Sneaking through the first guard    /  \        T1 = the effect of relationship 1) which will determine whether  S1    T1           the subsequent path F1a or F1b is taken   |   /  \     F1a = Fight first guard, variation a   | F1a F1b    F1b = Fight first guard, variation b    \ /  /     D2    /  \.  T2    F2 /  \    |S2a S2b  |  \  |  /    D3     ... and so on.  T3, T4, T5 are not shown, they correspond to the                     moments when the effects of 3) 4) and 5) take place

In general this structure is not called a plot tree, but it is rather common in describing the flow of a story. One thing to note is that this structure is more suitable in describing a static plotline where the variations do not significantly affect the plot. If it does, the top level of the DAT model will indeed be a tree, but as you zoom in you will see the substructures. In other words, I am saying that the variations TechnoGoth considered are not significant enough, they are all in the same branch of a tree, that is why there is an illustion that the plot tree is dissolved. The implementation of AER requires the designer to define the entire DAT graph (which is not substantially easier than defining the entire plot tree, because the type of variations are not comparable to those offered by a plot tree).

PCA is a more hardcore version of DAT, where the effect is indeed a plot tree, but the complexity involve is not. You can see why it is not a DAT graph when you consider this:

Since there are only three choices at any given level, one might draw the plot graph as follow

       D1       The restrictions where the same location cannot      / | \     be visited twice is implied. (i.e. If you chose K at D1,     K  S  L    D2 and D3 will not allow you to choose K again.)      \ | /       D2      / | \.         K  S  L      \ | /       D3      / | \.       K  S  L      \ | /      Endings

Simple right? Not really. This graph is not meaningful because the actual events duing each K, S, L are highly dependent on the decision history. For example, if you are the GameAI, at the S following D2, you will have to make a hidden decision:

      D2    /  |  \         S2D2 = the hidden decision node within the GameAI that    K  S2D2 L               determines which plot element to unfold      /  \          KSL = In this case, the mouse brought the cheese from the    KSL  LSK              Kitchen to the study, and that was why you need to                          stop the mouse                    LSK = In this case, the mouse had not done anything and you                          are simply observing it.

In similar fashion, each of the K,S,L nodes after D2 and D3 are split into two. If you count the number of nodes at each level, you will get: {3,6,6}. Now, draw the actual plot tree and count the number of nodes, and you will get the exact same numbers. This shows that the original graph does not show the true complexity of the design. This also shows that the decisions indeed significantly change the plot, to a point where none of the subsequent events are shared. This is the type of plot structure that PCA is dealing with, a true tree, not a plot branch with insignificant splitting.

(Again, PCA implements a plot tree without using a plot tree.)


To TechnoGoth:
What do you think the relationship between PCA and AER is? In what ways are they different? In what ways are they the same?

The player has one day to print out his C.V / essay.

-The printer isn't working
>He can use his neighbours, but it is a poor quality printer
>He can phone the repair shop but it will take two days,
>He can read the manual and find out that it is just low on ink

>He can buy some new cartridges and try to install them
>He can go to the computer department and use their printers.
>