The Verbweb
In its most basic form, the concept of interactive storytelling is simple - to get into the Protagonist's shoes and direct the story. Yet simple dreams sometimes require surprising paradigm shifts to realize. Take aviation - probably one of mankind's first fantasies, and certainly one of the most popular in any historical age. Not only was it easy to imagine aviation, it was also easy to imagine how to achieve it - just don't fly too close to the sun and you'll be fine! It later turned out heat-proof glue wasn't all we were missing.

Experience reveals interactive storytelling has its own complications: interactivity threatens to send a story plummeting by breaking its narrative flow, and a story's formal constraints threaten to drown and asphyxiate interactivity.

For a very simple example, suppose we interactivized Little Red Riding Hood. Further, suppose we had 22nd century technology which allowed us to give the player complete realistic freedom of every sort - to move, to speak, and to interact. There are probably a couple of interesting ways the player could take the story, but picture these situations:

Player refuses to go to grandma's. Story dies.

Player attacks wolf and gets eaten. Story dies.

Player runs away from the wolf looking for the constable. Although it's reasonable to assume that there was a constable somewhere near Red Riding Hood's home, he didn't appear in the story, so the player never finds him and eventually gets bored and quits.

Player gets lost in the forest. Story turns into a forest navigation simulator.

Feasibility aside, complete realistic freedom pretty much guarantees that the player will either do things which make no narrative sense, or do things which do make sense only to find that they aren't working because he or she isn't getting the proper response. It also guarantees that the player will have to make a lot of tedious decisions - choosing which flowers to pick or what turns to take isn't exactly the most dramatic dilemma.

The solution is to only give the player dramatic freedom - that is, only allow him or her to do things that make dramatic sense, and only require him or her to make dramatically important decisions. This paradigm views a story not as a realistic recounting of occurrences like Red Riding Hood's footsteps, but as a more abstract chain of dramatic events, Red Riding Hood's whole journey being one such event.

The main question becomes how to define what makes dramatic sense. For this purpose Storytronics employs the Verb. Each Verb is a dramatically-significant occurrence, and since only Verbs can happen in a storyworld, it is guaranteed that all interactions have a dramatic significance.

Let's contrast this Verb-based approach with the aforementioned realistic approach to story, using a simple activity as an example: eating. Eating is rarely dramatically significant, however In Red Riding Hood it is extremely important because it is a means to murder. So according to the realistic approach, to model the story of Little Red Riding Hood we must allow the characters the ability to eat. The Verb-based approach, on the other hand, says that it would make little sense to include a Verb Eat in this story - allowing Red Riding Hood to eat Grandma's food, or the flowers she picks, would not affect the story in any interesting way. Instead, we should use the Verb Kill to describe what the wolf does to Grandma and Red Riding Hood. It's a less specific description, but we can be sure that Kill will always be dramatically significant, no matter who does it to whom.

The Verb-based approach truly shines when we consider that every occurrence's dramatic significance depends on its context, and that each story occurrence must be relevant to its context. For example, while Kill is always dramatically significant, it's a Verb that suits a relatively small number of contexts. It makes no sense for Red Riding Hood to kill her mother, and it's impossible for her to kill the wolf. However, with the realistic approach there's no way to avoid it - realistic freedom means the player can do whatever he or she likes, including turning Red Riding Hood into a juvenile serial killer or a Rambo-like warrior, provided, of course, he or she can devise the practical means. For a realistic simulation, that's great, but for interactive storytelling it's terrible, as it results not in stories but in a free-form videogame filled with tedium and meaninglessness.

But, using Verb-based dramatic interaction, it's easy to ensure that a Verb can only happen when it is dramatically appropriate. In a Storyworld, Verbs are arranged into a "Verbweb". The Verbweb defines, for each Verb, which Verbs can succeed it in the story. Here's a simple Verbweb:

This Verbweb is a partial description of a bar brawl. It opens with someone bumping into someone else, the Verb at the top. There are three Verbs defined as possible reactions to this - the bumpee chooses one of them. The two Actors then alternate action/reaction according to the web. The Verbs AppealForPeace and DemandPeace can be performed by a caring bystander, to which the brawlers may then react by the Verbs Threaten, ShutUp, Grumble or NotScaredOfYou. Try tracing the possible Verb chains - you will notice they all make sense. You'll also notice the story is cut short - this is a very small portion of a complete Verbweb.

This graphic is only truly a simplistic depiction; a Verbweb doesn't only define which Verbs can follow which, but also the conditions that allow them to follow each other, the Actors who are allowed to perform these Verbs (this is usually not set in stone but depends on the context), the Actors who can be acted upon, and which optional Verb a computer-controlled Actor should choose in each situation (this usually depends on that Actor's personality).

To truly understand how rich, yet how well-structured Verbweb interaction can be, consider that the Verbweb is traversed simultaneously by all Actors in the storyworld. That is, at each point in the story, each Actor is performing a Verb, to which other Actors can react with those Verbs that flow from it in the Verbweb. Using just the simple Verbweb above, we can have dozens of different brawls taking place in the story, both simultaneously and one after the other, each of which involves different Actors and resolves in a different way. However, all of them will always play out according to the principles defined in the Verbweb, so they will all have narrative integrity. Of course, a story with nothing but brawls won’t be very interesting, but a complete Verbweb will have potential for a great many other interactions.

Subnets
It’s reasonable to expect that certain Verbs appear more in one context than in another. Verbs that have to do with brawling, for example, will probably happen in close proximity to each other, and only in combat situations. A portion of the Verbweb which tends to “cluster” like this is called a subnet. For example, the above graphic can be called a "bar-brawl" subnet, because it's a cluster of Verbs that only deal with bar brawling, and, therefore, they all make sense in a certain context, and only in that context.

A subnet contains three kinds of Verbs - "entry" Verbs, "Exit" Verbs, and "Inner" Verbs. The entry Verbs are those Verbs that flow from some other part of the Verbweb. In this case, this means they are Verbs that don't flow from bar-brawling, but do lead to it. There is only one such Verb here - BumpInto. Of course, we could imagine other ways to start a bar brawl - there could be a "Poker" subnet, which contains the Verb "Cheat". If someone catches the cheater, they can have the Option to Punch them, which leads us into the bar brawl subnet. In that case, Punch would become an entry Verb for the bar brawl subnet.

If we did that, Punch would actually take the drama from the Poker Subnet into the bar brawl subnet. We would then say that Punch is an exit Verb for the Poker subnet. The bar brawl subnet has two exit Verbs: SheriffArrives and Grumble. We can imagine that ShriffArrives is also an entry Verb into the arrest subnet. Grumble, on the other hand, doesn't lead to any other subnet - it represents one Actor simply backing off, so there's no reason for anyone to react with any Verb.

A Subnet's inner Verbs are its meat and potatoes - those special occurrences which only take place in this Subnet - they only flow from Verbs within it, and only lead to Verbs within it. Most of a subnet's Verbs are inner Verbs.

A Comparison with Branching Narrative
It's easy to confuse the Verbweb with Branching Narrative (discussed here). A Branching Narrative is also a "web" of dramatic occurrences, but the superficial similarity belies a different essence. A Storytronics Verb is not a specific occurrence in the storyworld - it is a general possibility, which can be realized an unlimited number of times within the storyworld by any Actor, in any number of different contexts and at any number of different times. Each such specific occurrence is called an Event. Because Verbs are reusable, the variety of narratives which can be created by stringing Events is incredible. The nodes in a Branching Narrative are in fact not general Verbs, but hard-coded Events. They are almost completely context-specific, and thus there is a relatively small number of different ways to combine them.

A second difference is in the decision-making - in a Branching Narrative, the non-human characters don't make any decisions. The human makes decisions by choosing which node to move to, and the non-humans' behavior is pre-determined in each node. This makes it impossible to guarantee a large number of interesting and coherent characters. The best novelists spend years carefully counterpointing the behaviors of a dozen main characters - it's a tough enough job without having to account for the thousands of different paths the Protagonist could choose in interactive storytelling. In Storytronics, the computer-controlled Actors each make their own decisions in real time according to the personalities the storybuilder gives them. This guarantees each individual Actor's dramatic integrity, and, together with a good Verbweb, it ensures the sum of these Actors' behaviors will be an interesting, well-structured narrative.