I’ve been trying to construct a history of storylets and other, related concepts since year. Each time I think I have a decent timeline of general events and notable works, more get added and an entire new range of historical connections become possible. There is, as a surprise to no one, a great number of historical projects and ideas that have influenced others works or converged at the same time. Some concepts even seem to be lost for many years, only to spring up again in another incarnation.

Major Terms and Events

Reigns (2016)

While the term storylet does not appear in its design or in interviews on the development of the series (that honor goes to the application and series Tinder) starting with Reigns (2016), it is often pointed at as an example of a game using content selection based on a combination of randomness, values of variables, and what content the player has previously experienced each session.

Storylet (2010)

In the most abstract form, the definitional work from Emily Short (2019) is commonly used to define a storylet system as composed of (at least) three parts:

• Content.
• Prerequisites.
• Effect on (world) state.

Short’s (2019) definitions, though, are based the earlier coining of the term by Failbetter Games in 2010. However, as noted by the developers at Failbetter Games (2011), they were influenced by King of Dragon Pass (1999). This, in turn, was written in a programming language called Opal Scripting Language (OSL) that uses the central metaphor of selecting the next scene from a larger collection based on the values of variables.

Sculptural Hypertext Model (2001)

Around the same time as King of Dragon Pass (1999) was work by Mark Bernstein (2001) in a paper named Card Shark and Thespis: Exotic Tools for Hypertext Narrative coining the term sculptural hypertext model where links are “sculpted” away instead of authored as connections between parts.

Bernstein (2001) described a model by which links in a story where not authored but calculated depending on different variables. As a user interacted with the story, these variables would change and the links between parts would adjust as a result.

Adaptive Hypermedia (1999)

Two years before Bernstein (2001), Bra et al. (1999) coined the term adaptive hypermedia to describe system by which a user might see different links based on a model of their expected goals. Instead of serving all links the same users, some might be changed or calculated automatically based on different values. Traversal of nodes would be different based on variables and the system could adapt to delivery different content in different ways.

King of Chicago (1989)

In Story vs. Game: The Battle of Interactive Fiction, Doug Sharp (1989) describes an “honorable failure” project named King of Chicago (1989). Based on values, the project would select a set of series of sequences composed of different episodes. Depending on different player input, the internal values would change and the next sequence would adapt based on the current values and different episodes might be selected from a larger collection.

Storyspace and HyperCard (1987)

The tools Storyspace and HyperCard, both first released in 1987, are famous for different reasons. Several fundamental hypertext fiction works were created with Storyspace, and HyperCard became popular because of its inclusion on every Macintosh computer sold by Apple from 1987 to 1992 and its connection to the development of Myst (1993).

However, while Storyspace would go on to eventually incorporate the sculptural hypertext model, HyperCard was the first of the two to include a programming language named HyperTalk that allowed for creating projects where the different parts of a project, called cards, could be computationally reached without explicit user navigation. While technically not the first use of computational hypertext, it would popularize the idea.

Dragon’s Lair (1983)

On the Dragon’s Lair Project’s (DLP) website created by Kinder and Hallock (1997), there is a breakdown of scenes from Dragon’s Lair (1983) as well as some notes about when they appear. This includes an introduction of some terms used by Kinder and Hallock (1997) to discuss Dragon’s Lair (1983): scenes, cycles, and the conditions in which scenes become part of cycles. Dragon’s Lair (1983) always starts, Kinder and Hallock (1997) explain, with a cycle including 12 scenes. After this is finished, a new cycle is run of the remaining scenes until each set of scenes (three per set) is exhausted and the game ends.

While heavily dependent on random selection from the different sets for each cycle, player input plays a strong role in navigation. If the player succeeds in the scene, the next is runs. However, if they lose, scene is queued for later play and appears as fourth cycle created from any scenes where the character died.

Aspen Movie Map (1978)

The first hypermedia system was called Aspen Movie Map (1978). It consisted of a database of different recordings from the city of Aspen. A user could navigate any arbitrary street through the city in ten-foot movements and view each location from four view. Because the recordings were captured in early fall and winter in 1978, the user was also able to switch between the seasons and view a three-dimensional model of the city while navigating it.

Information for navigation was stored as metadata within the larger database. Depending on user input, information was pulled from the database and the metadata used to reference additional information while a user navigated the virtual city.

Narrative Databases and Computational Hypertexts?

In a post earlier this week, I discussed my work on SimpleQBN, a generic quality-based narrative (QBN) system. Among the improvements to latest version is the use of the package Mingo and the incorporation of the MongoDB Query Language (MQL). The changes bring it much more in line to a narrative database system where different parts can be selected based on searches and the values of variables. This is reminiscent of Aspen Movie Map (1978), Dragon’s Lair (1983) and King of Chicago (1989). In considering HyperCard (1987), Adaptive Hypermedia (1999), and the Sculptural Hypertext Model (2001), these all share aspects of computational hypertexts. Links are computed per navigation and are influenced or created as a result of user input.

Is a storylet system more of a narrative database or a computational hypertext? As was mentioned the other day, three of the current four (that I know of) open source implementations of the storylet model are based on Twine 2. While certainly not a requirement, the existence of Reigns (2016) proves the two are not explicitly connected, there seems to be a strong overlap between the two.