The Narrative Structure of The Witcher 3

Let’s get this out of the way first. The Witcher 3 is an astonishingly good game: fun, engaging, massive, and worthy of just about every superlative you could chug down before overdosing on praise and reaching for your White Honey. It’s a very good game.

It also has a strong narrative, and while The Witcher 3 often gets praise for the scale and beauty of its game world, the scale and richness of its narrative is also quite extraordinary.

One of the ways in which the game design community moved beyond the arguments between narratologists and ludologists in the early 2000s was to accept that narrative was not just plot, but also included what the Russian Formalists call the Fabula, essentially the things in the story (especially the people, places and events). Games can benefit from a rich fabula without having to have a grand narrative arc, so for example League of Legends has a rich narrative element in terms of its character’s backstories and relationships, but no interactive story or plot to speak of at all.

The Witcher 3 takes full advantage of this approach in its own narrative world building, with strong characters, memorable places, and a host of details: from the notes on message boards, to posters on the streets of Novigrad, to the books scattered throughout the world, and your own evolving glossary and bestiary. Not to mention the visual details that hint at past events, such as the broken wagons and scattered goods you occasionally find, or elements that build a consistent picture of a world caught between two warring fronts. And the Witcher 3 does more than these Fabula level details. The many secondary quests, witcher contracts, and treasure hunts are also good examples of micro-narratives, small (typically simple) plot structures that can be easily dropped into a game world but are optional and largely independent of the larger experience.

But the Witcher 3 also has greater narrative ambitions, and layers on top of these a grand narrative arc in the shape of Geralt’s quest to find and help Ciri (his adopted daughter). Managing this arc within an open world environment so huge and yet so stuffed with Fabula elements is an impressive feat, and it’s worth looking at how the game manages it.

I have spent the last few years doing research in digital narratives using a particular model called sculptural hypertext (effectively a narrative state-machine); we’ve primarily been looking at location-based storytelling in the real-world (using smart devices and GPS), but the model works just as well at explaining virtual open world environments. In particular I think its interesting to look at The Witcher 3 in terms of some of the sculptural patterns we have identified.

The Witcher manages its grand narrative arc through two simple devices. A sequence of (mostly) linear story quests or threads, experienced within a broader three Act structure (four if you include the tutorial) which are effectively phases of the story. The diagram below shows the rough shape, although I haven’t gone as far as modelling all the quests, so the ones shown here are just indicative (you could do worse then this comprehensive guide to all the acts and quests on VG24/7). This model allows the player to move between different threads without necessarily completing them, and is hyper-textual mainly in terms of the juxtaposition of threads and the order in which they are experienced.


Essentially the player exists within one of the four phases, in which a number of potential threads are available, some of these are part of the chain that makes up the grand narrative arc and must be experienced in sequence, others are independent of the arc, and can be experienced in any order. Some of the independent threads are within the phase (and must be played during that part of the story), others outside of the phase (and can be played whatever stage the story is in – witcher contracts for example).

There are also a number of variations of the basic threads that are used in order to model narrative choices within the game. These are shown in the figure below.

Alternatives are used within a thread to allow the scene/dialogue that the player experiences to change according to their past actions and choices (typically whether or not they have completed a certain stage of another thread). Unlocking unpinned nodes (content that is not associated with a location) is used to build up a glossary of characters and a bestiary of creatures, these are also linked to encounters within the world. Unlocking Alternative unpinned nodes is used to update the glossary as the story progresses, for example to update a character entry. Multiple Unlocking and Parallel Threads are used to model sub-quests, where the completion of an overall quest initiates a number of sub-threads that can be completed in any order, but which all need to be done before the overall thread continues.


I have wondered whether sculptural hypertext might provide prototyping systems for game narratives, allowing them to be written, experienced, and refined without the need for a complete game engine, or rendered cutscenes. Most of the Witcher I think could have been designed in this way, but there are a couple of things that our current sculptural models do not accommodate. Firstly, the threads here are actually quests, and those quests have a name, descriptions, and other metadata (such as recommended level) that give the player a crucial way to manage them within the interface (as they select a current quest, thus focusing the player HUD on these particular locations and details). Secondly, there are a number of routes that can be taken to give the player narrative content, mainly this is dialogue and cutscenes, but also includes the glossary and bestiary. Our current sculptural models do not distinguish between different types of content or different types of node, and therefore cannot make choices based on this on what way content in a node should be presented to a player.

The biggest way in which narrative agency is expressed in the Witcher 3 is through dialogue choices. These can be modelled in sculptural hypertext as concurrent unpinned nodes (choosing a node to read unlocks different narrative branches) but this is a bit clumsy, and might indicate that the models we have need some sort of container node that acts as a kind of micro phase that the narrative slips into for a subset of the interaction (such as a dialogue), returning to the previous phase once the interaction is complete.

This analysis is very much broad brush strokes, but it does show how a grand narrative arc can be managed within a collection of smaller narrative segments, and driven forward by encounters and scenes experienced within an open world environment. It also supports the idea that Location-Aware Sculptural Hypertext systems might make good narrative prototyping environments for open world games.

In fact two of the designers at CDProjektRed gave a rather nice talk about the narrative design of the Witcher 3 at Big Techday 8 (embedded below), in which they give some really interesting details about designing the individual quests so that they work within the open world environment (considering location issues such as quest entry points). They also reveal some of the detail of the dialogue trees – which is considerable. The video starts in German, but is mostly in English.

It may be obvious by now that I am an unrepentant structuralist, mainly because that view provides so much fodder for computational modelling, so forgive me for eyeing the Witcher in terms of state changes and patterns. However, there is no denying that what makes the game narrative work extends far beyond the structural elements. Brave writing (e.g. The Bloody Baron), culturally grounded storytelling (e.g. Hearts of Stone), and an unusual and engaging protagonist – who is definitely not the everyman, give the narrative interest and weight. I was also impressed by the alignment of mechanics and reward with narrative progress (as XP and gear is awarded for completing quests, not for grinding), and the way in which narrative choices are seldom flagged, encouraging a genuine role-playing mindset.

Hamish Black of Writing on Games has produced a very good video essay (embedded below) that explores some of these other aspects of the Witcher 3 narrative, and is well worth a watch (btw – I was a tipper).

Finally, and perhaps most tellingly, it’s worth reflecting on how I felt once I completed the main game quests and was returned to the game world with the knowledge that the core quests and characters were no longer available to me. Mounted on Roach, and facing the path ahead, still rich with geographical detail but now empty of friends and companions, there was only one word that could describe it.

I felt lonely.

If nothing else, that shows how successful the Witcher 3 is in using narrative to breath life into its massive and complex world. Thank you everyone at CDProjektRed for all the care, attention, and skill that you put into this incredible game.