Algorithmic cultures in tabletop roleplaying games

Do roleplayers dream of algorithmic sheep?

This is not a peer-reviewed academic article.

This is the end-term essay for the Algorithmic Cultures seminar, a module within the broader Bachelor in Digital Media Studies.

Written by Alessandro Piroddi.

Submitted on the 22nd of May 2022.

Re-edited and published on Patreon on the 29th of August 2022.

Enjoy...

Introduction

When talking about algorithms we tend to first and foremost think about the digital ones stemming from computer sciences and applied programming. Likewise, when thinking about algorithmic cultures we tend to only focus on the social effects on human behaviour that digital algorithms can have: we think of Facebook, we think of face-recognition software, we think of big-data. This all feels new and unexplored, especially when we marvel (and shiver) at the unexpected and sometimes aberrant ramifications that algorithms can entail: racist face-recognition software, classist package-delivery services, racist and classist crime pattern-analysis tools, anxiety inducing social media.

But honestly, to me this feels like old news, a lesson we should have learned already, in a time when algorithms were less inflated, complex and inscrutable, so that now, moving forward, certain problems could be known, expected and thus avoided. Asimov’s sci-fi novels warned us about this kind of issues way back in the 50s, with stories of “defective” robots expressing “inexplicable” behaviour which invariably turned out to be the logical and predictable, albeit startling, result of product design and its interaction with the environment, a user or some other external variable (Asimov, 1950). In this paper I want to make a case for looking away from the electrical glamour of digital computation to instead consider a simpler but closely related form of analogic algorithm, one that could be used to model and understand more complex structures. I want to look at the algorithms at the root of tabletop roleplaying games (just RPG from now on): how they influence and are influenced by human behaviour, how their intersection with the digital age has allowed them to shape a variety of subcultures, and how they facilitate and reflect aspects of contemporary society.

Fundamental Concepts

This first section will introduce the core definitions and ideas needed to tackle the paper’s topic: What is an algorithm? What is an RPG? How are the two things related by the magic of design?

Algorithms

In its most general meaning, an algorithm is a step-by-step procedure for accomplishing some end (Merriam-Webster, 2022). This can take different meanings depending on the field of application. In mathematics it can be a formula to calculate a value, or a chain of formulas to solve a problem. In computer sciences it can be intended both as a description of the logic behind a piece of software, and as the programming code implementing the aforementioned logic; both instances are nothing more than the same step-by-step procedure, the former expressed in a human-readable language and the latter in a machine-readable one. A cooking recipe is, for all intents and purposes, an algorithm, and the same can be said of any text presenting a detailed procedure to accomplish some task: sheet music, ikea building instructions, the rules of a sport (Uricchio, 2017, 126-127).

Tabletop Roleplaying Games

As an activity, an RPG is a conversation about imaginary characters, places and events mediated by rules and mechanics. As an object, an RPG is a ruleset in the shape of a physical book or a digital document, expressing procedures that, when followed by the people participating in the game, will produce a specific play experience. Since the game is a conversation, defining who can say what, when they can say it, and how they can say it, is very important in shaping the fundamental structure of play. This core is often supplemented and enriched by additional elements, expressed through both textual and paratextual components, that further support and enforce the effect of the main design on the behaviour of the people playing the game.

Design

Algorithms are important because they are an inherent and foundational element of any design process: they are the plan, the blueprint, the instructions, the procedures. By thinking in terms of design it should appear obvious how the production of some artefact – a numeric value, a cake, a car, an activity – by way of a specific algorithm will have both direct and explicit outcomes but also a number of indirect and implicit ones, where the latter can often be more important than the former.

For example, there are dozens of possible sorting algorithms which could be used to craft a software that reorders the elements within a list, the direct result, but depending on their design – Which operations are carried out? In which order? Using which coding language? On which hardware? – each solution will also produce different indirect results. One algorithm has a simple logic but leads to many repeated operations, while another algorithm requires fewer operations but each is very complex. One algorithm has problems dealing with specific edge cases, while another is better suited to the handling of large amounts of data. Etc. These indirect effects will determine which algorithm is better for any specific use case (CrashCourse, 2017).

There are then many cases in which the real goal of a design is to achieve a set of indirect results, with the direct ones being just a means to an end. This is the case of the EdgeRank algorithm that Facebook uses. Its direct and explicit goal is to build and populate the News Feed segment of each user profile, but the true purpose of its design is to affect users so as to maximise their engagement rating, which in turn is a value designed to express and affect connected metrics such as session time, number of clicks, number of shares, etc (Bucher, 2012). The true goal of this algorithm is the extraction of value from Facebook users, with the production of a News Feed artefact being merely a tool to achieve an end.

What makes social networks and other complex algorithmic designs interesting and important is exactly their focus on indirect and implicit outcomes, which in many cases has, just like in Asimov’s stories, produced results that appear (at least to the users) aberrant and inexplicable: from the virulent proliferation on YouTube of junk videos “for children” that are quite child-inappropriate (Olson, 2017), to the obfuscation of LGBTQ+ discourse on the TikTok platform (Mulder, 2022), to the many ways in which Facebook causes harm in the name of company profit (Lima, 2021), to the more or less voluntarily racist, classist, sexist and overall problematic side effects of other software.

RPG Algorithms

In many ways, the algorithmic structure of RPGs resembles that of a social network. In their current form, most softwares like Facebook, Youtube and TikTok can be described as platforms for the creation and sharing of user generated content. Each post, each comment, each uploaded image and video, even the user profiles themselves, are all content. The social element is used to turn the content one user generates, into the content another user consumes, thus defining each user as both actor and audience at the same time. This dual nature turns the user from being a passive consumer of static artefacts resulting from the actuation of an algorithm, into a live “system variable” within the algorithm itself, establishing a feedback loop of mutual influence.

In the complex system of a social network, the process of running an algorithm doesn’t change the algorithm itself but only the inputs and outputs it handles, which are all inherently affected by the algorithmic mediation: an image is shown to some users, their interactions with it constitute a new input, which informs a new output in the form of the same image being shown more or less or differently, prompting new interactions, etc. There is no end. As long as the user engages with the platform, the loop runs, producing new results, which are then saved for future interactions, thus virtually continuing the same loop with its evolving identity.

Structurally, this is uncannily similar to how RPGs work. The game procedures constitute a complex system of self-affecting inputs and outputs where users are both actors and audience. The main difference is that in the case of social networks the algorithm runs on hardware, while in RPGs the users are the hardware running the algorithmic instructions, in addition to being variable agents within its parameters. Other than that, the procedural flow is very similar:

• A user generates some content – “My knight speaks to the crowd, demanding answers.”
• The algorithm takes the input and parses it for a number of conditions:
  • Is the input inherently valid?
  • Is the input contextually valid?
  • Does the input trigger a specific subroutine of the main algorithm? (combat rules, social rules, perception rules, special character rules, specific situation rules, etc)
• As a result, there will be an output – “Everyone shifts nervously until the crowd rallies behind a fat man, slowly walking towards you.”
• The users now react and interact with this output, again according to their end of the algorithm: content validation, identification and execution of subroutines, final output resulting in new content – “My knight observes the newcomer, head to toes, letting silence linger before saying anything. What kind of person does he look like?”

This structure is not only similar to that of a complex self-affecting digital algorithm, but is also being studied as a more easily understandable model to research and improve upon current digital Procedural Content Generation technology. In this view, systemic random noise is provided by the unpredictable choices and actions of users but, instead of being purely arbitrary, the process gets biassed by the game mechanics and rules. This biassed noise leads to the creation of content seeds (fictional characters, locations, events, etc) which will in turn further bias the system in a way that produces a space of possible outputs which feel intentional, coherent and adherent to an intended canon, rather than a disjointed mass of unrelated chaotic results (Guzdial et al., 2020).

Local Instances

The cultural impact of RPGs has grown and changed in time, partly due to the advent of the internet, and partly due to the reciprocal influence between the media and its consumers. To present this, in the following sections I will base my assertions both on personal experiences and observations, and on data coming from the series of historical publications by Shannon Appelcline: Designers & Dragons. 

The first RPGs were homebrew creations evolving out of the tabletop wargaming hobby. Because of this, they catered to a very niche audience: veteran gamers willing to decipher a large amount of complex and unclear instructions for the sake of playing a game. Although only one person within an average group of 3 to 6 people needed to be such an hardcore gamer, the hobby was still very limited in its reach. This meant that, through the 70s and 80s, the vast majority of RPG users were organised in small insular groups with little or no contact with others. Even in such small numbers and in relative isolation, the social-network-like properties of RPG algorithms were in effect, producing a “table culture” where each individual group would display its unique game-relevant jargon, traditions and history, meaningfully differentiating how the same algorithm (the same RPG) would be parsed at each table. This stage of RPG cultural history can easily be compared to a local instance of a decentralised social network, like a single Mastodon server. Only with the advent of the internet the isolated tables would really begin communicating with each other, sharing knowledge, stories, techniques and “code”, engaging in social identity mechanisms that would turn a galaxy of individual “clans” (a single table) into relatively homogeneous “tribes” (all the clans playing the same RPG). This was possible because most RPGs in the past used to be “forever games” in which every table would almost exclusively play the same RPG. As time passed this started to change, but even with a more diversified game diet most users would easily identify as primarily players of a specific RPG, keeping each clan anchored to a certain tribal identity and its (increasingly) shared culture. Only in the past couple of decades RPG cultures have ebbed, flowed and mixed enough to allow for a third, more meta, level of culture and identity to emerge where different tribes would coalesce into “nations” representing not individual games, but the underlying design philosophy at the root of different groups of games.

Algorithmic Affect

Let’s now look at a concrete example of how the design of an RPG algorithm can influence the behaviour of its direct users and, by extension, both express and affect the culture in which they participate. I’m going to use Dungeons and Dragons (D&D) as my main object of scrutiny, it being the most sold, played and famous RPG in history, worldwide.

The Reward Cycle mechanic of games such as D&D establishes (among other things we’ll not get into at this time) when and how a fictional character belonging to a player (PC) would accumulate Experience Points (XP). Once a sufficient amount of XP is accrued, the PC would reach a new level of experience and be rewarded with various improvements. In the 1974 version of the game, known as Original or OD&D, the main source of XP is the collection of treasure, where each gold coin worth of value grants one XP. Another source of XP, mostly a vestigial leftover from the tabletop wargaming ancestry of RPGs, is the slaying of creatures, with the reward being far inferior than that of treasure while the risk of failure and game-over (PC death) is much higher. The direct design effect of this structure is the crafting of a way to track the relative success and improvement of a PC, a meter counting points and delivering occasional treats. The indirect design effects are far more interesting:

• It clearly signals to Players what the game considers important.
• Within play, it encourages Players to seek adventure rather than slaving away performing a safe but boring pseudo-medieval job.
• Within an adventure, it rewards exploration and cleverness over violence and lazy-thinking. Avoiding direct conflict whilst pilfering valuables and surviving until the next safe area is arguably the core loop of the game.

But other unexpected behaviours can also arise as users start figuring out how to game the algorithm for their personal ends and purposes. One (in)famous example is the phenomenon of “murder hobos” where some players would find it easier and safer to rob and murder unsuspecting villagers than to venture into dark caves filled with monsters. These groups of fictional adventurers would commit fictional crimes while roaming from one fictional village to the next, increasing their power level before facing the challenges of a real adventure. Within the OD&D play culture this was unintended, but not necessarily problematic, as game characters were mere pawns allowing their players to interact with a fictional world full of challenges and opportunities. 

Cultural Feedback Loop

But with the passing of time, the hobby evolved. Story, once just a secondary dress-up element meant to set the scene for play action, became more and more important and central, with users trying to get “a good story” out of the old algorithms, affecting play in such a way as to create unique and peculiar experiences: some absurd and unexpected results would become, decades later, urban legends and memes about the strange old days, while at the same time some desirable and appreciated results have become the idolised fetish of whole subcultures devoted to bringing back in vogue the good old days.

So game culture changes and users find ways to squeeze new results out of old algorithms. This in turn prompts game companies to modify their algorithms to reflect and support new trends. This in turn affects the culture in many ways: some embrace the change, some reject it, some do unexpected things with it. Staying with the previous example we can see how across many years and many editions – eight editions in total counting the Original, Advanced, Basic, Advanced2, D&D3, D&D3.5, D&D4 and D&D5 – the focus shifted away from exploration and survival towards a newfound desire for power fantasy escapism, where survival was less of a reward and more of a given, the action was less bookkeeping-y and more bombastic, and the PCs were not just adventurers but actually “heroes” within an “epic” story. To this end we can observe how the XP rewards for slaying creatures increased, reaching and then eclipsing those for treasure, until XP for gold was not a rule anymore. Later still, we see the introduction of systems of XP for challenges, rewarding the mechanical overcoming of opposition no matter the method, and even systems of XP for milestones, rewarding the achievement of story-related goals and statuses.

These changes, along with many others, curbed the murder hobo phenomenon: partly because it was not effective anymore, as the XP granted by the wanton killing of common people was immaterial if compared to the XP amount provided by even low-level monsters; partly because the culture came up with tactics and “wisdom” on how to deter and handle such undesirable behaviour; partly because the culture’s newfound focus on story infused all game characters with more meaning, making their arbitrary killing a more questionable and problematic activity.

Algorithmic Ideology

Design conveys ideology. In the case of D&D, and many other RPGs inspired by it, this used to be one that nowadays is being criticised and actively addressed both culturally and through new algorithms. Whilst the textual message of D&D is quite positive and at times even uplifting, the subtextual messages expressed by its mechanics are more problematic:

• The idea that violence is the better and most expedient solution to all problems, with combat rules being prominent and more engaging (except in the oldest editions).
• The colonialist idea that it is OK to plunder other people’s art and relics.
• The racist approach to fantasy people, where all individuals of a certain “race” are seen as essentialist stereotypes rather than unique persons. It’s OK to slaughter goblins, for example, because they are all defined as inherently evil, they are never portrayed as children, or as conducting mundane and reasonable daily chores, or as expressing positive and relatable emotions. This way their murder can’t ruin the “fun” by questioning the PCs’ (and the Players’) morals.
• The ableist, sexist and hetero-centric portrayal of stories and characters, often full of male gaze and toxicly masculine tropes and stereotypes.

Arguably, none of these stem from intentional design. They are an understandable byproduct of the culture of their time, being reflected through the algorithms of the game. It is telling, though, that in the past decade the hobby in general, and D&D in particular, has tried to address and fix these problems, trying to be more aware and intentional with their designs, trying to focus and improve on all the positive values that RPGs have always expressed. It is also telling that these attempts have stirred more than a few heated arguments, often dividing RPG users across real-world political lines. 

An egregious example of this is the veritable “battle for the soul of the OSR”. The Old School Renaissance is an RPG subculture born around 2007 with the stated aim of continuing/rediscovering the traditions of the oldest D&D editions (Hann, 2021). Somehow this movement ended up attracting, over the years, an elevated number of users with far right views – from your average old white guy whose too conservative views are sometimes unsavoury, to obvious alt-right activists, to overt fascists and even bona fide neo-nazis – which somehow found the old algorithms to be a gaming media more welcoming and at times conductive for their personal views and ideologies. This became slowly more apparent and aggravating, tarnishing the OSR movement reputation as bad actors that were actively ejected from other RPG cultures found refuge in OSR-adjacent ambients. This in turn has, in recent years, prompted various OSR communities to take an openly political stance against fascism in order to reclaim safe and hate-free spaces, reappropriating the old games and traditions as vehicles of positive entertainment and non-toxic content by influencing the culture surrounding them: same algorithm but new and more aware user-input, less nostalgia and more inclusivity.

Conclusion

I hope this paper could succeed in showing how RPGs are truly an analogic and human-computed form of algorithm which closely models the complex self-affecting structures of social networks, to the point of being the cornerstone of diverse and vibrant algorithmic cultures. For this reason, further study of RPG design and the social phenomena stemming from it would not only be valuable and interesting in and of itself, but could prove useful in the study of digital-based phenomena, offering a simpler and more transparent model to work with.

Reference List

Appelcline, Shannon (2014). Designers & Dragons. Volumes: 1970s, 1980s, 1990s, 2000s. Silver Spring, Maryland: Evil Hat Productions.

Asimov, Isaac (1967). I, Robot. New York: Bantam Books.

Bucher, Taina (2012). "Want to be on the top? Algorithmic power and the threat of invisibility on Facebook." New media & society, volume 14, issue 7: 1164-1180.

CrashCourse (2017). Intro to Algorithms: Crash Course Computer Science #13. Published 20.05.2017. Last accessed 16.05.2022.

Guzdial, Matthew, Devi Acharya, Max Kreminski, Michael Cook, Mirjam Eladhari, Antonios Liapis, and Anne Sullivan (2020). “Tabletop Roleplaying Games as Procedural Content Generators”. International Conference on the Foundations of Digital Games. New York: Association for Computing Machinery. Article 103: 1-9. https://doi.org/10.1145/3402942.3409605

Hann, Keith (2021). A Historical Look at the OSR — Part V. Published 16.12.2021. Last accessed 20.05.2022.

https://osrsimulacrum.blogspot.com/2021/12/a-historical-look-at-osr-part-v.html

Lima, Cristiano (2021). A whistleblower’s power: Key takeaways from the Facebook Papers. Published 26.10.2021. Last accessed 16.05.2022.

https://www.washingtonpost.com/technology/2021/10/25/what-are-the-facebook-paper...

Merriam-Webster (2022). “online dictionary : algorithm”. Last accessed 05.05.2022.

https://www.merriam-webster.com/dictionary/algorithm

Mulder, Mia (2022). Self Censorship And Tik Tok. Published 29.03.2022. Last accessed 10.05.2022.

Olson, Dan (2017). Weird Kids' Videos and Gaming the Algorithm. Published 23.11.2017. Last accessed 10.05.2022.

Uricchio, William (2017). "Data, culture and the ambivalence of algorithms." The Datafied Society. Schäfer and Es. Amsterdam: University Press. 125-137.