premDAT

The blog's most assiduous readers heard about it three years ago, in an improvised post: I have been working for some time on a platform for comparing Player Characters (PCs) from tabletop RPGs. premDAT, said platform, has now reached a stage one could describe as beta, that is to say an early prototype, but already partially usable.

In this case, premDAT will ideally allow me to complete my thesis. If you want to help me with this goal, I will be grateful!

What is premDAT for? As RPG players, we are often confronted with two contradictory discourses about our characters:

1. On the one hand, we want to embody individuals as different from us as possible. These characters should also be very varied, to avoid any redundancy betraying an (almost shameful!) limit of our imagination. We perceive this distance as a guarantee of rich roleplay, as the possibility of getting into someone else's head, or even as an emotional safety net to avoid growing too attached to our PCs.
2. On the other hand, we admit that our characters are inevitably limited by our own personality, that they can only exist within the scope of what we can invent. How can one envision a character that one precisely cannot conceive? Expanding our repertoire is only possible through contact with other players, but here too, understanding what characterizes a character only happens through our "prisms". This tendency is more pronounced in some individuals, who sometimes consciously assume one (or several) favorite "PC type(s)": the bruisers, the summoners, the bards, and so on.

The notion of prisms seems important to me, as it determines how we define our characters. Some prisms are recurrent in practically every role-playing game: these are the sacrosanct "characteristics" (and their cousins, the "skills"), imposed by the rules, which allow us to identify a character as more or less "strong", "agile", or "intelligent". Some games also ask you to take a stance on more exotic elements, like L5R and its "game of 20 questions", which has us for instance assess each PC according to their loyalty to their family, their clan, or their friends. Finally, other prisms are our own, inherited from our previous gaming or literary experiences, or even simply from our way of sizing people up in everyday life. For example, some will almost systematically choose to quantify their characters' "level of madness", while others will never fail to mention their piety.

---

So in June 2017, I laid out all my characters (those I could still remember), to try to extract their "founding prisms". It appeared to me that some prisms work like an ordinal scale. A character can be "not at all brave", "a little brave", "fairly brave", or "very brave". These steps can clearly be ranked, without being able to objectively quantify the "distance" between each. Other prisms lend themselves less well to the exercise, like "military": either you are in the army or you are not, but there is no real in-between. To distinguish these two types of "prisms", I named the former "dimensions" and the latter "tags".

The highlight of the show was classifying my characters across all these prisms to compute their similarity. At the time, we were talking about 14 dimensions and a good dozen tags, on the basis of which I compared every pair of characters to establish their resemblance. The details of the similarity computation will not be discussed in this article, but the curious can discover its inner workings here.

This similarity score was then displayed graphically thanks to a "force-directed graph", a representation that is still in use on premDAT: characters are encoded by circles, their similarity by a curve connecting them. To show only the relevant similarities, the platform hides the weakest ones according to an arbitrary threshold that users are free to adjust to their liking. The radius of the circles encodes the number of sessions of a character, because a PC played over 60 sessions leaves more of a mark than its cousin destined for the trash bin after a one-shot. Finally, the colors indicate the level of a dimension: in my first prototypes, I used a diverging scale from pink (low) to green (high), but I ultimately opted for a sequential scale (dark red to light red) following numerous remarks on both the aesthetics and the readability of the colors. Finally, each character has a "radar graph", or "spider graph" (which sounds cooler), showing the values of its dimensions, the latter now being dynamically sorted by their proximity across all characters (a graph displaying sudden gaps therefore denotes an assortment of unusual dimensions).

---

premDAT is now a use case that must serve my thesis, which deals with interaction methods with data visualizations and, more specifically, visualization annotation systems. I need people willing not only to enter their PCs into the database, but also to participate in the analysis of the data thus captured. Should you wish to take part in the experiment, I would be grateful if you registered on the platform and tried to find your bearings there.

premDAT is no longer just a lonely graph. It is a web platform on which you can store your characters, but also the adventures they took part in and the games that served as their framework. I tried to make the experience enjoyable, with among other things a few features aimed at sparking the desire to participate, or at reducing the effort required:

You can spare yourself yet another login by signing in through Discord (as I write these lines, Twitter does not work properly);

You can archive your adventures, structured into sessions, and use these pages either to present your stories to people outside the platform, or to expose your exploits to the other participants, or even link them to your websites;

You can showcase your characters, develop their background, consult their similarities with others, or even create "slide"-style presentations to share on the web;

And when the visualization is fully implemented, you will also be able to discover your own role-playing habits, see whether they match your expectations, or even find some inspiration for your future PCs.

---

If you want to contact me about this experiment, to give me your feedback, report bugs or suggest ideas, you can do so on this forum thread.