Mr. S.C.Rheber Became A Videogame!

Let me introduce you The Strange Forgeries Of Mr. S.C.Rheber ONLINE! (click here to check the paper version)

At the time I am writing this article (February 2023) I can only grant private sessions to test the features, but

if you are interested in playing, let’s meet on my Discord Channel.

You are invited to join! click here and write your first post: I am happy to organise a game session for you and your friends where I explain you the rules and let you play. If you don’t have Discord, just post a comment here and we can find a different solution.

Once in the digital medium, Mr. S.C.Rheber had to change a bit: the paper version is a fast game, but the digital interface made it slower, unfortunately. However the access to

countless 3D assets at no expense to production allowed me to create a new way of inventing and solving puzzles.

If in the paper version, the puzzles to be solved involves any black-and-white sketch like those:

In The Strange Forgeries Of Mr. S.C.Rheber Online, players will create puzzles that look like Lovecraftian aberrations of an elderly gentleman, with his beautiful white moustache and beloved lipstick. Here are some examples:

I love them…

This is how an entire 7-players setup looks like:

Since there is no limitation in terms of cutting die, I customized the layout a little bit:

Tabletopia is a platform where you can play digitally many board games which are already available in paper form: it is a 3D space where you can move components on boards, roll dice, draw cards and everything else.

Discord is probably the simplest system in existence to talk to a group of people via chat, video or voice: for example, to create a place where you can talk with your friends while you are playing Mr. S.C.Rheber online.

But let’s get to the heart of the matter.

What differs The Strange Forgeries Of Mr. S.C.Rheber Online from the boxed version?

I typically advrtise two key features of The Strange Forgeries Of Mr. S.C.Rheber:

1.  Mr. S.C.Rheber is a logic game where a particular player (the GM of the game, called the Gallerist) can invent one of the rules. It doesn’t matter which rule it is, because the scoring system balances things out by itself: if the rule is complex or ambiguous (and the other players realise this) then the Gallerist loses, if the rule is simple and innovative (and the other players recognise this) then the Gallerist wins. For those who wish to explore this further I have written extensively about the genesis of this mechanism in this article.

2. Mr. S.C.Rheber in Gamefound’s version is optimised such that the rule to be invented can refer to any feature of a freehand drawing. Yet there is no need to limit oneself to a black-and-white sketch. In fact, as a consequence of (1) the rule can refer to anything: do you want to use Lego blocks you no longer need and send photos on Whattsapp to the Gallerist to judge their authenticity? Do you want to use the composition of the pieces on a chessboard instead? Do you want to use an art generator like Midjourney and use the emoji of an O and an X to bet on the originality or falsity of the bot’s image? All of this is theoretically possible, but the first version of The Strange Forgeries Of Mr. S. C. Rheber will not be optimised for anything more than freehand drawings while the game’s first expansion will attempt to explore the game’s mechanics in the context of a Cypher Book (I wrote about this in a past update).

Tabletopia, on the other side, has a major constrain and a distinctive features, which are, respectively:

• Freehand drawing is not supported

• But the number of components is virtually unlimited

So I wonder how I could leverage on the flexibility of (1) and (2) to adapt it to the new medium and I immediately thought of games like Xybrid, but with the versatility of Imagine.

In XYbrid, players compose the head-body-arms-leg segments of a hybrid monster by overlaying transparent cards on which the opaque drawings of those body parts are printed.

In Imagine, the act of composing something is independent of the head-body-arms-legs scheme and players are encouraged to create small animations.

This video game version of The Strange Forgeries Of Mr. S. C. Rheber will have a very similar system: superimposing a finite number of tiles in an infinite number of combinations without worrying that a gust of wind will blow them all away (more or less).

Here are the tiles from the game:

and here 23 of them had been combined to compose 4 Lovecraftian aberrations:

In an average match of Mr. S.C.Rheber players produce each around 20 different sketches: if the physical version would use tiles as the digital version, it would need to feature more than 600 tiles! Is also true that classification of free-hand drawings requires more trial and error and would require a certain speed of execution that a digital version of Mr. S.C.Rheber will hardly achieve.

In other words Mr. S.C.Rheber “Paper” and Mr. S.C.Rheber “Digital” are the same game, but they are optimized for completely different puzzles and unleash different type of creativity and strategy: the first is more similar to pure scientific heuristic (you can read more about from my article here), while the second is closer to the Raven Progressive Matrices of an IQ test.

Therefore is natural for me to wonder which New Mr.S.C.Rhebers can be invented thanks to the digital medium.

 Xybrid and Imagine are not the only things that came to mind in the context of a videogame version of The Strange Forgeries Of Mr. S.C.Rheber. The time has come to talk about the grotesque Cadavre Exquisite of the turn of the century, the procedural generated titans of Dwarf Fortress and the brand new text-to-image models.

A Cadavre Exquisite is an artistic technique in which several artists each compose a separate section of a drawing, but only know certain details of the other artists’ drawings: for example, by folding a sheet of paper into four sections, Jake and Dino Chapman achieved this in 2000

Imagine is nothing more than a Cadavre Exquisite made from cards.

Tarn Adams and Zach Adams, the designers behind Dwarf Fortress had developed, on the other hand, a hyper-complex version of XYbrid’s head-body-arms-legs segment system, but they only work with textual outputs like the one below which introduce a procedural generated Forgotten Beast to the player

Graphically those monsters are only a letter “T”s moving around the map, so the eclectic Dwarf Fortress community has created countless artistic interpretations from the game’s promts: you can find a full gallery of those beautiful interpretations in the official DF Wiki, while you can admire the artistry of my very own “humanoid figure has nostrils shaped like tarantula chelicerae and ears shaped like duck beaks, it has tarantula legs shaped like intestines and its body is covered in human skin”… bleach

An image&text-to-image software such as Dall-E or Midjourney is, in simple terms, the act of removing details from an image chosen by the user by blurring it and then refocusing it step by step by trying to add details that are congruent with the assigned text promt: an algorithm tasked with recognising things within an image will judge whether at each step, the software-drawer is drawing something similar to the assigned text promt, while the software-drawer will try to please its judge as much as possible at every step. For example, the software Disco-Diffusion took the following drawing of the Clockwork Baron

It blurred it a bit and then tried to redraw something trying to stick to my promt “a colossal worm having each of its hair as sharp as a knife and multiple human hands crossing its body”. Typically he will draw a bag of grass, grasshoppers or the legs of a centipede:

This is a kind of Cadavre Exquisite, but instead of relying on partial knowledge of a drawing in a section adjacent to the one it has to complete as is the case with a classic Cadavre Exquisite (although Open Ai’s DALL-E has just released a feature structured exactly like a classic Cadavre Exquisite), it will rely on partial (blurry) knowledge of a drawing in the same section it has to draw.

Of course, it is inevitable not to think of the ingenious prompts generated by the crazy minds of the Adams brothers and a Cadavre Exquisite made up of thousands of artists such as the latest version (number 4) of Midjourney, which is probably one of the most beautiful image&text-to-image I have ever seen.

And, in particular, this is what an image&text-to-image would look like starting with my drawing of Mr.S.C.Rheber’s Clockwork Baron (the same above) and the textual promt of a procedural generated titan from Dwarf Fortress (or something I made up).

I asked Midjourney to draw me…

“a colossal worm having each of its hair as sharp as a knife and multiple human hands crossing its body drawn in the style of this picture https://beyondwords.games/wp-content/uploads/2022/12/Green_Collector_4x5.jpg“

“a totem having multiple reptile heads, peacock feathers are extending from the mouth of the reptile heads like tongues, drawn in the style of this picture https://beyondwords.games/wp-content/uploads/2022/12/Green_Collector_4x5.jpg“

If another videogame version of The Strange Forgeries Of Mr. S.C.Rheber will exist, I would like it to exploit this technology in the context of a game about pattern recognition, but it must be said that there are dark implications behind the beauty of these images and it involves the definition of an artist and, in my opinion, the definition of a game in itself.

The assimilation of these models to an Cadavre Exquisite made of thousands of artists is by no means coincidental: the model was created using a Generative Adversarial Network (GAN), i.e. a two-AI-player system with some interesting analogies to the relationship between Mr.S.C.Rheber’s Collector and Gallerist.

In this game the AI-Gallerist is called Discriminator and invents a rule like “contains eyesORis standing next to a traffic lightANDtalking on the phone” or, more generally, like “<this>AND/OR<that> AND/OR… AND/OR<whatever>” where the number of AND/OR is probably close to a few billion. With this rule, it is able to classify millions and millions of images of artists on the basis of relevant textual descriptions (provided, let us say, in the title of the image or its alt-text): these millions of images are as unknown to the AI-Collector as are the 12 images of the Artistic Manifesto to the Human Collectors in Mr. S.C. Rheber‘s game.

The AI-Collector is called the Generator and will try to guess this first rule by finding a second rule of a different nature such that if a design is produced respecting the second rule to the letter (which has to do with “how to draw a thing”), then that design respects the first rule (which has to do with “what is drawn”). The design will be created using the aforementioned method of blurring-deblurring, which is more properly called Diffusion.

The AI-Gallerist/Discriminator must not allow the AI-Collector/Generator to guess that first rule, and each time a design by the Generator will fit that very first rule invented by the Discriminator, it must change the first rule in such a way that it continues to be true for the millions of designs hidden in its colossal Artistic Manifesto but false for each design by the Discriminator. Even in Mr. S.C. Rheber it is possible to change the rule invented by the Gallerista posteriori, but the action is limited to a finite number of times. In the GAN game between AI-Gallerist and AI-Collector this can be done a virtually infinite number of times. The game, however, is engineered for the Generator/AI-Collector to win by “checkmating” the Discriminator/AI-Gallerist: that is, when it is no longer possible for the Discriminator/AI-Gallerist to make a move that allows him to win… in this case the drawing method (embodied in the second aforementioned rule) developed by the AI-Collector always creates designs that are unrecognizable from the designs of the Artistic Manifesto.

The problem is that that Artistic Manifesto is not an AI-Artistic Manifesto: it was drawn (and classified with a text prompt) by someone else who does not belong to the game described. This someone else is typically an artist who owns a portfolio of his work on various online platforms such as Behance or DeviantArt. In some countries (like mine: Italy), there is no need to make it explicit that the use of an artist’s image can only be used after the latter’s consent, since this is implicit:

the reproducibility of a work of art is the author’s right, not the viewer’s.

It is true that Midjourney’s software product is authentic, as Oink Games does not sue its players for copyright infringement if they play the Modern Art game they produced or publish content using the game components they purchased*, but it is true that the works of Piet Mondrian printed on the Modern Art cards are for public use, while the Artistic Manifesto of the game of different GANs designed to train Midjourney or DALL-E probably is not.

“It would be cool if images had metadata embedded in them about the copyright owner or something. But that’s not a thing; there’s not a registry. There’s no way to find a picture on the Internet, and then automatically trace it to an owner and then have any way of doing anything to authenticate it.” (David Holz, Midjourney founder)

I would love to see a version of The Strange Forgeries Of Mr. S.C.Rheber created using this kind of products, but first of all every game has to be fair… every player has to know they are playing and decide freely whether to accept the challenge.

*For copywrite and trademark infringement in the context of games (and specifically about Wizard Of The Coast’s OGL) you can check this video from LegalEagle