Introduction

This is was one of my main interests, between 2016 and 2020, but it's still something that I want to explore deeply at some point, especially when applied to video games, but also more generally. It basically started when I took a Complex Networks course in university (about computational networks, multi-agent simulations on them, and game theory), and then read the book Evolution of Language, an edited set of paper on the evolution of language. This book had a huge influence on me, in particular the chapter by Simon Kirby on an Iterated Learning Model where fairly simple rules led to the creation of structured (recursive) language.

If you put the two together, you have a model to evolve structured art forms, and a way of relating them in space (through networks). Of course, it's easier said than done! Especially when dealing with arts and crafts, you need to add aesthetics in some way for it to look interesting to real people and not just a random assortment of bits.

Some other people whose work in the area I find really interesting (though more on the sociological side than the artistic) are Peter Turchin (on history), Peter Richerson and Robert Boyd (the de facto pioneers of the field), David Sloan Wilson (on religion), and I'll add more as they come to mind.

Also extremely important, though perhaps a bit tangential, is Milman Parry and Albert Lord and the latter's book Singer of Tales (see Epic Poetry below) which was extremely reinforcing to these ideas, showing a very clear real world example despite being entirely unrelated to the theory.

Iterated Learning

Iterated Learning is essentially the game of the Broken Telephone (or Chinese Whispers). Basically, you want to propagate information through agents: the first agent produces something, the second tries to learn that something (hopefully faithfully) and passes it onto a third. Things become interesting when the things you're communicating are perhaps too complex to learn completely. Then, you need to figure out rules to simplify the system, or accept imperfections. These imperfections can build up and lead to noise, or can lead to a more learnable system, because of its structure.

You can also add an element of expression. Imagine you're trying to learn, say, how to play the blues. There are thousands if not millions of songs and interpretations. You don't get to learn (or reproduce) them all, not even close. You typically find a smaller set that you love the most, and you develop a certain style based on that. Both the songs that are learned and the method through which they are learned is important and can lead to vastly different art forms as styles slowly (or perhaps rapidly) drift apart and are redefined.

Network

Simon Kirby's articles were mostly about a single chain of agents, not a "real" network. That is, you have a genealogy of single parent and offspring. The parent produces, the offspring learns. On the next iteration, the offspring becomes the parent and spawns its own offspring and the process repeats.

This can be generalized to a whole network, with a lot more complexity, too, if you add different kinds of relations. Perhaps there is a strong familial component, and so children still learn mostly from parents. Or maybe this is all about friends, played on a different kind of local network. Or maybe there are established schools which decide what to teach on a national level. Or maybe there are some really popular pieces of art that are copied everywhere just because people like them and end up being more influential.

You can imagine all kinds of situations where a network can play a role. The kind of network itself will influence what is learning and how things "evolve".

Artists and Critics

This is related to the network "quality", and taken from Geoffrey Miller and Peter Todd. The basic idea is that a network has not online artists, but also art critics - or consumers.

This can add a lot of complexity to individual agents, but can also be done simply to achieve some ends - for examples, you can distinguish a conservative art form from a more fluid art form by adding an element of "surprise" to the critics (which may be every agent). Conservative forms will be positively rated when they fit the previous patterns, less conservative ones will want to be as different as possible (though still built form similar principles and not entirely random).

You can also have central critics (say, TV personalities or Youtubers or whatever) whose opinions have a bigger impact on the network, and so on. The essential idea, however, is to have a kind of "steering" mechanism that dictates how the art forms move.

Cultural influence

Finally, and this is one of my favourite aspects of the whole model, comes cultural influence and exchange. Some cultures can live side by side and have completely different values and art forms, while others are far more homogeneous.

This is usually due to historical accidents or deliberate politics (sometimes atrocious, something less so). But there are interesting dynamics at play.

Consider Rome and the Celts or Germans. Those had very different cultures, and there may have been an interest in maintaining the distinction (more on the part of the Romans). No self-respecting roman would wear pants and have long hair like the barbarians. Rome, though, being a huge a rich city, would be more attractive to nearby peoples in many cases (as many Germans tried to emulate Rome's way). The direction of influence isn't always the same, either - the wonderful icelandic sagas and the Poetic Edda were preserved by exiles from Norway, who refused the christian monarchy that was beginning to take place there and probably focused their development on the oral pagan tradition.

Art is often grounded in history and politics (like it or not), and I think this could be a very interesting path to explore, especially in strategy games. Political relations can end up affecting how two nearby art forms evolve, if the agents feel more or less enmity between them. Bad relations may move them further apart, good relations closer together. If you are a king or president, you can also choose to implement policies that bring different cultures together (or stamp them out, or try to maintain heterogeneity). You may believe that uniformity is required for cooperation and enforce indoctrination, or perhaps you are simply distrustful of the world of larger politics and decide to close your borders (like Tokugawa Japan).

I think this increases the breadth of expression of a grand strategy game like Crusader Kings.

Agents and Humans

Despite my focus on procedural generation and video games, there is the more human aspect at the core of all this. I don't mean programs to replace people any time soon, obviously, but games necessarily put forward a world view, and I think this is a valuable one.

Beyond its application to single-player games, ideas like these can be used in multiplayer games or different kinds of social media. Perhaps art can take a more open approach than the typical idea of the lone artist (which is probably wrong in most situations despite being the most salient and high profile).

And finally, at the end of the day, all of these models are descriptive rather than prescriptive. They are ways of looking at the world, a different lens. There is no rule to say what you can and cannot do - you take whatever you believe valuable.

Real World Examples

Obviously, cultural evolution, of the kind I'm writing about, is just about everywhere you look. That said, there are some specific examples I find interesting.

Architecture

Architecture was the first domain in which I thought to apply this model. The idea came from Age of Empires, where the various cultures are visually identified by their architecture, mostly (with some forced standardization like the Mongols' walled cities). I envisioned a kind of 4X or City-building game where each culture could have its own living architecture, reflecting different their diplomatic and trade relations as well.

Architecture is especially interesting because it is very lasting (at least when built in stone). We can get a longer glimpse into the past and see how it changes in time (in the western tradition, with Greek and Roman architecture being similar, then to Romanesque and Gothic, then Renaissance going back to earlier models, Baroque taking ornamental detail to the extreme, and so on), but we also see differences very clearly in space (consider Japanese, Chinese and Korean architecture) and sometimes the various influences (For example the Khmer and modern Cambodian case of Indic and Chinese influence, or how Hindu architecture and Islamic architecture were hybridized in India). The examples are everywhere, and are beautiful to see!

Another aspect that favours architecture is that it is composed of fairly discrete elements with continuous variations - you can look at columns, roofs and stories at the highest level, then descend to capital, shaft and column (for columns), and so on. This makes it more controllable (one culture can borrow, say, columns wholesale, and vary the roof slope only slightly) though still a big undertaking, of course.

Miniature Painting

This is a kind of painting seen mostly in books, often featuring patterned textures and weird perspective. Just as interesting is the cultural relations - perhaps taken from China to Persia by the Mongols, where it developed in different ways. it further specialized by the Ottomans (Turkey) and Mughal (India) examples. There are also "western" miniature paintings, mostly for illumination of holy books, but I am not sure whether they have a common origin.

What I love about it is that it is that it is not only a very clear example of art being tied to political relations, but it is also a fairly specificialized form that nonetheless finds its own local flavor wherever it flourished.

Epic poetry

Epic poetry is one of my favourite examples, but also one of the most elusive - especially when you consider the oral epics.

Most people are familiar with Homer's Iliad and Odyssey, because they were written down and started a whole new tradition - that of the literate epic. This kept going with the Hellenistic Argonautica, the Roman Aeneid, to the medieval epics (Song of Cid, Song of Roland, and so on) and the later countless "national" epics (like the Lusiads, La Franciade, the Vietnamese Tale of Kieu, the Chilean Araucaniad and many many others). The Indian epics Mahabharata and Ramayana play a similar (but perhaps even stronger) role in India and South-East Asia.

That said, epic poetry was often oral and, to some extent, improvised, as described in the wonderful book "Singer of Tales". It was also extremely wide-spread. The Germanic and Scandinavians had this tradition (culminating perhaps in the Norse Poetic Edda and the German Nibelungenlied), and so did the Irish (seen in the written version "Táin Bó Cúailnge" and other cycles). It's also very active still in Turkey and Central Asia, though less accessible to me. The main problem is that they are often sung in small villages and don't get written and translated for the benefit of the internet!

The structure of this is ripe for cultural influence, though, be it in the themes (which battle are fought, against whom), the customs they portray, or the music they accompany. There are always many versions of the "same" story circulating, and the story may never be told in exactly the same way twice - but this is exactly what one wants when simulating culture. Constant tweaking and reevaluation (sometimes aesthetic, sometimes political, or whatever else). It is, granted, perhaps a bit harder to program at the moment - or maybe poetry is simply not my area (also true)!

Shadowplay

This one is more recent, and about shadow puppet theater. Specifically, the Indian and South-East Asian varieties. I became interesting in Gamelan music, and found out it is extremely common across South-East Asia (From India to Thailand and all across Indonesia). Like miniature painting, it is another example of a specialized art form that nonetheless is full of local flavor! You can see, for example, how the puppets themselves look very different in Java and Bali, which are extremely close together in space - the reason being perhaps Islam, which was against human representation in art; the puppets from Java, where Islam took hold, are a lot more stylized, whereas in Bali the puppets have more human-like.

It is also a compound art form, with the puppet design, the music and the stories themselves being free to vary across places. Javanese and Balinese Gamelan are different, but still very much about metallophones; the Indian tradition seems more focused on vocals and the percussion, while still important, is very different. Finally, most of the themes and stories come from the Indian oral epics (Mahabharata and Ramayana), but these are also reshaped in whichever culture they take hold.