Generation Next, Part 1: How Games Can Benefit From Procedurally Generated Lore

Mark Johnson is the developer of Ultima Ratio Regum, an ANSI 4X roguelike in which the use of procedural generation extends beyond the creation of landscapes and dungeons to also dynamically create cultures, practices, social norms, rituals, beliefs, concepts, and myths. This is the first in a four part series examining what generating this kind of social detail can bring to games.

In Critical Gaming: Interactive History and Virtual Heritage, a 2015 scholarly work by Erik Champion, a very intriguing point is made about the books in the Elder Scrolls series. In considering the interactive options given to the player, and the detail of the “lore” books the player can encounter, Champion argues that the books in the Elder Scrolls series describe a far richer world than the player is actually able to engage with. He proposes that the fiction in these works speaks to a social world with a grander possibility space, and a set of more detailed cultural, social and religious elements, than the fiction the player is able to create through their actions. Elder Scrolls books speak of remarkable unique stories with a tremendous scope of actors, events, cultures, and places; within the play of the games themselves, however, it is difficult to break out of several core gameplay loops of combat, conversation and exploration, to experience or even create the kinds of stories and social worlds we read about in the game’s libraries.

As well as feeding into this disconnect from gameplay, all this background world detail was hand-written, once, by the game’s developers. As deep and rich as many of the tales these books tell are, they remain the same no matter how many times the player starts the game. Few players will ever return to a previously-read book a second time, making their reading a one-off activity done at the start of the game, and even then something that most players gloss over, fully aware of the lack of immediate relevance to their character’s experience that most books (except for “skill books” that raise your stats) actually have.

The intersection of these two concerns – their disconnect from the gameplay, and their unchanging nature – severely limits the social detail of the Elder Scrolls world to being only incidental, a backdrop, a player-created side-quest, or something to be considered once and then entirely forgotten. But what if such sociocultural detail could always be new and fresh, and always be actually reflected in the world(s) the player explores? Would players want to engage with this kind of detailed and always-distinctive worldbuilding, and how could worlds with those sorts of elements affect a player’s actual experiences instead of being nothing more than background reading?

This four-part series will examine what procedural content generation (PCG) – a set of techniques for creating (game) content algorithmically rather than by hand – might be able to do for the creation of rich in-game social and cultural worlds, and how such procedurally-generated virtual worlds can offer far more gameplay relevance in their worldbuilding than their handmade cousins. PCG to date has tended to focus on generating spaces and areas of various sorts – we can witness this at many scales, ranging from the galaxies of No Man’s Sky, Stellaris and Aurora 4X, to the world maps of Civilization, Dwarf Fortress and Minecraft, and from the expansive dungeons of NetHack, The Binding of Isaac and Cogmind, to the individual buildings or streets of XCOM and Invisible, Inc. PCG outside those limits has explored the generation of narratives, puzzles, game mechanics, and in some cases entire games, but has tended to gloss over the generation of social or cultural detail of the sort related in your average Elder Scrolls book that tells of religious practices, historical events, everyday life, artistic or musical activity, scholarly enquiry, or theatrical performances.

When “sociology” gets a look-in in game design, it tends to follow a model of social-as-statistical – selecting a particular “policy” or “civic” gives an abstract boost to your civilization of some sort, as shown in most modern grand-strategy games. However, we can algorithmically create emergent and detailed in-game societies and cultures through what I’ve taken to calling qualitative procedural generation, and in the process achieve two goals: a far greater originality in sociocultural in-game elements and a far more convincing act of virtual worldbuilding. This opens up the potential for these elements to stop being backdrop and start to affect the world the player explores directly, and the possibilities and affordances of those worlds.

To begin, however, it’s important to have a quick overview of where games currently stand in this kind of sociological worldbuilding detail and its effects (if any) upon gameplay, which to my eyes comes in three flavours: histories and lore relating to the important events of a fictional world; in-game cultures; and in-game religions, beliefs, practices or political positions, which I’ll group under the heading of “ideologies” for this discussion.

Firstly, in-game histories. Role-playing games in particular are increasingly turning to rich historical detail in their worldbuilding endeavours, most often talking about ancient societies, cultures, wars, religions, and noteworthy people and ideas. In Elder Scrolls, as mentioned above, there is a history stretching across thousands of years underlying all the games, which is mostly acquired through reading the in-game books, and in some cases talking to particular individuals or performing certain quests. A different model is pursued in the Souls games where the player is given hundreds of fleeting glimpses of the game’s world through NPC comments, item descriptions and the environment through which they pass.

In these games the physical world itself is intricately tied to the world’s societies and history, and this kind of lore yields tens of thousands of forum threads dedicated to analysis of the fiction. In the Dragon Age and Mass Effect series players collect information for a “Codex” of data about their game-worlds through conversations with characters, finding items, or undertaking events that trigger the unlock of new knowledge; in many cases this information is quite closely connected to gameplay and can offer hints about combat or conversation strategies. Many other games use the “dropped documents” style of worldbuilding, whereby the player finds pieces of paper, or books, tapes, data files, or some other equivalent unit of data storage, as they progress through the game. These elements each give fragments of information about the game’s narrative, world, or both, and the player then builds up an idea of the game’s background fiction as they play.

Secondly, in-game cultures. Well-known examples can be found in grand strategy games of the Civilization and Europa Universalis ilk, where different cultures or nations serve both as markers of different players, methods for introducing and encouraging different styles of play (if different cultures have different abilities), and varying traits such as the inclination of an AI player towards war, peace, diplomacy, science, culture, and so forth. On games of this sort cultural traits are thereby condensed into a series of high-level global preferences for each faction.

Digital fictional cultures can also be found in MMO games. In EVE Online, for example, players choose from one of four factions – the fast and lightly-armoured Minmatar, the slow and heavily-armoured Amarr, the long-range sniping Caldari or the close-range brawling Gallente – who each have a large amount of cultural detail behind them to be found on the game’s website, Wiki, and forums. You’ll see that I’ve described the four cultures in gameplay terms; EVE’s worldbuilding blends the cultural practices of each faction (the Minmatar are quasi-nomadic, the Amarr absolutist and theocratic, etc) with their preferred gameplay style and the naming of their vessels, whilst also treating the four factions as the foundation of many of the game’s conflicts, the distribution of space and system territory, and what in-game communities players are likely to become a part of. A similar idea is present in World of Warcraft and many other MMOs where we see particular units, characters, abilities or items tethered to particular cultures, and players are encouraged to align with a specific culture.

Thirdly, religions and ideologies. Religions seem more common in games than detailed political ideologies of fictional factions, nations, or organizations, although both are present. For example, the Yevon religion in Final Fantasy X is the foundation behind much of the game’s story and is reflected in great detail in the actions of its characters and the places and monsters encountered during a playthrough. The Covenant religion in the Halo series, which believes the ancient “Forerunner” species had achieved godhead by transcending the material world through the use of the “Halo” devices (in fact deadly galaxy-sterilizing superweapons), serves as much of the narrative motivation for the series. In the God of War series classical Greek religion serves as the thematic and mythological foundation of the places, people and monsters the player encounters, but is presented as a reality rather than a set of theological beliefs. In Dead Space the player battles against the “Unitologist” religion, convinced that the human race should “converge” into the game’s zombie-like necromorphs, and their symbolism and iconography can be found throughout the series whilst their NPC agents advance the narratives and pose challenges to the player.

While religion is the dominant ideological vessel, there are also games of politics. Prominent examples include the many real-world nations in the Democracy series and other comparable games in the “government simulator” genre, and the ideologically-driven factions of Alpha Centauri (rigid militarist, deep ecological, authoritarian collectivist, anarcho-capitalist, techno-utopian, fanatical religiosity, and liberal democratic). In all of these cases, beliefs and ideologies serve as primary motivators for the games’ (sometimes fixed, sometimes emergent) narratives.

This is naturally a very terse overview of these elements in game worlds, but we can nevertheless see their increasing ubiquity and centrality to contemporary games. Some games have this worldbuilding layer in order to add detail and in turn, believability – making a world seem more alive, more worthy of the player’s time and engagement, and more accurately resembling some of the systems (political, religious, cultural) we see in the real world.

Worldbuilding also offers substantial roleplay potential for players who are less interested in the technical game mechanic elements of a game, but who might be more interested in exploring that fictional world’s social, cultural and narrative aspects. We can also see two basic models for player engagement with this sociological detail, according to what direct mechanical impact that detail has on the game world. In grand strategy games, a particular policy directly affects the technical specifics of play – such as “+X Gold Per Town” – rather than any deeper engagement with the kind of social structuration that terms like “Mercantilism” or “Serfdom” truly represent. On the other hand, many worldbuilding elements have no direct impact on play, but their discovery can become an additional play element for dedicated fans intrigued by the fictional universe. The hunt for lore can be an end in and of itself.

Despite online discussions around lore, fan-fiction, cosplayers who identify with characters, and even the popularity (or not) of spin-off books and films, it’s very difficult to know whether the majority of players are actually interested in these social, thematic and cultural elements. If not the majority, does a sufficiently non-trivial volume of the player base care enough to make these worth including? On the one hand, there is a tremendous drive in many quarters for the creation of games with deep and affecting stories (seen by many as a route towards the greater cultural legitimation of the media form).

On the other hand, heavily story-driven and relatively gameplay-light games risk being disparaged as “walking simulators” or as “mere” interactive fiction, and concerns of gameplay still dominate the reviewing processes of most critics, design decisions of most professional developers, and purchase decisions of most players. Despite these latter points, I think a lot of players would criticize an RPG, for example, for lacking this kind of worldbuilding detail, even if they never directly engage with it – there is an expectation of detailed fiction that players want to know is present, even if they ignore it. Many players of the Souls series have little interest in the cultural and thematic elements of those game worlds compared to the tight and precise gameplay, but still acknowledge that their very lack of knowledge of the game’s lore, coupled with an understanding that such lore does exist, is important to the overall “feel” of the game as mysterious and cryptic and to the creation of a convincing virtual world.

As outlined as the start of this piece, however, this kind of social, cultural, religious or ideological worldbuilding tends to suffer from two problems – it is often (though certainly not always) disconnected from gameplay or otherwise abstracted out, and once read the first time, players almost never return to it on a subsequent playthrough. To aid our thinking about how worldbuilding might be more closely integrated with gameplay, and kept fresh and original, in the next three parts of this series I’ll examine in detail the possibility of procedurally-generated social, cultural and religious worldbuilding and what it can bring to both background detail and immediate gameplay. To do this I’ll consider my own work and the work of others in these areas; explore where the cutting-edge currently lies, and how that cutting-edge might move forward in the coming years; and summarize what this can bring to worldbuilding as both intriguing detail, and gameplay-relevant information.

From this site

54 Comments

  1. hernique says:

    expecting fiercily

  2. Dj Assassin says:

    Procedurally generating lore sounds tough. I wonder if it becomes commonplace it might result in marginalisation of real world cultural history. In the ‘This expansive lore has been made far more interesting than real-life history’ case, you could point to games and books already doing that, but the more authentic *amount* that could be created by proc-gen would make playing those sorts of games more demanding.

    A little more worrying to me is the ‘I guess this is the way cultures advance’ sense, similar to how the civ series has a relatively rigid progression, the amount of variation in early examples of lore proc-gen might rely on recreating certain societal structures which could mean a lack of diversity in future iterations. I suppose a lot depends on the complexity of the generation and how much wackiness it allows for. Very much looking forward to the rest of this series!

    • LexW1 says:

      I think the first concern would be more of a problem with hand-crafted lore, which is generally far more engaging than PCG lore (and even when PCG lore is better than it is now, adding hand-crafting on top will inevitably improve it further). So unless you feel like that’s already happened (which you could argue it has), I don’t think it’s worth worrying about.

      The second concern is totally legit, though. The very idea that cultures “advance” is questionable, even just looking at the history of Europe. Was Britain circa 1100 AD “more advanced” than Rome circa 200 BC? Or even Britain circa 300 AD, say. So the idea that cultures follow some linear pattern of “advancement” is clearly false. Cultures often lose stuff and gain stuff – not until you get to around the 1600s do you see any consistent “advancement”, and even that is tightly associated with avoiding tyranny and barbarism, both of which love to burn down any advancements (or just shoot everyone who went to university or whatever). One could make a case that a number of countries in the Middle and Near East are less “advanced” culturally/intellectually (not merely in terms of being liberal or whatever – but even being complex or sophisticated in any way) than they were a few decades ago.

      • Unclepauly says:

        Africa has been pretty linear.

        • Coming Second says:

          Egypt might have one or two things to say about that.

        • Lamb Chop says:

          Also the whole part where many were colonized and often enslaved to extract their resource wealth. Progress!

      • Mark Johnson says:

        Re: the advancing issue, see my other comment on this bit of the comment thread; re: the PCG lore being engaging issue, I completely agree that the “huge database of small handmade elements stitched together” is actually really important to this kind of PCG, even if it might feel a little inelegant just writing down fifty different ways of saying “The gods I worship are called [X]”, or whatever. In turn, that classic thing of balancing the generator to produce outcomes 10/10/10/10/1/1/1/1 in likelihood, instead of 1/1/1/1/1/1, gives the unusual ones far more meaning when they do crop up!

    • Mark Johnson says:

      Thanks! Yeah, I totally agree about the dangers of linear progression, that kind of technological determinist narrative that games like Civ (re)produce; I’ve worked hard to avoid it in my own work, and it’ll get mentioned quite a few times in the future parts of this series. For me, something like Mesoamerican cultures and European cultures are great examples; each was massively technologically “ahead” of the other in different ways, and each lacked “technologies” that the other would have found life impossible without. I try to build off that idea and ask a huge number of “what if” questions pretty much constantly!

  3. savagegreywolf says:

    Yes, I recently played a game with procedurally generated ‘lore’ and enjoyed it quite a bit. It was called Not A Hero.

    The Madlibs approach to worldbuilding makes about as much sense as you’d expect.

  4. aozgolo says:

    I would certainly appreciate such endeavors greatly in games, especially if done right. Though in reading this I can already see the implementation of this having different meanings for different people. For many developers, the closest to this route I see being considered is a multiplayer hub of interactivity where someone’s first person grand narrative becomes another person’s third person lore. Yet for the purposes of this article having the game itself generate this richness of culture, society, and lore is a lofty goal in itself, and one I see suffering from many failed attempts before truly achieving success.

    As evidenced by a trend I’ve seen in other media recently where youtube videos and short films are being written by AI fed specific parameters, the results often end up as comedic abstract tales bordering on nonsense, which for a more serious game could spell disaster. Imagine if the procedural generation creates a rich cultural background for an entire faction whose key element centers around the worship of dogs in tutus and wearing clown make-up? A hard sell to someone hoping for Elder Scrolls or Dark Souls level of immersion.

    To create a PCG lore that isn’t immediately scrutinized for it’s lack of a human artist and nonsensical mix-mashing of disconnected tropes to build the semblance of worthwhile interactive lore is a true challenge. The many failures in this attempt could spell the death for this concept before it truly ever hits it’s stride.

    • LexW1 says:

      The “dogs in tutus” concern is simply addressed by limiting how far the PCG lore can go. Dwarf Fortress already does this, and whilst it makes some silly stuff up, it’s not at the level of ludicrousness, not remotely – and usually seems “in world”-consistent, even if a bit silly to us (though there are still issues – like small children winning historical sporting competitions – but again that’s about setting limits).

      I’d actually suggest that DF’s current lore generation produces lore very much in line with Dark Souls and to a lesser extent TES.

      I think the big thing you lose with PCG is any kind of thought or cleverness or allusion. You can’t be “inspired” by some particularly interesting moment from history or whatever if it’s all PCG. You’d never get something as interesting as Qunari from Dragon Age for example – well, you might, but only in an infinite monkeys kind of way.

      Again you already see this with DF’s PCG lore – very often you’ll have, for example, one very interesting and cool god in a pantheon of bores. Whereas a human artist with some talent could easily have created multiple interesting ones.

      • LexW1 says:

        Grrrr no edit so replying to myself, but basically PCG is a lot like giving the infinite monkeys a written set of guidelines, which they will follow absolutely as mindlessly as they were already bashing the keyboards, so those guidelines need to be carefully written.

        • Unclepauly says:

          I feel like if you need to basically guide the PCG so much you might as well just write an interesting story. Unless having randomness is SO important to reply value in a story game(I don’t feel it is but what do I know).

      • Mark Johnson says:

        Yes, totally re: “in-world” consistent for DF; all the stories make sense based on how we, as DF viewers/players/whatever, understand the way that the DF world functions and how people in the DF world “think”. I think it does produce more TES-esque lore, personally, than DS, but maybe that’s just because to me the philosophical/ontological/ethical underpinnings of the DS lore/background stories are so crucial (at least to my reading!). Allusion: that is an interesting point. In my own work (see future parts) I am trying to add in a few concepts like irony, insults, humour, allusion, etc, but “only” through large databases that can slot themselves, if possible, into the generated world.

    • Mark Johnson says:

      I completely agree that different people will “get” different things from the idea; the two points I outlined here (keeping lore interesting, and making lore gameplay-meaningful) are definitely only two possibilities of many. As you say, one does end up with a lot of weird/comedic/nonsense tales; I love reading them for their own value (if they’re complex enough), but obviously we need to move far beyond that for any kind of good PCG lore/social/cultural/etc detail.

  5. HuvaaKoodia says:

    “On the other hand, heavily story-driven and relatively gameplay-light games risk being disparaged as “walking simulators” or as “mere” interactive fiction, and concerns of gameplay still dominate the reviewing processes of most critics, design decisions of most professional developers, and purchase decisions of most players. ”

    And this is why we need Interactive Fiction to be its own thing with its own marketplace (or market category within current digital distribution channels).

    IF isn’t a genre/subset of games and thinking that way is severely limiting. Games are about winning, losing and decision making. Interactive worlds and stories can be about pretty much anything just like traditional, non-interactive fiction in general.

    • Mark Johnson says:

      Completely agreed; I think we will see this, but right now there seems to be such a mess of rhetoric and boundary-defining from all sides on the IF/games debate/issue! I also (and I’m sure many others think this) feel we’re only scratched the absolute surface of what IF can be.

      • HuvaaKoodia says:

        The good news first: The debate doesn’t matter. If there’s a market the developers will come. Critics and the public will follow automatically, or hazard being left in the dust (which they admittedly have the right to do).

        The bad news then: The market won’t appear on its own. There has to be a push, be it a movement within developers or just a few great titles that break the ice. Digital distributors will no doubt jump on the bandwagon when they assess the risk low enough.

        And yes, IF (and interactive digital media in general) will have a rosy future. I’m placing my bets on VR bringing about the first signs of the change (digital experiences, interactive movies, interactive simulations, you name it).

    • April March says:

      Go watch how 90% of people play GTA and then tell me games are about winning, losing and decision-making.

      “Walking simulators” did nothing more than to focus on a single aspect of games, namely the exploration. To argue that they should be separate (more than they already are by being its own genre) is like arguing that because I only like games that can be beaten without fighting, fighting games shouldn’t be included in the same category as ‘regular’ games.

      • HuvaaKoodia says:

        You bring up two very good examples of the kind of muddy waters that interactive digital media is indeed mired in at the moment. Everything is called a game because there’s a big market for games, not because everything is a game. Let’s take a look at said examples.

        GTA is of the kitchen sink variety of interactive media or interactive multimedia as I like to call it. You have a city simulation, interactive story bits, minigames/puzzles and of course the main dish: causing mayhem.

        I don’t mind calling GTA a game, but surely enough it is more of a toy a lot of times. You are exploring the simulation and seeing how it reacts to all sorts of input, just for fun and giggles. The mayhem part is most certainly a game though. You try to reach a high wanted level (winning), by using any means possible (decision making) in order to avoid the cops (losing).

        Walking simulators as their name suggests are “non-games” (terrible term, wouldn’t you agreed?), somewhere between interactive fiction and simulation, depending on how much focus the developer has put on the narrative in contrast to rules and interconnected systems (simulations, that is).

        If there’s no winning, losing or decision making why call it a game then? Every non-digital game has those three things so why do digital games get away with having none?

        The reason, like stated before in the article, is the market; plain and simple. We change the market and soon developers are more free to create (and, most importantly, sell) all sorts of interactive media. Wouldn’t you prefer that?

        Concerning the last point. This is not about exclusivity, but rather inclusivity. By using more specific names we are not trying to shove certain titles away, but rather giving them the stage that they deserve. An over saturated market, geared towards rule based interactive media (games, toys, puzzles) isn’t that stage.

  6. froz says:

    To be honest, it all feels to me like advocating procedural generation of novels (or even some background “lore” within fantasy novels). Surely everyone will agree that would be a stupid idea?

    I don’t get the appeal of quantity over quality. Especially in a world where we have so huge quantity of everything, including the best works of art.

    • igorhorst says:

      I wouldn’t say the procedural generation of novel is a stupid goal. NaNoGenMo (National Novel Generation Month) does exists (disclaimer: I did participate in it last year), and people have tried to generate novels before (The Policeman’s Beard is Half-Constructed, Just This Once). The procedural generation of a novel can be just as hard as the manual creation of a novel…even more so since you have to code the very constraints necessary for a novel to be written and you’re dealing with an entity that lacks common sense. Some of the novels that are generated can be seen as “conceptual art”, while others are legit attempts at novel-generation (primarily for the challenge involved).

      I do agree with the whole “quantity over quality” argument though. One of the biggest problems that PCG wants to solve is producing quantity, except we are already experiencing “Content Shock”, with a flood of very good human-generated content already in existence. But this “quantity over quality” can already apply to human-generated content…why should you bother handwriting your own content when other humans already did it for you? Could an algorithm find good examples of human-generated content on the web (possibly from the Gutenberg Project), and then arrange them in a pleasing manner for a human to consume?

      But I don’t think the goal of PCG would be to produce quantity. Other goals could exist…such as maybe providing uniqueness to each person’s gameplay experiences (as the popularity of roguelikes shows), displaying interesting concepts that a human would not necessarily think about, or even just producing the illusion of a world greater than itself (the role that current hand-generated “lore” plays). This blog post (link to procedural-generation.tumblr.com) actually tries to argue against treating “infinite content” as the major goal of procedural generation.

      • igorhorst says:

        (Disclaimer: I did not write that blog post, it was written by Isaac Karth. He also was a regular participant in NaNoGenMo as well.)

    • HeavyStorm says:

      Hey, not to sound pendant, but I don’t think you get the point. (or I didn’t)

      There’s a place for this. Generating lore procedurally means more than generating the static books that dully fills the world of TES. What I think the ultimate goal is, is a game capable to generate lore reacting to player interactions, and build the world around the player.

      Well, at least that’s what I think valuable. Filling those books with words are meaningless to me. (I’m one of those players that gloss over the books; only I’m not and I actually read them)

      • Unclepauly says:

        “pendant” Surely you meant pedantic? Also I’m pretty sure we have worlds that react to the player already. I don’t think that’s what PCG is striving for.

      • Mark Johnson says:

        > is a game capable to generate lore reacting to player interactions, and build the world around the player.

        Personally, my interests lie more in having a world where the lore and the world/gameplay are closely integrated, but in many ways I think these are two sides of the same “X affects Y, and Y affects X” coin, really!

    • Scio says:

      To be terse (but sincere,) it’s only a question of ‘quantity over quality’ if it’s plausible—or even possible—to hire a talented and hard-working author for every player’s every playthrough; to come up not only with a new world every time but to also keep advancing it as the player interacts with the world, including their own effects.

      In lieu of this it is more of a question of doing something that cannot be done—at all. It is exploring a very new and personally very exciting form or writing.

      And it goes beyond just lore, even if it’s the topic of this series. “Proper” emergent worlds (as opposed to half-baked or buzz-wordy emergent worlds :) excite me to no end.

      • froz says:

        Personally I would always choose to have great experienced which is similar (or even identical) to that of other players, than to have mediocre experience that is unique. To be honest, I completely don’t mind if what I see in game is the same what other players see. Perhaps that’s even better, as later I can discuss it. The only reason for me to have each playthrough different would be as a goal to let me reply the game several times, with different outcomes. But then again we are back to my original statement.

      • Mark Johnson says:

        I agree! I think it is very exciting to see what can be created in this way, AND how it can affect gameplay as well as being cool/interesting background stuff.

  7. mercyRPG says:

    I’ll just use an Autoplay bot on the game then so I can go pokémon hunting, if: “also dynamically create cultures, practices, social norms, rituals, beliefs, concepts, and myths.”

  8. MajorLag says:

    The problem with procedural generation is that there’s always this sort of samey-ness about it you know? Yeah, technically this procedurally generated landscape is entirely unique in game X, but I can find an infinite number of just-as-unique landscapes none of which are really anymore *interesting* than the others. They’re all just randomized mish-mashes of the same familiar pieces and the whole of their combinations doesn’t add up to more than the sum of their parts.

    I think the core of the issue is that the player, with a little experience, can easily see how this piece of generated content came to be. They can see all the parts and how they fit together in an instant, so it isn’t interesting. I’m not sure there’s a way around that. You need to introduce some kind of creativity. Something that grabs the player’s attention and makes them say “I’ve never seen this before, and I have no idea how it can even exist!”. Make them question it, poke it with a stick, and want to learn its secrets. Think something like the alien artifacts in Elite Dangerous, but but deeper. If every time a discovery was made about them it lead to more clues, but also more questions, more “how is that even possible?”.

    In lore, you’d want to see completely unexpected historical events, that none-the-less could be researched and lead the player down a rabbit hole of information. There was a kingdom, say, that the lore says was ruled by an immortal. That’s odd, because there is no immortality in this game, dig deeper. Ok, the king reigned for 600 years and only spoke to the people through an intermediary. Ah, so probably a construct then, not real, just a convenient fiction for those with the real power. But wait, the king demanded large quantities of lye every week. What would they need that for? Dig deeper…

    And so on. Ideally ultimately leading to a revelation about the world the player inhabits in the present. This is probably impossible with any conventional algorithm, of course, but if we can teach an AI to play Go at Grand Master levels, then maybe we can eventually create one that can extrapolate creative histories from a given baseline world. When that sort of thing is possible, it’ll be interesting times.

    • Unclepauly says:

      Fully agree with the missing creativity stuff. How can PCG ever hope to create a story like Planescape Torment or any of the other great and unique stories? Without so much guidance as to where you didn’t need the PCG in the 1st place it can’t happen.

      • igorhorst says:

        Then again, is *every game* like Planescape: Torment? For every game with excellent human-generated lore, there are hundreds of other games with excuse plots (also human-generated). It is possible that PCGed stories are not going to produce the “best stories ever” (TM)…but it could produce ‘good enough stories’ that would be on-par with most “game stories”.

    • Mark Johnson says:

      I replied, but must have misclicked somewhere – it’s at the bottom of the thread!

  9. HeavyStorm says:

    Very interesting piece. Congrats, I’m really anxious to see the next the parts.

    I wish you do some case studying, specially of games that have already took a shot of generating lore, (even though probably not by using algorithms, since I don’t know any that do).

    One game that I think fits this idea nicely is King of Dragon Pass and RimWorld (or DF, since they are analogous).

  10. Reverant says:

    I’ll of course reserve judgment until the rest of the series is done, but my knee jerk reaction is this interest in procedurally generated lore sounds a bit like a saying, “I’m not keen on the nature and function of literature, and I’d rather we didn’t have to keep one of those nasty writers around. Can’t we just automate it all?”

    • Nasarius says:

      Or: a human writer does exactly the same amount of work they usually do, but procedural generation fills in all the unwritten details that the player chooses to explore.

      Procedural generation can be so much more than generating random garbage from scratch.

      • Reverant says:

        Nasarius, my second comment below was posted before I saw yours. I still wonder—is that procedural mad libs still worth reading? It’s all so soulless.

        • Nasarius says:

          Again, you’re thinking of the random garbage generation you’ve already seen. We’re only just beginning to explore the possibilities here.

          • daveyd says:

            Correct me if I’m wrong, but it sounds like you’re suggesting that game designers essentially write a scenario that could one of several ways… So say I happen to get version A when I play, whereas you could get version B or C…

            As a gamer who almost exclusively plays RPGs, this sounds much, much less compelling to me than a game where what my character does or says ultimately determines which version of the events I see. In other words, “choice & consequences”… What benefits does randomness bring to the table?

            “Age of Decadence” has amazing replayability. Eventually a player who makes every possible choice might see everything, but they would also eventually see everything in a game with kind of PG lore / quests, you’re talking about.

    • Reverant says:

      Really, why labor to generate an infinite amount of fluff only worth reading once, when you can strive for your writers to create one story reading over and over again?

    • Mark Johnson says:

      Heh – well, I appreciate you reserving judgement, and I do understand the critique, of course. That’s not in ANY way my interest in the area – I mean, I’d say then the PCG designer is the author, but that’s a side point. It’s more about… the process of writing-gameplay integration, and the writing-and-player-experience integrations, that I think PCG could help us with.

  11. kaisergav says:

    The idea of variable in-world lore is interesting.
    However, if I recall correctly, some of the lore in Morrowind’s books (though perhaps not in later Elder Scrolls games) actually *does* contain real information about things that exist inside the game world. For example, some of the rare books about vampires describe the very real and secret vampire clans that actually exist and can be sought out in the game. It’s hidden amongst meaningless fluff, but I still found it exciting when I first realised that breaking into the libraries of Vvardenfell was an actual way to potentially learn real secrets about the game. Makes me wish that Wikis didn’t exist, in order preserve the mysteries and get the most from the experience.

    • Unclepauly says:

      That sounds glorious. I never played Morrowind :(. I’m waiting on a proper up to date remake.

    • Oozo says:

      That’s actually one thing were I could see procedurally generated lore being interesting: Undermining wikis, restoring a sense of discovery and mystery that I remember games having had before the internet came around. If wikis have to tackle lore the way game guides have to tackle strategy games — that is, talking about general concepts, but not minute details –, it could be exciting.

      Also, I’m just reading The Name of The Rose, and I couldn’t help thinking that one of the reasons why those lore books often seem so cheap is simply because they have absolutely no value in the game world they can be found in. It would be probably way more interesting to have books relegated to specific places, libraries, monasteries and so on, so that finding them would be exciting again. (Especially if you connected them to proper quests.) If you can find them virtually in every nook and cranny, no wonder they seem to be disposable garbage.

      • Mark Johnson says:

        YES. This is a huge reason I do it – the game I design (which I’ll look at in the later entries) is in many ways designed to be un-wiki-less, to the greatest extent possible. You MUST learn about that in-game world to progress. Name of the Rose is a huge inspiration on my work, actually, and it has heavily influenced how I’ve designed my work and how/where the player acquires information!

    • Mark Johnson says:

      I believe you’re right re: Morrowind – and it’s basically a PCG version of that I’m after. If you read a book that Queen A was buried in a tomb with Design B on it in Crypt C under city D, and then you go there, and that is indeed there – now THAT is what I’m after.

      • daveyd says:

        I’m just not seeing what the benefit of these variations will yield. So the name of the Queen changes. The location and layout of her tomb changes. It’s still fundamentally the same exact quest: Go to a dungeon and retrieve this trinket.

        Is this all about invalidating wikis / strategy guides?

        I’m far more interested in complex quests with multiple resolutions. Suppose I’m after a magical item which I find out is in the possession of nobleman. I can sneak into his home to steal it. Just break in, kill his guards / threaten him and take it. Talk to him and somehow trick him into giving it to me, or find out what I have to do to convince him to part with it. Each method would have different ramifications. Plenty of replay value, and no need to change his name or where he lives..

        • FuzzyBlueBaron says:

          Ah, but surely that’s where the power of PCG truly comes into its own! Why stop at changing the nobleman’s name & location? Why not mix up who (queen, innkeeper, hermit, travelling prizefighter) or what (dragon, golem, alien/non-human race, crumbling ruins, suspiciously quiet forest clearing) has the item? Why not generate a personality for the owner of said treasure–are they easily angered, overly fond of [vice], or rumoured to have fallen out with an old friend? Research and gossip could allude to past accounts of the item, or owner, or location, or a third party that could completely alter how you approach your task.

          The only thing limiting you, really, is your imagination. And the human body’s unfortunate need for things like food, sleep, and occasional potty breaks.

  12. Mark Johnson says:

    > I think the core of the issue is that the player, with a little experience, can easily see how this piece of generated content came to be.

    Oh yeah, that’s a HUGE issue. Again, not to get ahead of the other parts I’m writing, but I’ve put quite a lot in place to try to handle this issue through a range of techniques, and I think it’s going pretty well. As an example, I often get people pointing to something in my game and asking “How was that PCG’d?”, when in fact the element was almost entirely handmade, and people saying “That’s clearly a handmade element” when in fact it’s heavily PCG. As a design goal, almost, I’ve tried to blur those as much as possible and mask generation processes for precisely this reason – seeing “how it was made” breaks the immersion/suspension of disbelief/etc.

    As for completely unexpected historical events – oh yes, that would be amazing if it was *truly* PCGd, but that would require an incredible detail and granularity impossible for even a dedicated one-person developer, and honestly, for most teams too. The closest I’ve come to this is writing out a list of “cool unique things”, having the game pick only a small number each time, and then vary them massively. It’s not perfect, but it seems to work.

  13. Traipse says:

    I’m quite torn on this. On the one hand, procedural generation is a great tool for freeing up creators’ creative energy for the game itself, and it’s possible that it might be capable of making gameplay deeper, as you suggest. On the other hand, the results have — to date, at least — been invariably rather less interesting than hand-crafted content. The ultimate example, for me, would be the 36 Sermons of Vivec (link to newwhirlingschool.com) from Morrowind, a sprawling collection of mad, rambling, brilliant texts about one of the game’s central characters that are an absolute gold mine of story and style. “Story” because it gives you backstory which makes the game’s main plot much richer; “style” because it’s an inimitable work which establishes the theme and feel of the setting. Twenty years later, people are still debating its philosophical ramifications on forum threads. I just don’t think we can get that degree of depth out of procedurally-generated lore.

    Furthermore, as a player, when I know that something is procedurally generated, it seems less interesting to me. The first couple of times you see a cool-looking mountain in Minecraft, say, you think “Wow! That looks neat!” Thereafter, you look at it and think “I appreciate the unlikelihood of that particular feature’s generation”, but the magic is gone. I expect the same would happen for me with generated lore; I would start spotting the patterns, picking out the game-relevant bits, and ignoring the rest. It’s just not as much fun, somehow, to interact with something that I know is procedurally generated rather than something I know someone deliberately laboured over and spent their time and intention on.

    • TheAngriestHobo says:

      The first couple of times you see a cool-looking mountain in Minecraft, say, you think “Wow! That looks neat!” Thereafter, you look at it and think “I appreciate the unlikelihood of that particular feature’s generation”

      Did the first mountain have a dictionary in it or something?

  14. geldonyetich says:

    Speaking of a dabbler of the craft, I’d say the main appeal of procedural generation is just knowing I could experience content I never conceived upon. It’s all the joy of teaching the computer to be your own dungeon master.

    The trouble is getting it all to fit together. Histories, cultures, lore of various kinds… if the gameplay doesn’t support or require it, it’s rather futile. Or, as you put it:
    This opens up the potential for these elements to stop being backdrop and start to affect the world the player explores directly, and the possibilities and affordances of those worlds.
    The main thing stopping it from happening is that it’s a lot of work to teach these overgrown calculators to connect the lore to the gameplay with any kind of fidelity.

    I’m working on it. Dammit, every time I get to work, life reminds me that it’s the ultimate procedural generation engine. I swear, it’s like God himself is worried he’d be upstaged. The mortal dareth? Cue lightning bolt.

    Anyway, I’d say our closest mark thus far is probably Mount & Blade. It doesn’t have procedural generated books, sure, but all the sociological aspects have ties to the gameplay.