Madoc
Project Lead
The next big update is due out in days, but before I just give you a bunch of release notes, I want to address the elephant in the room. We all know the update has taken an extraordinarily long time to get here, and I want to touch on that. What started out as just the next update became a turning point in development, and from there the work just piled on.
There is a lot that comes with this update. We've overhauled huge parts of the core game, added important new features, a lot of new and revisited content, improved on almost every aspect of the game, we've also expanded and improved our development tools, and paved the road for big features to come. Most of this will be in the release notes, but I wanted to touch on a big topic that I think needs some context, and represents the biggest effort development wise: our AI.
To understand why AI is important you need to understand our core design philosophy (many of you do, I know!). You might think the game's physics were designed for combat. They were not. They were designed to support an interactive and dynamic world, where things aren't just symbolic representations with specified outcomes, but actually functional. However important, combat is just one aspect of something broader.
This applies to everything we do. We don't create a world from preset interactions and scenarios, we create things with functional effects, which can produce certain outcomes, but also many more that we never even thought of. Things just work, this gives the player creative freedom, it makes the world more "tangible" and in a sense more real. We want the choice to do anything we can think of, and witness plausible outcomes. Whether this is desirable or important is not relevant, it is important to us, it furthers our immersion.
AI is such a vast topic that I've never figured out how to talk about it. Perhaps I will have to do a series of more in-depth discussions of specific aspects of it. We've been figuring this out for years, we built some core features of the game to support it, we've had simple approximations of some behaviour models in the game for some time. But there was still a lot to figure out, and a massive amount of work to make it functional.
We've now built a complete core behaviour model. This covers personalities, social behaviour and influences, interpreting the actions of others, relationships, memory, emotions and other states of mind, and many subtle but notable things besides. It is a functional core for some life-like emergent behaviour, and it also supports separate but important things like manipulation with mind thaumaturgy.
But this alone is not nearly enough. There is behaviour, and then there is intelligence, understanding endless possible situations and knowing what to do. The more scenarios you imagine, and the more player and mechanical freedom you allow, the more this seems insurmountable. We've designed an incremental approach to this problem, which we call the role system. It is in a sense a library of behaviours befitting various situations. Big topic, I will discuss it more later, but it is something we can build gradually, expanding on what our AI is capable of doing.
Realistically most of this won't come into play until the more complex setting of Sui Generis, but it is important that we are now using these systems to drive all AI, and even in Exanima's context, there is much we can do with it.
We have some ambitious goals, and we've built high expectations in the community over the years. Of course we had to start somewhere, and we inevitably entered a cycle of feedback and refinement which distracted us from other goals. We had to break this cycle, we had to stop building on top of incomplete systems to move forward.
It's easy enough to imagine simple solutions by simplifying the problem. But we're not interested in that. The game is full of open ended tasks, many things you could spend years on, but I am just one man programming everything, developing the game and our content tools, and doing much more besides. There is even more I ought to do, such as communicate better with you, but marketing material and hype is not my thing, and I also hate spoilers. I do think communication is important, perhaps even more so with this unique game. I do want to try to do better, I mean it, but things rarely go as planned, and I am often overwhelmed already.
Anyway, the update really is just about ready for release. And we have a tonne of stuff lined up for smaller updates in the near future, things that are very nearly complete, or that we will be fleshing out significantly.
There is a lot that comes with this update. We've overhauled huge parts of the core game, added important new features, a lot of new and revisited content, improved on almost every aspect of the game, we've also expanded and improved our development tools, and paved the road for big features to come. Most of this will be in the release notes, but I wanted to touch on a big topic that I think needs some context, and represents the biggest effort development wise: our AI.
To understand why AI is important you need to understand our core design philosophy (many of you do, I know!). You might think the game's physics were designed for combat. They were not. They were designed to support an interactive and dynamic world, where things aren't just symbolic representations with specified outcomes, but actually functional. However important, combat is just one aspect of something broader.
This applies to everything we do. We don't create a world from preset interactions and scenarios, we create things with functional effects, which can produce certain outcomes, but also many more that we never even thought of. Things just work, this gives the player creative freedom, it makes the world more "tangible" and in a sense more real. We want the choice to do anything we can think of, and witness plausible outcomes. Whether this is desirable or important is not relevant, it is important to us, it furthers our immersion.
AI is such a vast topic that I've never figured out how to talk about it. Perhaps I will have to do a series of more in-depth discussions of specific aspects of it. We've been figuring this out for years, we built some core features of the game to support it, we've had simple approximations of some behaviour models in the game for some time. But there was still a lot to figure out, and a massive amount of work to make it functional.
We've now built a complete core behaviour model. This covers personalities, social behaviour and influences, interpreting the actions of others, relationships, memory, emotions and other states of mind, and many subtle but notable things besides. It is a functional core for some life-like emergent behaviour, and it also supports separate but important things like manipulation with mind thaumaturgy.
But this alone is not nearly enough. There is behaviour, and then there is intelligence, understanding endless possible situations and knowing what to do. The more scenarios you imagine, and the more player and mechanical freedom you allow, the more this seems insurmountable. We've designed an incremental approach to this problem, which we call the role system. It is in a sense a library of behaviours befitting various situations. Big topic, I will discuss it more later, but it is something we can build gradually, expanding on what our AI is capable of doing.
Realistically most of this won't come into play until the more complex setting of Sui Generis, but it is important that we are now using these systems to drive all AI, and even in Exanima's context, there is much we can do with it.
We have some ambitious goals, and we've built high expectations in the community over the years. Of course we had to start somewhere, and we inevitably entered a cycle of feedback and refinement which distracted us from other goals. We had to break this cycle, we had to stop building on top of incomplete systems to move forward.
It's easy enough to imagine simple solutions by simplifying the problem. But we're not interested in that. The game is full of open ended tasks, many things you could spend years on, but I am just one man programming everything, developing the game and our content tools, and doing much more besides. There is even more I ought to do, such as communicate better with you, but marketing material and hype is not my thing, and I also hate spoilers. I do think communication is important, perhaps even more so with this unique game. I do want to try to do better, I mean it, but things rarely go as planned, and I am often overwhelmed already.
Anyway, the update really is just about ready for release. And we have a tonne of stuff lined up for smaller updates in the near future, things that are very nearly complete, or that we will be fleshing out significantly.
