To Build a Better AI

sb4-1

Storybricks, the company which we have heard so much about but we still know so little. For those of you who have not yet heard of Storybricks, they are collaborating with Sony Online Entertainment (SOE) to develop the AI for the upcoming EverQuest Next title. The Emergent AI system that has been mentioned as a main feature of EverQuest Next, that’s the Storybricks team.  Thus far there have been few opportunities to actually see what they are about, largely because of the black box that SOE is working in. However, we were lucky enough to get a chance to ask the Storybricks team of Rodolfo Rosini, Stéphane Bura and Brian ‘Psychochild’ Green a few questions to learn a little more about them. Keep in mind, with their work on EverQuest Next they are limited in much of what they can say, but I think you will find that our collective gaming futures appear to be in good hands.

Recently we have heard a lot about what the future of Storybricks may bring to the gaming industry. But something we have not heard much of is the company’s beginnings. What led to the creation of Storybricks?

The company had a rather simple beginning.  We came together because we wanted to see an improvement in games.  Rodolfo Rosini, the CEO, is an avid MMO player with a tech background and was disappointed after he’d defeat raid bosses and none of the NPCs in town recognized his achievement.  Stéphane Bura brought his ability as game designer and AI expert, eager to bring new stories to a new medium.  Brian ‘Psychochild’ Green has over a decade of MMO development experience and a frustration with the stagnation of MMO development. There are many more talented people that have joined our company as well.

The initial focus was on AI, as this seemed an area ripe with possibility.  Being able to bring the spark of life to NPCs was our first goal.

I have seen mentioned that you are working on other projects in addition to your well known collaboration with SOE.  Are there any plans to branch out into other genres?  How about consoles or perhaps tablets or smart phones?

Right now EverQuest Next is our main focus.  We admire SOE for having the foresight to really look forward to what will make a truly next-generation game. We are not hiding the fact that we are working on other projects but we are not ready to talk about them yet.

You have mentioned that the Kickstarter campaign didn’t seem to take off, possibly because of marketing. What role do you think having such a new and different idea played, where people did not understand the potential?

There were many reasons why the Kickstarter for the first iteration of our technology did not take off. We learned some lessons and helped us move forward. Blaming marketing would be too simple as there were usability issues as well. Failing gave us focus and we learned what we were good at and where we needed help. And we’re very grateful to the people who did support us and shared our vision at the time, many of them from the game industry who had no problem understanding, as you say, the potential.

Kingdom of Default (Not EQNext)

maxresdefault

A recent article mentioned that the Storybricks we have seen is different than the initial vision. How has it changed?

It’s not much that our vision changed over time but more how it became better defined. We started with a very wide and fuzzy statement about the “illusion of life” that came from an old book about Disney Animation. It turned out that we saw similarities between the state of computer animation pre-Walt Disney and the games industry today. And some of the challenges that they had were very inspiring. We figured that the way to deliver the resemblance of a spark of life was not through better graphics as others have tried but with better behaviors, thus the focus on AI.

Being able to prototype our ideas allowed us to learn a lot about what worked, what needed revising and, most importantly, what the actual priorities were in our development plan. This is how game developers who have the luxury to iterate on their design reach high quality. We’re trying to apply the same techniques to our own development.

We have specific goals and priorities for EverQuest Next, and you will be able to see what this unique and fruitful collaboration yields in the future.

In what ways has your partnership with SOE helped you to develop Storybricks beyond its original concept?

Our partnership allows us to focus more on the core of the experience that the AI provides, instead of having to handle a full game development. We come to a project that is rich with game design, art, lore, world design, all created around a solid vision.

Great AI doesn’t exist in a vacuum. It gives life to a world and uses game systems and thus is shaped by them. Conversely, because we joined the project early enough in the design process, we can weigh in on the design of features that would be impossible to create without AI. This type of synergy between AI, design and lore doesn’t happen by chance: you have to pursue it. That’s why we are very happy of working with the EQN team because that’s exactly how they work.

There is clearly a lot that has gone into the development of Storybricks. Is there a simple way to explain how Storybricks works to the average gamer?

What Storybricks does is giving characters understandable but not entirely predictable behaviors.  We will have accomplished our goal when players see a character’s behavior and say, “I didn’t expect that to happen, but it makes sense.”

Today Storybricks for EQN is technology and design. It’s a lot of work and some of it might not make it in the final build. Sony will talk in detail about features that make up their Emergent AI model when they are ready. We prefer not to talk about features until those are ready for primetime.

Kingdom of Default (Not EQNext)

There has been a lot of speculation that Storybricks can influence almost everything in a game world. Is there any part of a game world, other than logging in and out, that Storybricks does not in some way interact with?

We would rather not add to the hype and build unrealistic expectations. The only thing that we will mention is that Dave Georgeson spoke about ‘tagging’ elements of the game so that NPCs react differently to it depending on the context. You will have to wait and see for the actual applications of this technique.

It seems that with Storybricks we are able to have highly variable experiences during interactions with the game world. I imagine that I could be playing with a friend on a 50 step story arc, get to the last step and have completely different experiences all because we interacted differently with an NPC back on step 11. Will that be possible?

To be fair, that is possible with traditional game even without any AI, just not in MMOs.  There have been games released in the last few years where you make a decision and that decision is important later, or even in the sequel to the game.

The problem is that you have to script it manually and because of that all these different possibilities have to collapse to avoid combinatorial explosion. And the outcome of that script is usually not saved anywhere so it can’t affect future relationships. Letting developers (and NPCs) access this data can alter the gameplay experience in new ways.

From what I have seen of Storybricks, it could very easily be used in player generated content. Was that one of the goals during its development?  What limits, if any, would gamers, with no programming knowledge, have using Storybricks in their player made content?

As mentioned above, the original version of Storybricks was developed as a way to help players tell their own stories.  We wanted to make the interface easy and fun to use, so that it did not require deep programming knowledge.

Regarding EQN, you will have to wait to see what aspects of our original design are present in EverQuest Next and how they will be integrated.

What else would you would want gamers to know about Storybricks?

We just want to keep expectations at a reasonable level and drive down the hype. We have seen some posts suggesting things that would be impossible in 10 years time and others that are possible but it would be very disruptive to have in a multiplayer game with a persistent world.

(This does not mean we do not welcome feedback, in fact we read and share internally every post about EQN and AI we can find)

With the Emergent AI feature of EQN, we want to deliver a new type of gameplay experience and given that this has never attempted before at this scale, we expect to iterate on our design a lot.

For more information on Storybricks, visit their website or follow them on Twitter @Storybricks.

Have a comment about the article? Let us know!