Previous work in automatic content generation has traditionally relied on procedural methods and has succeeded in creating realistic game environments. For at least thirty years, many different procedures have been proposed to automatically create such content as terrain, tree and plant models and urban environments. Results show that a common shortcoming in traditional procedural methods is the lack of control over the generated output, due to the randomness that these algorithms carry. Therefore, researchers are now aiming at more controllable procedural methods, allowing designers to steer content generation.

Work developed by Muller, Parberry and many more, on controllable content generation, is enabling procedural methods to become more flexible and accurate. While maintaining its automatic nature, these methods are allowing game designers to steer automatic content generation by means of a better expression of their intent. However, control over the generated output only relates to defining physical properties for the desired environments.

These results encourage further work on using controllable content generation to adapt game environments. To do so, the challenge ahead becomes twofold: (i) controlling content generation by using knowledge about all sorts of high-level targets (game goals, player expectations, qualitative features), and (ii) ensuring that this knowledge is generic and applicable to each player. Solving these problems will make it possible to create customized content, in the sense that game environments would be generated before game time, according to knowledge about individual players. This knowledge should go beyond player preferences (e.g. favorite colors) and focus on what affects his playing motivations or purposes. As an example, consider a world domination strategy game and a player who aims to be an economic leader and use trade strategies. A matching virtual world for this could be generated so that, for example, the player homeland is located close to natural resources, in an intermediate position between opponent cities, and has a flat and oceanic geography (that allows the fast development of travel strategies).

Current research is already tackling some of the challenges identified above, and its methods could be valuable to future work in customized content creation. Semantic and declarative modeling techniques are already capable of controlling procedural methods by embedding knowledge in virtual objects. This encourages further research on customized content generation through semantic techniques. By developing such semantic schemes, further specifications can be used to embed meaning about player purposes in the virtual world and its objects. This meaning could be used to control the generation of customized content.

Chen and Michael have already identified the main challenges that assessment in serious games is facing. Most traditional methods for assessment are not accurate enough for serious games, since they are inspired by the simple feedback mechanisms used in their entertainment counterparts. Identifying and reflecting on mistakes and decisions is especially important when considering serious games. As such, for assessment to improve the player experience, log information and teachers / instructors knowledge should be fully explored and, in same way, converted to assessment information that can be incorporated back in the game, to guide its course.

So far, research in assessment for serious games has been mainly centered on After Action Review (AAR) methods. However, results already demonstrate that the direction identified by Chen contains a lot of potential. AAR systems for military simulations are already being used in innovative ways they were not designed to, not only for assessing past behavior, but especially for planning new future training exercises. In these systems, real-time in-game and AAR assessment information establish an emergent domain culture that could allow the co-creation of future game scenarios. Assessment information could be better explored and even incorporated to potentially influence content in serious games.

There is typically a lot of valuable information in game logs and emerging from AAR sessions, in serious games. This information is far from being fully explored by the game itself, to improve game-play. This happens because logs usually offer an enormous amount of unstructured game data that is therefore difficult to interpret and use. Moreover, AAR information emerges to engage communication between trainees and their instructors and it is not incorporated back in the game. Using this information as a source to guide adaptivity seems a promising, unexplored area. Assessment information can became valuable not only per se, but also to improve game-play, both while playing and in future interactions.

The challenges ahead indicate multiple research directions on what and how to adapt. On the one hand, assessment information could be used to re-generate ”try again” game scenarios, adapted and focused on what the players failed on the previous session. So, offering a re-generated game scenario could simultaneously allow a better understanding of what went wrong, and better opportunities to succeed. Work on this direction should tackle, for example, customized content creation, e.g. adjusted to better achieve a learning goal. On the other hand, on-line adaptivity can also be influenced by assessment methods. Game scenarios, and even intelligent agents, could adapt to the assessment of how players are deviating from these purposes. More than a matter of measuring player performance, it would involve interpreting how that performance is being achieved. One example would be to adapt the game because the player is succeeding in learning, but in a slow pace (instead of because he is just performing too good or too bad).

As an important note, researching the relations between adaptivity and assessment seems to be limited to the serious games domain. We still need plenty of human expert knowledge to make sense of the correct assessment information, either during or after game time.

Serious games are becoming increasingly established, but they are still coming of age in terms of player experience. Most serious games are developed ad-hoc and lack sound theoretical foundations, which leads to a number of drawbacks: they are predictable, impersonal and limited by stereotyped training scenarios. In particular, serious games should be designed to prevent (i) training modules from following rigid patterns, (ii) unattractive and predictable game-play, (iii) little advantage being taken of user data collected throughout the game and, worst of all, (iv) little knowledge being employed to guide the course of the game. For example, it would not be effective if all medical trainees in a certain course would have to follow the same timed procedures, in the same standard scenarios, independently of their personal skills or difficulties; and it would be (pedagogically) even worse if their final scores could not be traced back to particular in-game moments. Trainees might just learn how to play the game, instead of how to think and act in similar scenarios.

Many researchers agree that serious games have to become more challenging, unpredictable and user-centric, to be fully embraced as an effective way of knowledge transfer. To prevent the shortcomings above,  serious games should include virtual scenarios that adapt to what and how players need to learn in a given context. This scenario adaptivity should benefit players, by providing them with more flexible challenges and a broader range of (pedagogically) meaningful ways to solve them.

With my research, I expect to contribute with a methodology for supporting the creation of such adaptive virtual scenarios. This methodology should focus on adapting scenarios to: what players need to learn, how they should learn it and what did they failed to learn. We argue that Instructors (or Trainers) already possess this specific knowledge and, as so, are in a privileged position to steer scenario adaptivity to the expected benefits. With our methodology, the Instructor will use his knowledge on what should be learned by a specific player, to automatically generate virtual scenarios that are suited to player characteristics and learning goals. The Instructor will create in-game situations where objects and events adjust, in real-time, to the player performance and the way he should be learning. After a game session, the Instructor will assess on players performance and that evaluation will be used to re-generate game scenarios where players can play again, focusing on what they failed to learn.

My aim is to embed such knowledge into game worlds and objects, which will become more meaningful in different ways of adapting to benefit players. My research focus on supporting such virtual worlds that are enriched with meaning (or semantics) about learning goals, player performance, game decisions and assessment evaluations.

The game uses perspective shifts between 2D classic views and the 3D “real-world” view. Despite Super Paper Mario similar perspective changes, the innovation in Fez is turning the “view shifts” into a central piece of the gameplay. To succeed, the player has to solve visual 2D/3D projection puzzles and become a gamer version of M.C. Escher.

Looking at the video, you can’t stop feeling captivated by such a simple game. And why? Because it is remarkably different from anything we seen, but at the same as familiar as The Beatles. Personally, it makes me wonder about the development of such games. It’s not a surprise that creativity is more abundant in the indie games community, when compared to commercial games. After all, there are less restrictions in deadlines and the hunger for profit: it’s more craft than industry.

But this little example makes me thing that game design and development in indie games might actually offer bigger challenges, and therefore better innovation opportunities, than in commercial games. I risk saying that imagining, designing and building from scratch a 2D/3D puzzle world for Fedge might be more difficult (and rewarding) that yet another map for another version of the Call of Duty more than stable game engine.

There have been some efforts in this field that led to some meaningful conclusions. Theory of flow, immersion, harmony, Malone, Koster are the words to search to get a glimpse at what the academia is researching on fun in video games. However, I also think that the industry is a good perspective to look at. In theory, it should be the better place to look at: people buy games which are fun to them. But I have enough years on my back to find myself confused by the game industry.

Let’s take a look back at the big massive successes in the industry. Cutting the story short, I am intrigued on the genre shifts that occurred. After puzzle games, platforms ruled the world, everybody stills knows Mario, Sonic and Prince of Persia but are they still big hits today? Then came (and mysteriously disappeared from the face of the Earth) the graphical adventure genre: Curse of The Monkey Island, Day of the Tentacle, Sam & Max, Larry and a lot more were all we played. Then, first person shooters (Doom, Quake, Unreal) and this is probably the longest surviving genre I have seen, in terms of massive adulation, it still holds up. But gamers did eventually moved on to MMORPG, after all World of Warcraft is a genre on its own. But now, in 2010, the big hits seem to navigate around the third person action genre: Uncharted, Assassin’s Creed, God of War, Devil May Cry, Batman. Take a look at what are the best games of 2009 here. Apparently third person action games are the most fun, for the time being.

I am very curious on the reasons why genre adoption evolves like this. Is it a generational thing? After all, music hits also evolve in genres. Or is it actually just a technology factor and third person action now incorporates the best of platforms, puzzles and shooters? Maybe it is even simpler: a big game has huge success and the industry is automatically flooded with copycats to mirror that success. The market eventually saturates, anxious for a next hit.

Anyway, I find genre evolution interesting because it makes me think that it might be closely tied to skills. And forgetting the industry perspective, we can transpose it to the abstract notion of fun in video games. Tomorrow I will find that another game is much more fun than today’s game because it appeals to a new skill that I haven’t acquired yet. After mastering the game, the genre, I have that skill, it becomes easy to me. Now, is there a new skill out there for me to buy? You could call it challenge, learning, curiosity or the human unconscious seek of knowledge.

In conclusion, researching the possible relations between fun in video games and skill curiosity is something that is definitely interesting and could have surprising results. But I do have other stuff to do. Well, I might even came back to this in a few years… but in the meantime, anybody wants to try?

Ok, it does seem a bit too much, the video is produced like we are looking at the most revolutionary and best thing that ever existed. But, being fair, there are enough good signs already  to show us we are looking at a serious project and a sustained release in the winter of 2009. Onlive’s team has people that were responsible for several stuff in Quicktime, Eidos Interactive, Mozilla and Activision. And At&T and Warner Bros are financial investors in Onlive. And it is announced that Electronic Arts, Ubisoft, Take-Two, Warner Bros. Interactive Entertainment, THQ Inc., Epic Games, Eidos, Atari and Codemasters are already lining up their games to be served through Onlive.

Of course that, in these days, everybody who is talking about this is either suspicious and/or skeptical about the feasibility of this huge technically complex game platform. Hardware required for Onlive servers, video compression, broadband speed, latency in processing commands (after all the game is computed on Onlive servers!), all are thoroughly pointed out as concerns. There were successful demos in GDC 2009 and E3 but still there are plenty of haters out there.

But, let’s not focus on that, let’s assume that everything is technically fine and that we always have a good Internet connection (Onlive talks in 1.5 Mbps for SDTV resolution and 4-5 Mbps for HDTV). Yes, my reasoning is much simpler: why should we buy an Onlive service instead of an XBox, Wii or Playstation?

Why is this better than having our own games running in our own consoles, which are physically ours even if Sony, Nintendo or Microsoft explode or stop producing? How can we replace going in some random frat boy reunion, lost in the middle of nowhere with no broadband access, with XBoxes under ours arms? How can we borrow and trade games with friends? How can we sell our games on eBay? How can we hack Onlive with illegal chips, like we do with our Playstations? Can we just pause for hours and go to sleep without worries? And why should we give up on having Wii remotes, bullet proof HDMI graphics, Blu-Ray movies and Halo? These are the reasons why we buy these things in the first place, instead of a PC and a bunch of hacked games.

We should give up all this… in trade of? Answer these questions, Onlive, and I am a believer. Until there, you are just ahead of your time.

This seems like a valid and interesting statement, it could be on some nice abstract, but it does raises some concerns and probably more among gamers, who are always very protective about their world. If it’s adapting, does that mean it’s making it easier and taking the challenge out of it? And if it is a game tailored individually to everyone, how to compare performances? Does this means the end of bragging rights?

The research on adaptive games must strongly tackle these issues before anything else. Games are there to serve players. There are some who advocate that the player should not know he is playing on an adaptive environment (problem solved!!!), but that is not the point. The goal is to improve the whole game experience (including bragging rights) and immersion for the player, so adaptive games should clearly be presented as a better solution for that, without loose ends.

I do think that is possible. Adaptive games should keep a strong boundary between what evolves and what remains, so that players are still enjoying the same game with the same rules. Also, how and what to change in the game are the cornerstones for the “too easy vs. too hard” argument. Of course, constraining the possible adaptation to a specific game, its players and a set of requirements in terms of game experience is doable and it might not even touch on the difficulty issue. However, the real challenge, from the research point of view, is to construct sound theoretical foundations and generic (with minimum constraints) mechanisms for adaptive games.

But Rome wasn’t built in a day and serious games seem like a good place to start. Their objectives are clearer than entertainment games: training and education. The goal of the game experience is that the player learns everything it needs to learn or, at worst, something new. Scores are only kept to know what and how to play again. It’s not about evaluating performance, it’s about learning. And adaptive games can provide more flexibility and thus higher success rates, in terms of this serious game experience.

Let’s try and jumpstart the future: adaptive gameplay.