We are a Computational Creativity research group, committed to investigating the technical, philosophical, societal and commercial issues that arise in this fascinating field. A working definition of Computational Creativity from this paper is as follows:

We’re a very practical group, and the majority of our research relies on building software prototypes which take on creative responsibilities in one way or another. These prototypes produce artefacts and are therefore makers … and this makes us the meta-makers of course!

Within the context of the EC-funded GRO project, where the focus is on digital games, our three main aims are to:

  • Show that Computational Creativity research is ready for the next level whereby creative software affects people’s lives in positive ways.
  • Democratise game design so that anyone, regardless of their background can produce interesting and engaging digital games.
  • Promote videogames as the most interesting art form of the 21st Century and to present game design as an important creative activity.

To achieve these aims, we are developing software for advanced procedural content generation in games. In one strand of research, we are using these techniques to enable people to efficiently construct games with the software acting as a creative collaborator. In another strand, we are using the techniques to enable the automated generation of entire games, where the software acts as an autonomous game designer. We use our practical research in addition to scholarly study in order to blur the lines between video games and other culturally important art forms, and to perform outreach to people who wouldn’t normally be able to write games.

We are very interested in the question of whether creative software and/or the artefacts produced by it can have an impact commercially. To this end, our research prototypes will have a double life. In one direction, they will be used as platforms on which we will conduct research into what it means for software to be creative. In another direction, we will commercialise the software through sales and consultancy.

GRO Project Manifesto

The following manifesto was a deliverable for the GRO project, which funds much of the activities in The MetaMakers Institute:

We are very ambitious for the research and outreach activities that we will undertake in the GRO project. Our two main research goals are:

  • To take Computational Creativity applied to digital games from research to commercial practice.
  • To democratise video game design, enabling people of all abilities to build their own games.

Our main research vehicle with which to achieve these aims will be the automatic generation of digital games. In particular, we will investigate the processes behind building the kind of more unusual, thought-provoking and quirky games being produced by independent games designers. Such games can be seen as high-art games, where enjoyable engagement with particular ideas, aesthetics and mechanics are more important than high-end graphic/audio realism, vast game worlds and complex non-player character behaviours, which are mainstay components of AAA game titles. We will focus our investigations on games where the player’s imagination needs to fill in some of the game details, and we will design and test player imagination models which draw on the use of game analogies with the real world that the software can use to make more engaging and thought provoking games. In addition, we will explore new genres of games which are opened up in an era of automatic game generation, and we will work at the meta-level by developing software to help us ideate such genres.

Some of the reasons for the above choices of focii for our research are practical, and others come from an ideology of blurring the lines between more traditional art forms and videogames. For instance, entire teams of hundreds of people are employed to build large AAA games. As this will be beyond the power of automated game designers for the foreseeable future, Computational Creativity researchers necessarily need to focus on smaller games. However, we note that the more abstract nature of other art forms such as paintings, poems and music enable audience members a more unique experience and more ownership of each piece, as they project individual interpretations and use their imaginations to engage with the piece. The same can be said for many games, especially those coming from small independent studios, where the impressive scale, realism and complexity of AAA titles are beyond their resources, but where the beauty, quirkiness, innovation and sheer fun of the games produced are unrivalled. We believe it is more interesting and important to research how existing and to-be-invented Artificial Intelligence techniques can be used to automatically invent such inventive and beautiful games, than it is to build software to enable quicker production of game content for more straightforward titles.

While there have been some successful projects in the area of whole-game generation, it is still very much a cutting edge field, and therefore ideal for the GRO project. Success with such re- search will provide a powerful vindication for the benefits of Computational Creativity as a whole, as building games involves so many creative activities across all forms. Moreover, focusing on the more artistic, thought provoking games of the type developed by independent developers, forces a higher emphasis on simulating the innovative aspects of creativity rather than the craft aspects. Both craft and innovation are important to game design, of course, but artificial innova- tion is certainly more important to Computational Creativity as a field at this point than artificial craft. Software for producing visual art, music, textual works and game components can currently produce output to a high standard under the guidance of people, but it is rare to see software that is genuinely innovative. In addition to promoting and advancing Computational Creativity research, success in the area of automated game design will drive forward videogames culture, as communities widen their perspectives to engage with more advanced notions of Artificial Intel- ligence in game design, as they did with more advanced notions of in-game Artificial Intelligence driving non-player character behaviours.

One of the benefits of researching automated game design systems is the building of tools to assist people in writing games. In order for software to build entire games, it will need to be able to invent and engineer all elements of games, including content, rules, high-level ideas, characterisations, dialogue, directorial instructions and mechanics. This means that we will be able to write interfaces to the game engineering routines which enable people to guide the au- tomatic generation of game elements, and we will research how people can use these tools in semi-automatic game design. We envisage a system where the user acts as the creative force in their game’s design, and various helper routines enable them to achieve their aims. However, the system will be able act as more than a mere set of tools. In particular, it will be able to drive the game generation process through game ideation, drawing on related work in the EC-funded WHIM project. This means that the user could come to the game design project without a full idea of what they want to achieve – perhaps with a piece of music or a particular aesthetic – and still use the system to produce games.

Our main impact vehicle with which to achieve the two main aims above will be the presen- tation of digital games as important cultural artefacts which are inherently as important as books, paintings, films, musical compositions, etc., and similarly presenting game design as an important and fulfilling creative activity. This will help our research achieve maximum impact in the following ways. Firstly, it will increase the demographic for the consumption of small, thought provoking and artistic games of the type that our software will produce. That is, as games are increasingly taken more seriously as intellectual stimuli by an increasingly large portion of society, so the demand for more bespoke games will increase. This will widen the market for automated game design, with an emphasis on individualisation requiring mechanical generation of games to keep up with demand. It will also drive demand for further innovation in game design, as people want to be continually impressed by the intellectual sophistication of the games they play. This will again provide impetus for research into how Computational Creativity techniques can innovate in game design.

We believe that one of the main reasons Computational Creativity research has not yet had a arger economic and societal impact is a lack of understanding in society that software can and does exhibit creative behaviours which produce artefacts of real value. There is a popular mis- conception that creative activity is a human-only endeavour, and that bit-and-bytes nuts-and-bolts machines cannot possibly be capable of such behaviours, due to their mechanical, programmed, nature. This is a serious issue which we have addressed, along with various suggested remedies, in a number of philosophical papers. With the GRO project, through The Meta- Makers Institute, we will take our philosophical standpoint from theory into practice. In particular, through events aimed at both showcasing the research results from The MetaMakers Institute and promoting games as art, we will always have an emphasis on public engagement with the idea of software being creative. Moreover, we will develop software that not only produces games of real value to players, but is able to invent new processes, describe its behaviours, frame its work and place its output and process in broader artistic contexts, adding to the impression of creativity.

With a widening of participation in game design enabled by the automated tools we have planned, and the productivity of existing game designers enhanced through our tools, so that they can be more innovative, more people will be able to express themselves creatively through the medium of games in more ways. In time, with computationally creative help, we expect that it will become as easy to design a bad videogame as it is to write a bad story, paint a bad picture or compose bad music. Of course, once people get over the hurdle of producing their first, bad, game, then their creative potential will be unleashed, and we should expect innovative, personal, surprising and thought provoking games to come later. With initiatives such as the No One Left Behind project where children from age 8 to 17 will be encouraged to explore science, maths, history and languages by designing and building games, game design will be promoted more and become more popular. This expansion of interest in game design will need to be coupled with advancements in game design tools of the type we plan to build. It could be a very exciting time indeed ahead for the arts, and – as always happens – the creative industries will be at the heart of this economically, with benefits for society all round.

Within the context of the two main research goals above, our strategy for how research is undertaken, disseminated, commercialised and sustained in The MetaMakers Institute is given in the next section, but it is worth highlighting how the choices in this manifesto will help achieve those two goals. Firstly, as a medium, digital games are very important, both culturally and economically. Of all the types of media studied as application domains within Computational Creativity research, we believe that games offers the best opportunities for that research to have cultural and commercial impact. In addition to opening up new research avenues, automated game design will open up important new commercialisation routes, in addition to existing ones within the context of procedural content generation. This will lead to success with respect to the first research goal, of taking Computational Creativity from research to commercial practice. Secondly, by developing interfaces to automated game design methods, we will engineer tools that enable software to act as a creative collaborator in building games. Writing a digital game is a difficult and time consuming process – indeed, entire undergraduate and postgraduate degree programmes are devoted to teaching game design. With respect to the second research goal above, automated tools which take on more of the creative responsibility in game design will cer- tainly help to democratise this activity, enabling more people from wider backgrounds to engage in game design. Finally, for the benefit of both of the research goals, we plan a programme of outreach activities to promote game design as artistic activity, games as important artistic arte- facts, and software as being able to act in autonomously creative fashions. This will lower any barriers to acceptance of automatically created games, hence will simultaneously broaden the market for such games and broaden the demographic for potential game designers as users of the semi-automatic tools we will produce in The MetaMakers Institute.