The MetaMakers have been on the road to scientific conferences and events this summer, presenting our research and showing off technology prototypes.
But first, before our own travel commenced, we started the summer off with a high-profile visitor here. Yang Lan, host of a massively popular talk show on Chinese television, visited to film footage for an episode they’re doing on artificial intelligence. The crew had visited Google DeepMind’s offices in London, home of the recently victorious Go player AlphaGo, and then headed west to see us in Cornwall, in order to add a segment covering the take we have here on the more creative, artistic side of artificial intelligence.
Later in June, I (Mark Nelson) traveled to Hong Kong to give a talk at technology incubator InnoCentre, entitled “Creative Tools in the Smartphone Era”. This talk laid out a vision for how the landscape of creative design tools is changing due to the fact that smartphones are becoming a (or even the) primary computing device for many people, and how the Gamika technology we’re developing in The MetaMakers Institute fits into that shift.
A week later, almost the whole Institute traveled to Paris for the International Conference on Computational Creativity (ICCC). We presented two papers on aspects of the Gamika project at the Computational Creativity and Games Workshop. In the first paper, “Towards a Computational Reading of Emergence in Experimental Game Design”, presented by Simon Colton, we traced through three design sessions where a novice, intermediate, and expert user made new mobile games. By identifying the moments of emergent design in these sessions, our goal here was to outline a technical research agenda for the techniques an AI game designer would need in order to produce similarly emergent moments in automated game design. The second paper, “Automated Tweaking of Levels for Casual Creation of Mobile Games”, presented by Ed Powley, focused on a more specific part of the automated game-design problem: if you have a game that is almost good, but not quite, can it be automatically tweaked and fixed up? We demonstrate a method for doing so, using auto-playtesting sped up with offline machine learning, to evaluate a large parameter space.
Most recently in August, Michael Cook and Mark Nelson went up to Dundee, Scotland, for the first joint meeting of DiGRA and the Foundations of Digital Games (FDG) conferences. Michael presented two papers at the Procedural Content Generation Workshop. The first discussed two experimental games, CLAY and Tombs of Tomeria, in which the player being able to reparameterize a level generator is used as a form of exploratory gameplay. The second presented initial work on Danesh, a tool to help game developers explore, improve, and understand procedural generators. Mark presented a paper on the weird and wonderful patent for the 1983 arcade game Tapper, whose existence he recently discovered, which he reads as demonstrating a sort of unintentional method of formalist game studies. Our Falmouth colleague Michael Scott was also in attendance, participating in a panel on navigating postgraduate research in game studies.
As we head into autumn, two more events are coming up. In September, Mark heads to the Computational Intelligence and Games conference in Greece to present two of our papers. One is on semi-automated level design for handheld games using Gamika technology, and the other investigates scaling properties of the Monte Carlo Tree Search (MCTS) algorithm on the GVGAI corpus of arcade games. And then in October, we’re hosting here in Falmouth a two-week festival, Games as arts/arts as games. This year is the inaugural festival of what will be an exciting annual event. You’ll hear lots more about this festival from us soon, but for now check out the festival’s website, and consider coming for a visit to Cornwall this October!