[cross posted from my cacm blog]
Do you remember the overweight hamster in the film Bolt who travels everywhere in a Perspex ball? Can you imagine being stuck in such a ball, rolling through a pleasant country meadow? Curiously as you travel, the ball deforms into sausage shape, then inflates again in a different direction. It does this regularly, with no discernable pattern. Your goal is to survive until the sun sets as you combat gremlins and various other airborne threats. Your only means of defense is a grappling hook, which you can use to deter the enemies. You can amass points by firing a grappling hook from the centre into the furthest away point of the deformed ball. To put it another way, can you imagine finding an optimizing pair of points for maximal re-growth of a Bloch sphere, and thus solve a problem in quantum information back flow?
This is not a terrible dream in which you are the victim of crazed Lillupitian quantum physicists, but rather the prototype of a scientific discovery game we are working on at Heriot-Watt University, Scotland*. Scientific discovery games “focus on leveraging human problem solving ability to solve computationally difficult scientific problems” (Cooper, Treuille & Barbero, 2010;1). There is a nice review article about the use of such online games to solve scientific problems in the March issue of CACM. It describes how 57,000 players of the Foldit game helped biochemists to discover the structure of a protein which helps the human immunodeficiency virus multiply, a long standing problem in biochemistry. It also summarises other crowd sourcing games in which players tag photographs, translate phrases and identify astronomical objects. All of these games take advantage of humans’ superior image processing capacity. In the case of Foldit and other biochemistry games, the target problem is highly spatial: the 3D representation of the protein in the game closely matches the structure of the actual protein in real life. But can scientific discovery games still be useful if the target problem doesn’t have an obvious 2D or 3D representation which our visual systems can get a grip on? In general, what sort of scientific problems are amenable to representation in game form? Which sorts of problems benefit from non-expert human intuition and to what extent does this approach outperform computational techniques?
Back to the crazed physicists and their hamster ball. We’re attempting to solve a problem in dynamics of information flow in quantum bits which physicists often visualize using a three dimensional sphere (see Haikka et al, 2011 if you care to know the details). For now, we are creating the game around a three dimensional version of the problem, for which there is a known solution so that we can easily check whether the players arrive at the right answer. But as I understand it, there is a whole set of more general problems in multiple dimensions for which no solutions are known. This will be a real test for the scientific discovery game approach. Quantum physics is notoriously un-intuitive. While humans are evolved to be dab hands at reasoning about three dimensions, most of us would feel quite ill at the thought of reasoning in thirteen dimensions. In later stages of the project, we need to think of ways in which we can divide and conquer the problem in such a way that individual players can each solve a small aspect of the problem within a 3D world, but somehow the solutions of many players can be put together in a coherent solution to the more general high dimensional problem.
For now, even designing a game for the simple three dimensional version of the problem is proving quite challenging. Because the physicists often think of the problem in terms of geometric shapes, we decided to exploit this in a first person shooter style game instead of the puzzle game genre which suits Foldit. We started off by modelling the Bloch sphere according to a set of equations, stuck a camera in the middle, and made a grappling hook which the user can throw at the sides of the sphere to get points. We logged the required data for the physicists to study later. This was completely faithful to the physics problem we were trying to solve, but – boy! – that was a boring game. In fact, it probably couldn’t be described as a game at all, as there was no challenge for the player, no rewards, no ways to lose, and no forms of punishment. So we had to insert classic game mechanics into the basic representation of the game world in such as way that the player would have fun while still achieving the problem solving requirements of the physicists. We ended up with the design described in the opening paragraph. We’re refining the game now, and will be testing with users over the next few weeks. If you fancy trying out the hamster ball whilst solving open problems in quantum physics, you’re most welcome. Leave me a comment below.
*The project team are Sabrina Maniscalco, Suzanne McEndoo, Stephen Kilbride,John Truesdale, Sandy Louchart, Theodore Lim, Judy Robertson.
Cooper, S., Treuille, A., & Barbero, J. (2010). The challenge of designing scientific discovery games. Foundations of Digital Gaming. Retrieved from http://dl.acm.org/citation.cfm?id=1822354
P. Haikka, S. McEndoo, G. De Chiara, G. M. Palma, S. Maniscalco, Phys. Rev. A 84, 031602(R) (2011)