Alternate Reality Gaming: Can we apply the Swarms paradigm to translational medicine?

I haven't written about serious games in quite a while, but that doesn't mean the topic hasn't been on my mind in the interim. The Feb 2008 ACM Communications issue has a cover article entitled "Alternate Reality Gaming", by Jeff Kim of UWash, Jon Allen of UCSF, and Elan Lee, creative director of two of the pioneering ARGs - The Beast and I Love Bees.

As far as I could recall, I had never heard of the concept (but then I'm pushing 60, so who knows what I knew two weeks ago), but googling ARGs led me to a post in Christy Dena's blog, which pointed me to Jane McGonigal's blog, where she was celebrating her recently being featured in the Harvard Business Review's annual "Top 20 Breakthrough Ideas" List.

Jane knows whereof she speaks, it seems - according to the article by Kim et al, she played the "community lead" role in the ILB game, which sounds like it basically meant being in-world (the virtual world) 24/7/365, observing events and making sure the game reacted in timely and innovative ways.

Here's a key excerpt from Jane's HBR article:

In the coming decade, many businesses will achieve their greatest breakthroughs by playing games—specifically, alternate reality games, or ARGs. Custom-designed ARGs will enable companies to build powerful collaboration networks, discover solutions to specific business problems, forecast opportunities, and innovate more reliably and quickly.

Why? ARGs train people in hard-to-master skills that make collaboration more productive and satisfying. Playing an ARG teaches 10 collective-intelligence competencies. These include cooperation radar, a knack for identifying the very best collaborators for a given task, and protovation, the ability to rapidly prototype and test experimental solutions. Using these skills, players amplify and augment one another’s knowledge, talents, and capabilities. Because ARGs draw on the same collective-intelligence infrastructure that employees use for “official” business, games will map directly to a familiar reality—no translation required.

As these competencies mature within a business, ARGs will provide a truly stimulating framework for doing everyday work. Few meetings are as engaging as an ARG, whose emerging narrative evokes players’ shared sense of urgency and whose puzzles and clues deepen their curiosity. The structure for collaboration is clear, with players rallying around explicit goals and continually sharing theories, tactics, and results. Playing also generates compelling momentum: The puppet master monitors and rewards participants’ efforts, and times the release of new challenges so that players experience multiple cycles of success.

OK, so business will soon be a lot more fun than it was during the first 35,000 years since the last Ice Age. What does this portend for translational research. Quite possibly, it's big news.

Let's digress for a moment and consider the Swarm paradigm from the world of robotics, the promise and challenge of which are elegantly described by James McLurkin of MIT's CSAIL:

As robots become more and more useful, multiple robots working together on a single task will become commonplace. Many of the most useful applications of robots are particularly well-suited to this "swarm" approach. Groups of robots can perform these tasks more efficiently, and can perform them in fundamentally different ways than robots working individually. However, swarms of robots are difficult to program and coordinate. My work focuses on developing software and programming techniques applicable to large swarms of robots, with populations in the 10 to 10,000 range. Software is evaluated on the 100-robot iRobot swarm.

So, the problem is that swarms are difficult to program and coordinate? Maybe that's because robots are still pretty darn stupid, and in my opinion will remain so for some time to come. Humans, though, are both very smart and very imaginatively adaptable - the willing suspension of belief that is essential to theater is very likely integral to the human psyche. Massively multi-player online role-playing games (MMPORGs), ranging from the Dungeon game of the '70's to such byzantine modern games as World of Warcraft (WoW), have long been a vehicle for creativity in which thousands of people have invested considrable time and energy collectively solving problems (and generating others, to keep the game going).

The programming behind WoW is no doubt enormously complex, but very little of it involves telling the players what to do, or even devising the "business rules" that govern their actions. Humans need minimal guidance because they are capable of DWIM (Do What I Mean) behavior: they can employ tacit knowledge and context to extrapolate meaning from whatever signals the game provides. Moreover, if they don't understand the signals correctly, they will improvise, and a game like WoW is designed to permit a certain degree of just such improvisation.  The key challenge in robotics lies in surmounting the inability of robots to mimic these most human of traits.

ARGs as a category subsume MMORPGs, but are not constrained to the online environment. In I Love Bees, for example, bottles of honey sent to Silicon Valley game designers and a near-subliminal URL reference in a a video game trailer were key inciting elements in the game, though it also had very important online components. (Did you visit ilovebees.com yet?) Kim et al. describe ARGs as follows:

Alternate reality games represent a new genre of digital gaming designed to blur the distinction between a player's experience in the digital world inside the game and the real world outside the game. Combining online information and real-world events, ARGs bring gamers together to collectively solve puzzles and advance a game's storyline. Part of what characterizes an ARG is that the game universe is not explicitly limited to a particular piece of software or set of digital content. A typical ARG would not even acknowledge or promote the fact that it is a game, yet every Web site or discussion group may contain and reveal a potential clue...

Two features of a successful ARG are a compelling storyline and collaborative game play. In it, a plotline is narrated and delivered through multiple communication channels, including Web pages, email messages, phone calls, and print-based mailings. Gamers use them to track the story's progress. In two of the first ARGs to achieve a critical mass of more than a million participating gamers—"The Beast" and "ILoveBees" (2004)—compelling plot development made it possible for the designers to attract, retain, and increase the number of players. The collaborative nature of game play is another important ARG feature. Game players construct their own means of interacting through Web-based discussion boards, email messages, and real-world gatherings. Through this interaction, players communicate with one another, share their knowledge, offer interpretations of the storyline, and gather the information necessary to progress toward the game's conclusion.

This commingling of the virtual and the real is characteristic of WoW and other such games as well. Which brings me to the translational science angle.

One key problem in translational medicine is to tie knowledge from basic science to the signs, symptoms, lab reports, and other analog elements that form the messy information architecture of the clinical encounter. Suppose we have a drug with dangerous, horrible, and fatal side effects, but deterministically capable of curing an otherwise terminal disease in the fraction of patients (let's say 60%) it does not kill. Until now, such drugs have never made it to the clinical setting because of the you-bet-your-life element: if you have six chances in ten of being cured, and four chances in ten of dying a horrible death, you might willingly take the chance.

Now, suppose the patient is your three-year-old daughter. Feeling a little more conservative, and a lot more hopeless? You bet you are. Can you live with yourself if you consent to the treatment and dies a horrible death, looking into your eyes as she slips into the blessed oblivion of the hereafter? No. But can you live with yourself if she dies of the disease, looking into your eyes as she slips into the blessed oblivion of the hereafter, with you never knowing if she was one of the six in ten who would have been cured? No. I'm glad I'm not in your shoes...

It is very likely there is a way, through genomic, proteomic, or metabolomic analysis, to deterministically identify the characteristics of the patients who will respond to the treatment. Some configuration of molecular structure and function will characterize the survivors, and its absence will characterize those who succumb. That's translating science into medicine. The problem thus far is that the search for that combination of factors is much, much harder than looking for a needle in a haystack. Unimaginably massive computational power is needed. Scientists are attempting to devise creative solutions, like David Baker at UWash with his Rosetta@Home project, which propagates instances of his Rosetta algorithm for protein folding analysis to home computers, to operate as a screen saver when the computer is idle, applying the algorithm to some tiny subset of the combinatorial explosion that must be waded through in any given problem.

Sounds a lot like the Swarm paradigm, doesn't it? This is what makes me wonder if there is potential to apply the ARG to the problem.

Imagine if very bright young people who have mastered some critical mass of knowledge of molecular biochemistry, who take on the above-described translational science problem as a game. It's a social game, so they get together with friends - in both the real and virtual worlds - and share knowledge and ideas. The search becomes a lot more intelligent, because humans are much better than computers at pruning search trees.

I have to admit I haven't pursued this thought very far. This post is the result of three days' lunchtime poking around and virtual scribbling. Thus far the applications of the ARG approach I've been able to find are in the realm of entertainment, business, and of course the military (which my dyslexic but subconsciously intelligent fingers keep wanting to type as "the limitary"). What if we could apply ARGs to just such problems as the one I described above? The story was a hyperbole, of course,  but there are examples from the real world that may be less extreme but just as problematic, such as the recent furor over Vioxx.

It makes me wish I was young enough to get hired as a game designer... <sigh>.

Lunch is over - time to get back to work.