A screenshot from a game for change called MOLT, in which players diagnose malaria.
There’s something magical about the math of crowdsourcing. You bring a bunch of people together and collectively they often outperform the group’s smartest member, or even a computer. And if you consult this chart you’ll see that that’s an increasingly challenging feat in games these days.
So I was excited to come across a crowdsourcing game for change called MOLT, from the folks at UCLA’s Ozcan Research Group. Players get some basic instructions and then they have to identify blood cells that have been infected with malaria. The team at UCLA created this game for change because pathologists usually have to check between 100 and 300 different views of a simple blood smear under a microscope in order to diagnose it as negative or positive.
Obviously this takes buckets of time and decreases the number of cases they can examine. An intelligent algorithm learns from the players’ decisions, and with groups as small as 20 gamers, all biological and epidemiological laypeople of course, the algorithm has been able to get within 1.25% of the accuracy of a trained medical professional by combining the player’s answers.
The UCLA team feels that if they get a larger pool of people playing their game for change, they can boost the accuracy even higher. However, as exciting as the technological implications are, one potential barrier is that the game is not engaging enough to encourage long-term gameplay, which might restrict the amount of data they can collect.
Some fixes they can implement to make the game more engaging:
1) Have a player’s score accumulate over time rather than restarting at the beginning of every round.
2) Institute a leaderboard. My score keeps going up in a vacuum and it means nothing to me. How am I doing compared to my peers and even compared to my own past performance? They can take some pointers from this great post on Undercurrent on “de-suckifying leaderboards.”
3) Give the player a sense of his or her impact. When a player plays a game for change, they deserve some assessment of the change they are making. Free Rice, for example, puts the score in terms of grains of rice donated. If the unit of change is increasing the intelligence of the algorithm, maybe there’s some way to quantify that for the player.
The UCLA team obviously has a strong science and technology background to have achieved these impressive results, so far, but I think if they consult a game designer, they will be able to take their game for change to the next level.
