Winners and losers from 83 players of
an introductory level in Foldit. These visualizations
are complex, which is not surprising given Foldit's
many degrees of freedom; however, we can still dis-
cern some characteristics of player behavior. Even
though winning structures are near the starting con-
guration, players still explored wide swaths of the
state space, suggesting they did not immediately see
the path to victory. As such, one might guess that
those who quit did so because they became lost in
poor protein congurations, which would show up
as movement away from goal states. However, play-
ers who eventually quit often explored quite exten-
sively near goal structures. Thus they had the cor-
rect high-level ideas, but for some reason did not
perform the remaining optimization on these 'good'
structures to nish the level.
74