That the homo ludens ('playing man') can give expert scientists a run for their money at least when it comes to predicting protein structures has been shown by the collaborative multiplayer online game Foldit. But can lay people also compete when it comes to the development of algorithms? A new paper by Khatib et al. tests this by giving the Foldit players the opportunity to create, edit, share and rate simple macros that automate specific strategies. Successful algorithms were quickly adopted by many players. In particular, two recipes for the energy minimization of the protein structure were used extensively. Interestingly, the two algorithms share many features with what was, at the time, an unpublished method developed by the laboratory of David Baker and were more efficient than the previous state of the art. As the players could manipulate only a small fraction of the possible variables and base algorithms of the underlying Foldit program, this is a remarkable achievement. In the next version of the game, the players will have control over many more features and it can be expected that even more interesting ideas will be developed by the Foldit community. (Proc. Natl. Acad. Sci. USA 108, 18949–18953, 2011)