Over last few months we spent large amounts of time analyzing the patterns of recipe use by folders. We discovered many interesting things and drew conclusions. One of the conclusions we came to is that overall folders don't share enough. We want you to share more. To that end we want to provide an incentive for recipe authors to share their creations despite the potential for losing their competitive edge. We haven't figured out any specifics yet. So at this point, it would be interesting to find out your ideas about how we can encourage authors to share more.
ALL recipes should be freely available. Eliminate the Group "private" recipes.
As it is now, only a few teams participate in Evolving, so maybe the whole Evolver thing shoudl just go away. The Groups can still exist, but compete on the basis of the individual player scores.
As it is now, there are many players whose only contribution is to wiggle up a fraction of point here and there, and get highly ranked for that minimal effort.
Regarding the "players are highly ranked for minimal effort issue": I've discussed this before the following feedback posts.
http://fold.it/portal/node/986077 (mentions the issue of picking up a cheap evolution from a XXX.8 or XXX.9 structure)
http://fold.it/portal/node/986220 (the calculation for evolver scores are incorrect, in the sense that the denominator takes in the number of soloist players rather than evolver players)
As for the idea of making evolver solutions available to everyone: this has been tried before in the form of "all hands" puzzles, where the top solutions that are sufficiently dissimilar in structure are automatically uploaded every 2 hours or so. What the developers of Foldit have found, however, is that these puzzles actually discourages having a variety of solutions because most people would simply focus on only the top solution. Hence, as counter-intuitive as this might sound, having many different teams limit themselves to evolving only their own structures is actually better from a scientific standpoint because a more diverse set of structures would result.
==> Given this, a better idea might actually be setting caps to the group sizes… because the marginal benefit to adding an extra group member will approach zero at some point.
Finally, regarding the issue of encouraging sharing: here's an idea.
One possible way to give players credit for sharing their recipes/scripts is to award more points (which could be a category separate from global soloist and evolver points). The difficult part, though, is enforcement– what does that stop people from creating a script that consists of only global wiggle and global shake, which will often generate a lot of points at the very beginning when given a bad starting structure? To protect against such abuse, a possible workaround might be awarding points to the automated methods only if they prove pivotal– for example, if they raise a solution's score from the 75th percentile to the 99th percentile. This requires extreme amounts of extra work, though, as it will require tracking (and analyzing!) score vs. time for every single player; plus, what's "significant" is subjective. But there's always the option of manually awarding achievements to individuals and/or teams after the fact for scientific breakthroughs made possible by automated methods– for example, achieving an RMSD below 2 angstroms relative to the native structure.
Elevator summary: diverse strategies make the search better.
the point of the game to me is to create a hybrid parallel search of the structure space. Cooperation and competition are thematic ways of characterizing different search strategies. Parallelism is the point. If too much correlation between different approaches then you get herding. herding isn't evil but it doesn't span the space very well - it makes search threads cluster.
The only thing Mao is much good for is the quote "Let a thousand flowers bloom."
Elevator pitch summary: the open source model of sharing might be helpful.
one might adopt a practice similar to open source licensing - like gnu - for example that permit anyone to use certain materials but that requires such uses to inherit the open source character too. Under that model anything I develop solo is subject to my discretion in the matter of publishing, but anything based on someone else's published work would need to be published too. A drawback might be tweak spam though.
