I didn't see this in the FAQ - are teams required to publish their algorithms once they make an entry to a milestone?  I know they have to submit it to the organisers obviously, but will it then be published publicly?

Hi,

I have a kind of reversed question to competition organizers. May I publish my algorithms (give the description of algorithm and maybe source code - that every participant would use) on my website/blog?

I know that this would not help me to get any prize, however I think sharing the existing solutions as foundation for further research  may allow concentrate other participants to find better solution. In this competition I think the fusion of 2 worlds is needed - the "data hackers" (this is rather my case) and domain experts (physicians) - so also sharing the algorithms early might help to get better results. Teams that come from these different worlds might share ideas.  

I remember that during Netflix competition one person published the solution that allows many other people to get higher results and brought more "competition" into the competition (not on the top of the leaderboard but below).

Marcin

Wouldn't it be better to simply publish a description of the algorithm?  To publish your competitive advantage at the milestones simply puts everyone back on the same playing field.  I think it could also inadvertently take people down a rabbit hole by "locking" in a particular method, rather than the concept that allowed the advancement.

This question may have an obvious answer, but it isn't obvious to me as a late-comer to this competition. The papers of the milestone winner are incredibly detailed as required by Rules: D. 12. JUDGING PROCESS. My question is: doesn't this full disclosure practice benefit the ultimate Grand Prize winner who can review the methodolgy of the milestone winners and gain ideas and concepts that may not have otherwise been determined?

Yes.

That's why the Milepost winners get prizes. It compensates them for giving up any "proprietory intellectual capital" to all their competitors.

Or another way of looking at the situation is that there are three small contests and a big one, independent but all in the same domain of knowledge. Each team (nominally) starts each contest on an equal footing, since they all share a common understanding of the "state of the art" from the prior contest.

(In this case, the "small" contests are appreciably larger than most single-phase contests. The prizes are significant.)

Amulet:  In principle, you are right.  There may be some disadvantage (in the long run) to taking a milestone prize.  Anyone concerned about winning the milestone prize can specify a mediocre submission as their entry for the milestone prize to ensure they don't win.

Whatever disadvantage Market Makers incurred from publishing their first milestone paper wasn't great enough to stop them from winning the second milestone.  Similarly, Edward and Willem took second place in both milestones so far.  

Either the disadvantage from winning a milestone is smaller than it appears, or those teams are just very impressive (or, as I suspect, both).