$50k competition to spot icebergs off the coast of Newfoundland
cbc.ca
cbc.ca
Cool crowdsourced challenge!
Edit: Removed US Coast Guard reference, see peeters link below.
http://laws-lois.justice.gc.ca/eng/regulations/SOR-2005-134/...
Automatic Identification Systems (AISs) 65
(2) Every ship, other than a fishing vessel, of 300 tons or more that is engaged on an international voyage shall be fitted with an AIS.
(3) Every ship, other than a fishing vessel, of 500 tons or more that is not engaged on an international voyage shall be fitted with an AIS, but if it was constructed before July 1, 2002 it need not be so fitted until July 1, 2008.
I've emphasized the part about fishing vessels because that is particularly relevant here. The waters around Newfoundland are huge fishing grounds.
AIS has been slowly rolling out for almost 13 years now, and is pervasive enough that you can be fairly confident most, of not all, large ships are announcing themselves with it. I know several blue water cruisers even with sailboats in the 30-50ft range that have it installed to prevent “death by cargo ship” in the middle of the night.
This isn't getting people to work for free. It's simply getting people interested in a problem.
I'll speak for myself, I do it because I love trying to solve interesting problems and I love data science.
thus far every competition I joined I learned something new and expanded my ds code base. I also really love the fact people post kernels and discussions I can learn from as well.
On the other hand the problem sets are 'real' and it's excellent machine learning experience - importanly because you have the competition element, you get an idea about how good your solution actually is.
however, this was my first real kaggle competition, and I only spent a few weeks on it, (I'd say about 3 total). it was also one of the more competitive competitions, with a million dollars on the line.
Its exponentially difficult to move up the ranks, but I do think its realistic to win one (though hard - as it should be). a lot of time, effort, and skills required but not impossibly difficult.
I think kaggle would do everyone a favor if they gave out awards to more people, instead of the winner take all mentality. at the very least, everyone in the top 10 should win money, not just the top 3, with 1 getting the lions share.