“Kaggle competitions measure objectively a very subjective metric: model performance. Model performance on a metric is not Subjective ” I am keen to disagree. How? You could say they only evaluate an aspect of data science. The selection of the metric might be subjective to what a client wants (which is a business decision, the company I work[dunnhumby] for has hosted 2 kaggle competitions.) — but once agreed it is fair among competitors. I mean — it is as objective as you get.
“If all data scientists are trained on Kaggle, the global economy will suffer.” I disagree with that statement — it is very subjective, but in any case Kaggle is a resource , is not a university ! Finishing a university does not make you a complete data scientist , you obviously need more resources, actual experience and so forth. But is it definitely helpful to do it .
“The metric that I propose for Startcrowd, market performance, is measured very subjectively.” That is personally my disagreement with the other platforms. Subjective…When I know I have limited time to spare to develop my skills, I would prefer to do it in a way that is comparable to what the others are doing . Even what you get from learning a course is not quantifiable. Hence you need to rely on reviews, connectivity with the industry , people’s opinion, university reputation and so forth . In that sense Kaggle is more quantifiable and clear.
“They look like ordinary forums and repositories, without any clear vision and purpose behind.” Again I disagree, I think kaggle has moved ways into democratizing AI as you can basically participate in competitions or do exploratory analysis without having resources yourself — relying only on kaggle’s machines.
I don’t think the yield is bad in higgs, because apart from the winning solutions, the papers, the innovations that took place , you had many people learning things too — something not visible in the winners’ tab.
I guess copying my exact words is funny… in a way — good luck with Startcrowd , I hope it works out for you!