I can relate to the "bittersweet Kaggle memories" phenomenon.
I participated in their first-ever competition, which I thought I would have a good shot at because Kaggle was brand-new (thus not much competition), and because it was in my wheelhouse, a biological application of ML. And at that time, c. 2010, ML was not all that well-known.
I did OK (placed somewhere in the top-middle IIRC) but it was quite humbling. Now it's not really worth doing except for fun or to be recruited by someone because the competition is so fierce and there are people with a lot of time to devote to it. The difference between 1st and 25th place is often measured in the 3rd decimal place of performance, making success kind of random. But the postmortems by winners are always good to see some real-world best-practices and different workflows.
As for the business model, I'm pretty ambivalent about it. My wife is a graphic designer, and in that field, "compete to see who has the best design" is a somewhat common thing. But it's scummy and designers hate it because it's a way to basically get free work out of lots of people and it erodes salaries in the industry.
Work should probably not be gamified, especially when the gamification takes the form of "you only get paid if you win". And "hey, you might get recruited if you do well without winning" is not a lot of consolation. It's pretty exploitative for anyone not A) doing it purely for fun/learning or B) willing and able to assume the risk of making their money from competitions. (Just to be clear, I've never done Kaggle except for fun, but I know others do it for serious career purposes or money, as those are obviously express intentions of the site)
Good point. I work in the "relatively safe" area of academic research, but the point still holds.
Even more broadly, your thought has made me wax a little philosophical about capitalism: we believe that 1) everyone should work, 2) I only want winners to work with/for me, and 3) not everyone can be a winner. I guess you can't have all three, but we sure try.
If you put it in that light, maybe Kaggle isn't so bad. But OTOH, we do make the distinction between employees and entrepreneurs for a reason.
I participated in their first-ever competition, which I thought I would have a good shot at because Kaggle was brand-new (thus not much competition), and because it was in my wheelhouse, a biological application of ML. And at that time, c. 2010, ML was not all that well-known.
I did OK (placed somewhere in the top-middle IIRC) but it was quite humbling. Now it's not really worth doing except for fun or to be recruited by someone because the competition is so fierce and there are people with a lot of time to devote to it. The difference between 1st and 25th place is often measured in the 3rd decimal place of performance, making success kind of random. But the postmortems by winners are always good to see some real-world best-practices and different workflows.
As for the business model, I'm pretty ambivalent about it. My wife is a graphic designer, and in that field, "compete to see who has the best design" is a somewhat common thing. But it's scummy and designers hate it because it's a way to basically get free work out of lots of people and it erodes salaries in the industry.
Work should probably not be gamified, especially when the gamification takes the form of "you only get paid if you win". And "hey, you might get recruited if you do well without winning" is not a lot of consolation. It's pretty exploitative for anyone not A) doing it purely for fun/learning or B) willing and able to assume the risk of making their money from competitions. (Just to be clear, I've never done Kaggle except for fun, but I know others do it for serious career purposes or money, as those are obviously express intentions of the site)