Two factors cause the Netflix prize to be an exception to the rule:
1) The prize was so large as to throw usual rules of motivation out of the window. Anyone would attempt almost anything for $1 million.
2) The prize is a once-off, not an individual form of motivation. There's only one prize and to get it means prestige, beating others. The competition element counts for a lot. If your boss offers $10 to all workers who beat a certain quota, that's not the same thing.
I think that there's something to either one or both of your arguments, and that there are lots of further arguments:
3) The "candle experiment" applies only to medium-scale creative tasks
4) The "candle experiment" is one of a small class of lateral thinking problems for which monetary reward is negatively correlated with success
5) The "candle experiment" result only holds on tasks that last a short time
My question is, which hold, what's your evidence, and what experiments should be done next? I can think of a bunch, so if anybody wants to fund some research, let me know :)
I don't think a significant prize is what the talk is about. It's talking about being paid a salary, being rewarded with extra vacation vs. not being paid and freedom/mastery/purpose.
A significant reward past some threshold (which will differ by society) is bound to motivate.
> A significant reward past some threshold (which will differ by society) is bound to motivate.
This seems questionable, though likely, without empirical evidence. Especially given what data we do have.
I'm merely trying to think about where that threshold could be, or what the differentiating factors are in the netflix prize, that could cause it to be fundamentally different than Ariely's rewards. Was his "large reward" simply not large enough? Is it on "medium difficult" tasks that his result holds, but not on very difficult tasks combined with very large rewards?
Would you care to defend the claim that the netflix contest would have attracted anywhere near the amount of effort it did, without a million dollar prize at the end?
I understand that the Netflix "experiment" has no control, but that doesn't mean we should dismiss it as evidence. (And start talking X-Prize, Darpa grand challenge, etc. My assertion is that payments work sometimes, and clearly don't work other times, and I'm wondering why that is.)
Personally, I think the million dollar prize is just a way to justify it to your spouse / boss / whoever holds the carrots and sticks in your life. It's an appeasement tactic so that those who love tinkering with this stuff can do so guilt-free, because it "might" lead to That Ultimate Approved Outcome.
Same with the X-Prize, and the Darpa grand challenge. Those teams spent way more on their entries than the challenge money brought back. Throwing in a few million just legitimizes the fun that everybody's having; it's not actually the motivator. I mean seriously; you think Red doesn't build robots when Darpa doesn't have prizes?
> you think Red doesn't build robots when Darpa doesn't have prizes?
I assume Red is some dude who builds robots. I apologize for not knowing who you're referring to.
But anyway! I'll assume he's a Darpa grand challenge winner and some awesome robotics dude.
I don't think he would have gained as much as he has, technologically, without the competition that most certainly wanted to win the Darpa challenge.
These competitions don't change the dominant players in the game; what they do is move the margin, and force the dominant players to be that much better, because they want to remain dominant.
My theory is that $1 million -> prestige -> more entrants in DGC -> better tech than would have been created otherwise. Thoughts?
You're still missing what the experiments showed- money expands the time taken to do a finite task. Indeed, the base argument is that without a million dollar prize, much less effort would have been expended... yielding the same degree of success. The defense rests on the experimental data and the similar evidence of Wikipedia and kernel patches.
The X-Prize and Darpa challenges are qualitatively different in that the tasks require (expensive) material components, not just the undervalued resource of time. Nothing in the talk covers this scenario, and I'd suggest extra experiments.
> The defense rests on the experimental data and the similar evidence of Wikipedia and kernel patches.
This is where we disagree; I don't think the netflix prize would have been successful without a monetary reward.
I don't have a good argument at the moment, mainly because I don't think anybody has a good argument why Linux and Wikipedia worked.
I also think the X-Prize and Darpa challenges are more similar than you think they are, but need some time to think. Thanks for responding intelligently and giving me food for thought.
(Also, if you want to follow Pink's argument exactly, you need to claim that success would have come faster not merely as fast had there been no monetary reward.)
Comments
Counterexample: the netflix prize.
Two factors cause the Netflix prize to be an exception to the rule:
1) The prize was so large as to throw usual rules of motivation out of the window. Anyone would attempt almost anything for $1 million.
2) The prize is a once-off, not an individual form of motivation. There's only one prize and to get it means prestige, beating others. The competition element counts for a lot. If your boss offers $10 to all workers who beat a certain quota, that's not the same thing.
I think that there's something to either one or both of your arguments, and that there are lots of further arguments:
3) The "candle experiment" applies only to medium-scale creative tasks
4) The "candle experiment" is one of a small class of lateral thinking problems for which monetary reward is negatively correlated with success
5) The "candle experiment" result only holds on tasks that last a short time
My question is, which hold, what's your evidence, and what experiments should be done next? I can think of a bunch, so if anybody wants to fund some research, let me know :)
I don't think a significant prize is what the talk is about. It's talking about being paid a salary, being rewarded with extra vacation vs. not being paid and freedom/mastery/purpose.
A significant reward past some threshold (which will differ by society) is bound to motivate.
> A significant reward past some threshold (which will differ by society) is bound to motivate.
This seems questionable, though likely, without empirical evidence. Especially given what data we do have.
I'm merely trying to think about where that threshold could be, or what the differentiating factors are in the netflix prize, that could cause it to be fundamentally different than Ariely's rewards. Was his "large reward" simply not large enough? Is it on "medium difficult" tasks that his result holds, but not on very difficult tasks combined with very large rewards?
don't downvote without explaining why you disagree! It's a creative task for which a monetary prize was extremely effective.
A true counterexample would show that the monetary prize gave a faster or better result than no prize. Without a control, you haven't shown anything.
Would you care to defend the claim that the netflix contest would have attracted anywhere near the amount of effort it did, without a million dollar prize at the end?
I understand that the Netflix "experiment" has no control, but that doesn't mean we should dismiss it as evidence. (And start talking X-Prize, Darpa grand challenge, etc. My assertion is that payments work sometimes, and clearly don't work other times, and I'm wondering why that is.)
Personally, I think the million dollar prize is just a way to justify it to your spouse / boss / whoever holds the carrots and sticks in your life. It's an appeasement tactic so that those who love tinkering with this stuff can do so guilt-free, because it "might" lead to That Ultimate Approved Outcome.
Same with the X-Prize, and the Darpa grand challenge. Those teams spent way more on their entries than the challenge money brought back. Throwing in a few million just legitimizes the fun that everybody's having; it's not actually the motivator. I mean seriously; you think Red doesn't build robots when Darpa doesn't have prizes?
> you think Red doesn't build robots when Darpa doesn't have prizes?
I assume Red is some dude who builds robots. I apologize for not knowing who you're referring to.
But anyway! I'll assume he's a Darpa grand challenge winner and some awesome robotics dude.
I don't think he would have gained as much as he has, technologically, without the competition that most certainly wanted to win the Darpa challenge.
These competitions don't change the dominant players in the game; what they do is move the margin, and force the dominant players to be that much better, because they want to remain dominant.
My theory is that $1 million -> prestige -> more entrants in DGC -> better tech than would have been created otherwise. Thoughts?
http://www.frc.ri.cmu.edu/users/red/
Yes. Your theory is correct. But the key link is prestige. The DGC entrants aren't really expecting make a profit.
But I think the prestige is dependent on the million.
You're still missing what the experiments showed- money expands the time taken to do a finite task. Indeed, the base argument is that without a million dollar prize, much less effort would have been expended... yielding the same degree of success. The defense rests on the experimental data and the similar evidence of Wikipedia and kernel patches.
The X-Prize and Darpa challenges are qualitatively different in that the tasks require (expensive) material components, not just the undervalued resource of time. Nothing in the talk covers this scenario, and I'd suggest extra experiments.
> The defense rests on the experimental data and the similar evidence of Wikipedia and kernel patches.
This is where we disagree; I don't think the netflix prize would have been successful without a monetary reward.
I don't have a good argument at the moment, mainly because I don't think anybody has a good argument why Linux and Wikipedia worked.
I also think the X-Prize and Darpa challenges are more similar than you think they are, but need some time to think. Thanks for responding intelligently and giving me food for thought.
(Also, if you want to follow Pink's argument exactly, you need to claim that success would have come faster not merely as fast had there been no monetary reward.)
A true counterexample would show that the monetary prize gave a faster or better result than no prize. Without a control, you haven't shown anything.
Fair enough. A counterexample: "the paycheck." :)