Fog Creek asserts that the study shows a negative relationship between commission and performance. This relationship was found - for very large rewards. Small rewards increase performance when compared to no reward.
Here is the original paper: http://m.pss.sagepub.com/content/15/11/787.short.
In a follow-up study, he showed that effort increases with reward almost as predicted, but error rate increases exponentially: http://onlinelibrary.wiley.com/doi/10.1111/j.1467-937X.2009..... The effect is also variable depending on whether it is a primarily intellectual or mechanical task.
I'm not making a ruling on commissions here, just saying the Ariely quote corroborates Fog Creek's null hypothesis.