She forwarded me to this website which discusses a particular type of merit-based pay: one that compares teachers to themselves only and provides compensation to all, regardless of how they perform in relation to another. Since one of the major criticisms is that this type of pay-for-performance would foment competitiveness between teachers and therefore negatively affect students, it seems to be a good idea. Additionally, if the evaluation criteria for "merit" includes test scores from students, there is plenty of potential of cheating and manipulation in order to boost salary.
What evaluation criteria are used is probably the most important point in this debate. Specific things like test scores or "using hand gestures" might be easy to check off a box, but it does not tell the quality of a teacher. But what if we were evaluated by peers using evidence from the classroom itself - would that be better?
Even though teachers overwhelming care about other aspects of teaching more than merit pay as evidenced by the Bill and Melinda Gates Foundation study, it is still a hot-button issue. As a math teacher in Philadelphia I get an annual bonus of $1500 due to my teaching in a "high need field." Additionally, every year one teacher from each school receives the Lindback Award for Distinguished Teachers, which comes with a check (I believe it is over $2000). So what if there were better criteria that didn't make me focus on test scores or compete with other teachers - would I want that? Perhaps.