November 28, 2007
Reputations On (the) Line
practicalethics: Honest Opinions or Bullying? is another of my practical ethics posts. This time about on-line rating systems for teachers. Teacher's unions hate them, but at least in Germany they are protected as free speech. I argue that they are relevant as a way of balancing the formal power of teachers with informal reputation power, and that persistent pseudonyms might make everybody happy.
However, there is an issue I did not have the space to get into: local versus global reputations.
We have always had reputations, and they are essential for the function of society. Without reputations trust would not be possible, and economy would be highly inefficient. In the past reputations were always local to groups, subject to constant reappraisal. Today we have emerging global reputations outside our own group. These global reputations suffer both from lack of context, lack of reappraisal and a high degree of noisiness. The teacher evaluation may be due to temporary affect, and be interpreted very differently by parents, administrators, pupils and teachers. Evaluations may remain long after the situation has changed. And it is easy to lie.
The combination of Internet, free expression and effective search systems means that online, global reputations are inevitable. The problem is that we tend to treat them as local reputations. They are not reliable, consistent, frequently updated or reflecting a particular group. Hence we need to learn to look for the reputation of reputation sources: is someone saying something about me anonymously on an open forum, or doing it under his own name?
We ought to collectively (out of enlightened self interest) look into the possibility of not just making distributed reputation systems but ways of estimating reputation source reliabilities and aggregate scattered reputation pieces appropriately. Much of this is likely already being done in collaborative evaluation research, but to make any kind of software work we also need the social interest and understanding why it should be used. Just as we have gradually learned not to trust something because it is printed or on-line, we need to build a memetic immune system against low-quality reputation fragments. That may prove very tricky, since reputations are just the kind of "hot" social and emotional fact that we tend to fall for.
Posted by Anders3 at November 28, 2007 11:41 PM