In an ideal world, I would agree with the recent letters to Eos by M. Beck (1 July), C.J. Robinove (29 July), and R. E. Criss, and A. M. Hofmeister (29 July) calling for all reviews by referees and associate editors (AEs) to be signed. But in an ideal world, they as well as the rest of us, would invariably be fined for petty infractions such as exceeding the speed limit by even 1 km/hr, so it might not be much fun to live there. Human nature being what it is, offering anonymity to referees is probably necessary to ensure the smooth operation of journals.
The above correspondents did not distinguish between the problems created by anonymous referees and those created by anonymous AEs, but I believe this is an important distinction. The AGU publication guidelines state that the editor is the sole judge of what is accepted and what is rejected, while all other opinions are advisory. This is fine in theory but may not always reflect reality. When there is, say, only one editor assisted by a team of 20 to 30 AEs, as a practical matter, the AEs are the real decision-makers and the editor cannot possibly find the time to study each manuscript carefully.