Refereeing is not only one of the most important activities of the working scientist; at the same time, it is getting more and more problematic.
The referee spends considerable time on a paper he accepts to review, but this time is never accounted for. You cannot put it on your cv. It hardly makes sense to mention it at your yearly performance review, since no one can check what you did. Refereeing must be anonymous.
This is quite different from the position of editor of a journal, or of some conference proceedings. Editorship is an academic distinction, and frequently mentioned on cvs.
One would expect, therefore, that editors treat prospective referees with the utmost deference: they ask for a favour that cannot even be acknowledged.
Instead, many publishers nowadays do lots of efforts to ease the life of editors, at the expense of referees. Typically, the referee nowadays receives a "no-reply" email message (or, in any case, an automatically generated message), in which he is invited to visit Easy Chair or some such web-based editorial system; where he must log in and do all kinds of complicated things before he can even see which paper they want him to look at. Of course, the referee has forgotten that he has been at this particular site before, has to ask for a new password, confirm that, etc., etc. But worst of it is the insulting impersonal, robotic type of communication that is being practiced here.
Perhaps, in the future, referees will be paid. Something like 5$ per reviewed page seems reasonable. I don't see any disadvantages of this. Surely no one will abuse this in order to write hundreds of reviews a year, because referees are chosen for their expertise in a particular topic, and have to maintain their reputation by being active researchers.
In the meantime, I have formulated my own policy. I will have nothing to do with any web-based editorial system; when I referee a paper, I will deal with the editor alone. I communicate with the editor via his/her personal e-mail address. I will not respond to any prompt from a machine. In the case of a "no reply" e-mail message, I will of course respect that and not reply.
Back to base.