From: Andrew Odlyzko []


Thank you very much for the extensive correspondence about new electronic publishing proposals you have sent me. I am about to leave on a trip, so will respond very briefly.

I agree (as usual) with most of what you say. In particular, I am very much in favor of separating the distribution function from the peer review one. In general, separation of different functionalities is the best way to go when faced with rapid change. The separation of applications from transport was the key to the success of the Internet, for example. Such separation does have costs, but those are outweighed by the extra flexibility. I think it was an inspired move on the part of Paul Ginsparg to keep his preprint server separate from any overlay journals. That way a failure at a higher level (such as in an abortive attempt a while ago to start an overlay journal) did not cast a shadow on the archive itself.

In scholarly publishing, a separation of distribution from peer review is especially desirable since the costs and structures involved may be widely different. Paul has shown convincingly that preprint distribution can be handled very inexpensively. The other parts of the enterprise are still in doubt.

I also agree with you that some of the proposals are unlikely to succeed because they do not provide sufficient incentives for scholars to abandon existing journals. If excessive profits were the main problem, and one were trying to set up inexpensive journals to compete with them (as SPARC is trying to do, for example), why would this be more likely to succeed than some commercial publisher coming in and trying to undercut the profiteer? The function of capitalism, after all, is to reduce abnormal profits. Since such competitive moves have not been carried out successfully in the past suggests that some of the current proposals will have a hard time.

The one area where my views differ substantially from yours is (just as it was half a dozen years ago) peer review. I am much more skeptical of conventional forms of peer review, and much more willing to entertain ideas of novel procedures. That is one of the reasons I am a strong supporter of separating peer review from distribution, to allow maximal freedom in experimenting with the former.

I suspect that what will lead scholars to participate more fully in preprint distribution through widely available archives is evidence that this will really let them reach a much wider audience. I think such evidence is only now becoming available (I will send you a note about this later), and it might have a major effect.

Best regards,




Andrew Odlyzko

AT&T Labs - Research voice: 973-360-8410 fax: 973-360-8178