J. Smith to S. Harnad, 5/2/99


It is the refereed literature that "is and has always been an invaluable resource to researchers". The journal is only involved as a filter (by organising the refereeing) and as a carrier (delivering the information of the reader). Unfortunately it is the combining of these two roles, thus putting too much power in the same hands, that has led to the situation we have now where scholars (or their institutions) have to pay large amounts of money to read what their colleagues gave away earlier. In the past the only way to be widely distributed was to get your work into a journal - the net bypasses this problem. However the other major need is to have your work recognised and validated by your peers - currently you still need the journal to do this. The LANL pre-print server still has this problem - it can distribute but it can't validate (which is one reason I think it is not the way to go). As long as you have 'journals' you have exclusivity because their funding model is based on ownership of 'their' articles. What is needed is a system that cleanly separates the validating role from the distribution role. I have proposed such a model in the Deconstructed Journal:


The LANL pre-print server could operate within this system as my model does not dictate who will run the servers.

However you are proposing the model LANL embodies become the model for other disciplines. I have the following reservations about this:

- The LANL model leaves it dependent on journals to validate its content. The very journals you condemn for limiting access to knowledge.

- The LANL model is centralised. I guess this is a leftover from the era of the mainframe. Any centralised model is vunerable to control - who shall say you can deposit your work here? Just like the journal. The net is distributed and any publishing model based on it should take advantage of this.

sh> The gist of my recommendation to the Scholar's Forum is that that is the principle it
sh> should implement, not any untested model for the future of journals…

Because it is not not free of the 'journal' - the model it embodies is not self-sufficient.

If everything is safe in a "safely distributed, redundant and mirrored storage architecture" why do we need a "LANL-style Archive"?

All we need are the 'overlays' (I don't know if this term was originally used by you, or Ginsparg,


or someone else) but not an'overlay' on a centralised archive but an overlay on the whole of the net. Which is what my proposed Subject Focal Points are intended to be.

sh> But if you needlessly overstructure the simple target….

Why have 'half a loaf' when you could have a whole one?

I propose a publishing model in the Deconstructed Journal which replaces all the roles of the current model. The implicit model within LANL service only replaces the distribution half of the main roles.


When I think about the operation of the LANL pre-print server over time it seems it must become like the Deconstructed Journal since the 'journals' that currently do the validating will not have a distribution role at all so what will they be selling? All they will have is their validation role but how can they charge the reader for it since it is the article that is validated and once it is validated it is validated for all readers (since it is freely available from the server). They will therefore have to sell their services to the author or author's institution because it is these that benefit most directly from their validation. At this point they will cease to be 'journals' and become 'independent evaluators' - a concept central to the DJ model - as you might expect :-)


John Smith,

University of Kent at Canterbury.