Ike Nassi, an Executive Vice President and my former manager at SAP, writes in an email:
By accident, while reviewing a very old CACM paper “Programming Semantics for Multiprogrammed Computations” by Dennis and Van Horn from March 1966 (!) reprinted in the CACM 25th Anniversary issue (Volume 26, Issue 1 (Jan. 1983) Special 25th Anniversary Issue) I found a letter to the editor from Galler et al. dated 12-DEC-1968 (“Proprietary Packages: a Point of View”) in which the authors call for what is essentially open source:
“… In any case, we believe that the protected nature of such a proprietary package and the fact that it cannot be examined and evaluated impartially imply that descriptions and evaluations of it deserve to be related as sales literature and by nature are not appropriate for professional publications. When a scholarly report on such work is published in a professional publication, all supporting data (and programs) which are referenced in a professional publication should be freely available to all readers, as is usual with such a publication …”
There were follow up letters on the same theme. I wonder if this is the earliest reference to the concept.
Ike may well be right and this may be the earliest reference arguing against closed source, at least in research. Pointers for earlier references, if there are any, are welcome! And yes, open source licenses had still to be invented…