/ ' Letters

Editor The article "IJCAI Policy on Multiple Publication of Papers," by Alan Bundy in the Spring 1989 issue, misses one of the most important functions of IJCAI (or any other conference for that matter): The opportunity to present research to a wide audience of one's peers in person. No journal article can wholly replace the valuable give-and-take discussions that occur after a paper session. The proposed policy would deny this opportunity to researchers who have successfully published their results in a journal. Should AI researchers delay journal submissions until portions of their papers have been accepted by IJCAI or other major AI conferences? Given the vicissitudes of peer review in a methodologicallydivided field, the proposed policy seems like an excellent way to prevent dissemination of recent results Also, as much as the AI community might regard the IJCAI proceedings as an archival publication, the same thing can hardly be said of university tenure and promotion review committees, which in general prefer refereed journal articles over conference papers of whatever kind.