Proposal for Moving the HCI Bibliography to ACM.ORG

Gary Perlman, Director, The HCI Bibliography Project

DRAFT FOR SIGCHI EC COMMENT 30 August 1996

Document URL: http://www.acm.org/~perlman/bibprop.htm.

Table of Contents


Executive Summary

Gary Perlman, founder and director of the HCI Bibliography (HCIBIB), proposes to move the HCIBIB from Ohio State University to ACM.ORG, much as the SIGGRAPH Bibliography is offered via the SIGGRAPH.ORG site. Currently, the project can be accessed at: http://www.hcibib.org/
Benefits
Costs
Contingencies

Background

The HCI Bibliography (HCIBIB) has been on the Internet since about 1990 and with over 12,000 bibliographic entries, most with abstracts, has been a primary source of bibliographic information for the HCI community. There are over 500 registered users on a mailing list, representing perhaps 10 times that many users.

SIGCHI has funded part of the development of the HCIBIB, first by funding a graduate student for a year to help organize the project, and later through the Chair's Discretionary Fund to provide a flatbed scanner for the project.


Motivation

The most immediate motivation is that I have left OSU, and the HCI Bibliography needs a new home. Another motivation is that the performance and features of the OSU CIS servers have been inadequate, and there have been discussions of moving the HCIBIB to ACM for years. The HCIBIB has been available primarily via ftp, and the OSU CIS ftp server often has sever performance problems. The CIS Web server has often had performance problems, too.

Given that ACM SIGCHI has funded the HCIBIB, and that it serves the HCI community (a community SIGCHI wishes to continue to serve), I think it makes sense for the HCIBIB to move to ACM.ORG over other potential sites (e.g., other universities, my current employer OCLC, its own server). Stu Zweben (past ACM President and OSU CIS chair), offered his view that the HCIBIB has, in his mind, always been more associated with SIGCHI than OSU.

Another ACM SIG, SIGGRAPH, offers a similar free service, at its web site: the ACM SIGGRAPH Online Bibliography Database. The SIGGRAPH Bibliography, like the SIGGRAPH web site, is on the SIGGRAPH server, not the ACM server. As an ACM SIGCHI member and officer (Vice Chair for Publications, 1995-97), I believe it is important that the ACM Internet presence be able to offer its members and technical communities what they need, so that they do not perceive a need to create their own servers.

In addition to simply providing the data in the HCIBIB, the Web server at ACM.ORG could provide more functionality, such as a search service (which the OSU CIS staff have been unwilling to allow), or alternative display formats. This would further enhance SIGCHI's role as THE place to go for HCI information.

A search service added to to database file access might be used as a way to promote ACM/SIGCHI membership by being available only to ACM/SIGCHI members (and perhaps also be freely available to users in .edu or similar domains) after being available for free on a trial basis. Limiting access to ACM/SIGCHI members would help justify a potentially larger load on the ACM server. Note that the database itself must be distributed free of charge because it is part of the agreement with copyright holders (a part that helped them permit the HCIBIB's royalty free use of their data).


Legal Issues

The HCI Bibliography Project has obtained written permission from all publishers for which it provides abstracts of publications. Moving the HCI Bibliography Project to ACM.ORG would allow ACM.ORG to provide this service (with the provision that there could be no charge for the service, as per the license agreements). Because of ACM's non-profit status, there should be no complaints from publishers, but in case they might complain, they would be warned in writing in advance of the move, and given the opportunity to pull out of the HCI Bibliography. Given the comprehensive nature of the HCI Bibliography, publishers should be willing to continue to allow royalty free distribution of abstracted entries because it serves as a form of free advertising for their publications. Previously, some publishers who have been reluctant to grant permissions, eventually granted permission after getting letter of support from their editorial boards (who are HCIBIB users).

The license with publishers contains the condition that there can be no charge for the HCIBIB. ACM may object to this condition, because presumably, it would mean that access to the files could not be restricted to members only. (The reason for the free-distribution condition is that it allowed the HCIBIB to operate without a budget; if we charged for the HCIBIB, the publishers would expect a payment, and the basic nature of the project would change, possibly to one that could not afford to support itself.)


Proposed Services

I propose that ACM SIGCHI distribute the HCIBIB database files, free of charge and without access restrictions, via two mechanisms:
  1. WWW: Linked into the ACM SIGCHI Website
  2. Listserv: Via the ACM listserv as HCIBIB@ACM.ORG or possibly as HCIBIB-CHI@ACM.ORG (to match the SIGCHI listserv conventions). The HCIBIB list would be moderated and allow:
Additionally, I propose that:
  1. ACM set up an hcibib login on turing.acm.org, with email address (hcibib@acm.org) being directed to the ACM listserv
The above would be offered freely to all Internet users. If search capabilities were added, the decision of access restrictions would be left to the SIGCHI Executive Committee, based in part on feasibility recommendations from the SIGCHI Information Director, and the ACM Network Information Manager.

Cost Estimates

The main costs are in space and in potential impact on the performance of the Web and listservers. We do not have good statistics on accesses to the HCIBIB because most accesses have been via anonymous ftp, with is not tracked by the OSU CIS servers. The existing listserv for the HCIBIB gets an average of less than one request per day for an average of 5-10 files. Given the low cost of storage and the currently minimal listserv load, the main risk is in load on the ACM Web server. Until there is a perceived problem, I recommend gathering data on number of accesses and load impact, so initially, the cost estimates are at most in the hundreds of dollars.

Size of the HCI Bibliography

Server Traffic for the HCI Bibliography