Visions for SIGCHI Electronic Publishing

Jakob Nielsen, SIGCHI Vice Chair for Publications
Steven Pemberton, SIGCHI Bulletin Editor
John "Scooter" Morris, SIGCHI Information Chair

December 1993

When discussing possible SIGCHI ventures into electronic publishing, it is important to keep in mind that the content providers have provided SIGCHI with their original intellectual property in order to have it widely disseminated. Thus, we need to consider electronic publication as a way of providing access to the materials for a larger audience and of giving this audience better ways of finding the HCI materials they need. In other words, we cannot consider electronic publishing as a profit center to generate additional revenue by restricting access to the information or by maximizing access charges. On the other hand, we realistically have to consider that there is in fact some cost associated with the conversion of heterogeneous source materials into a high-quality integrated hypertext, just as it will cost money to run and maintain the server and its network connections. In the long term, we will have to resolve the conflict between these two opposing views and find ways of generating sufficient revenues to cover the cost of supplying SIGCHI information electronically.

Possible cost recovery mechanisms include the following:

1. No cost recovery! The cost of electronic publishing is paid through conference profits and membership dues, viewing electronic publishing as a kind of publicity medium that can attract sufficient revenue of these two kinds to cover its costs without any additional income. The advantage of this approach is obviously that accounting is simplified tremendously and that our information is given the widest possible dissemination since everybody will be able to access it.
2. Indirect cost recovery through membership dues. We could restrict access to SIGCHI electronic publications to members in good standing and then cover the costs of the service through increased membership dues. The advantage of this approach is again its simplicity (no additional accounting required, though access controls and;/or passwords of some kind will have to be placed on the electronic publication system).
3. Separate subscription fees for SIGCHI electronic publications. People and/or companies pay a certain amount of money each year and are granted unlimited access to SIGCHI electronic publications in return. The advantage of this approach is the potential for holding down the basic membership dues while allowing non-members to access the materials (if they or their organization pay the required subscription charges).
4. Usage fees. The system could charge users (or their organizations) a fee each time they use it. Charges could be per megabyte downloaded, per hour of connect time, per search or document, or some combination of these. The main disadvantage of this approach is that people are known to hate the feeling of "the clock ticking" as they use an online service. Most people prefer a flat rate, even if they in fact have to pay the same. Also, accounting and billing is potentially more complicated and expensive for usage fees, though the Internet doubtlessly will develop mechanisms similar to the U.S. telephone companies' 900-numbers for simplified usage fee collection.

There are different technological solutions to electronic publishing, including the following:

1. CD-ROM publishing: The advantage is that access to a CD-ROM is free once it is acquired and that it is easy to charge a price for the sale of the CD since it is a physical object. Also, many libraries have established means of storing and accessing CD-ROMs. For the user, an advantage of CD-ROMs is that the transfer rate of information between the CD drive and the computer is higher than the transmission rate for most current Internet connections, making it easier to scan large amounts of information and view digitized videos. A major disadvantage for users is that CD-ROMs are not integrated across disks, meaning that users will have to manually search their growing piles of CDs to mount the appropriate one, and that hypertext links across CDs are difficult to support. Typically, a single CD contains less than one hour of video, making it impossible to integrate the proceedings and conference video of multiple CHI conferences on a single CD-ROM. Also, CD-ROMs are isolated solutions that do not naturally lend themselves to collaborative usage such as annotations or relevance votes (that are otherwise potential added-value features of electronic publishing as compared with paper publishing). A final disadvantage of CD-ROMs is the profusion of standards, making it necessary to produce multiple versions of each CD if all major platforms are to be supported.
2. Internet publishing: Internet publishing overcomes all the disadvantages of CD-ROMs, since it is possible to integrate virtually infinite amounts of information by simply buying additional harddisks. Not only is it possible to support hypertext links across large amounts of material, it is also possible to support dynamic information such as user-supplied annotations or bidirectional hypertext links (e.g., an old paper would "know" what newer papers cited it). If we use the World Wide Web as our Internet delivery mechanism, standards problems go away since a single back-end format can be viewed on front-end viewers optimized for each platform (there are WWW viewers for at least the following platforms: IBM VM mainframes, plain-text Unix [VT100 etc.], X-windows, Macintosh, MS Windows, Amiga). Currently, Mosaic seems to be the most popular and powerful WWW viewer. In addition to Internet publishing mechanisms include FTP, Gopher, and email servers, but these alternative mechanisms provide stone-age user interfaces (OK, Gopher is bronze age )) and are not recommended for our initial experiments, though some publications (especially text-only files) may be made available through these mechanisms as a supplement to the WWW hypertext.

Given the trade-offs, we think Internet publishing is the most promising approach for our first experiments.

Given that we want Internet delivery, we will need a server. There are four possibilities:

1. We piggyback onto some existing server run by an organization that is willing to donate the necessary number of megabytes. The obvious two advantages are that we will not need to maintain the server and that there are no costs. The disadvantages are that we will be at the mercy of the manager of the host organization and that the host can pull the plug at any time. Also, we will not gain the experience with the logistical aspects of running a server that we will need for our long-term plans.
2. We establish our own server, sigchi.org. This would be fairly easy to do and would involve buying a workstation (or even a muscular PC) with a few hundred megabytes of harddisk and installing it in somebody's basement with an ISDN connection (possibly upgraded to T1 if the server proves successful). The costs of this model are not particularly large, but having to maintain the server will be a major pain. There will be a need for backups and a need to negotiate with transport providers (currently, ISDN provisioning is not well supported by most telephone companies).
3. We use acm.org since it is an established server maintained by ACM staff. The advantage of this approach is name recognition and the ability to be one of the driving forces in upgrading and modernizing ACM computer services. The disadvantage is that acm.org is currently implemented using obsolete hardware.
4. We contract with a commercial service provider like Mead Data Central or America Online. The advantage of this approach is that these service providers already have extensive experience in running information services that pay for themselves, so we gain a major advantage with respect to cost recovery. The disadvantage is that most of these services are not easily accessible via the Internet (most require modems for access) and that they would be comparatively expensive for our members compared with the other three solutions. Also, since SIGCHI is supposed to be a resource for the entire computer industry, it would be doubtful for us to single out any individual service (and putting our content on all services would be a logistical nightmare).

We recommend using acm.org for our initial experiments.

IMPLEMENTATION PLANS

The plan for implementing SIGCHI electronic publishing is divided into three parts, covering the short term (this year), the medium term (the next two years), and the long term. We know concretely what we want to do this year, and we have a fairly good grasp of what the technology will support over the next two years. For the long term future, our plans should only be seen as preliminary and the concrete plans will depend of developments in the hypertext field and the Internet.

Short Term Plan, 1994:

Convert those issues of the SIGCHI Bulletin that exist in FrameMaker format into HTML hypertext for delivery over the World Wide Web, served through acm.org. Initially, this service will be free to all Internet users. Desirable advanced features of this project would be automated searching through WAIS or other means as well as the establishment of tools to convert future issues of the Bulletin from Frame format to HTML format for integration into the online hypertext. We expect to need no more than 10-20 megabytes of storage space on the HTTP server for this limited project.

Conversion will most likely take place as a student project at CWI, supervised by Steven Pemberton. It is proposed that the costs be paid by SIGCHI out of its development funds.

Action items: Steven Pemberton: Find out what financial support would be needed to get a student to do this (and start thinking about who might want to do it). Scooter Morris: Find out when acm.org can be expected to have an HTTP server and what (if anything) SIGCHI will be charged for putting material on the server.

Medium Term Plan, 1995-1996:

The proceedings of CHI'95 and CHI'96 are published electronically on the World Wide Web in HTML form, including some non-traditional media types. Additionally, the future issues of the SIGCHI Bulletin are placed online as they are published, and we should experiment with converting at least one old volume of the CHI proceedings into online format. Some form of search mechanism will definitely have to be implemented for this medium-scale collection, most likely using WAIS, Latent Semantic Indexing, AppleSearch, or some similar information retrieval-based paradigm. The storage space for the medium-term project will likely be less than a gigabyte.

Major parts of the conversion effort may be outsourced to staff recruited through the Moscow chapter of SIGCHI, taking advantage of the temporary disparity between the quality of the available personnel and their cost when local salaries are converted into western currency.

If needed, we might experiment with access to our online materials for people who have email but no Internet connection. Depending on technical developments, we might also experiment with non-WWW Internet access.

Most likely, this material will be made available for free to all Internet users, but it may be necessary to restrict access to SIGCHI members in good standing, depending on the cost structure of running the acm.org server. For the medium term period, we do not envision the use of explicit access charges, so the most we can do is to recover some of the costs through membership dues and restrict access to the members. All of the conversion costs as well as most of the costs of running the server should probably be covered through SIGCHI development funds given that the work is scheduled to take place during a transitory period where Internet servers are comparatively expensive and without well-established charging mechanisms.

Long Term Plan, 1997 and beyond:

All proceedings of major SIGCHI conferences are available online in an integrated manner (e.g., with hypertext links for cross references), including not just text and traditional printed materials but also dynamic media like animation and video. The SIGCHI Bulletin, conference announcements, and other materials are also available. Older information is converted to as large extent as possible, hopefully including all CHI conference proceedings since 1983. Advanced search tools are provided, possibly using agents, information filtering approaches, readwear, or other paradigms that go beyond standard information retrieval. Also, features like group annotations and ephemeral interest groups may be supported in addition to the traditional publication of original source materials supplied by the initial content providers. The required storage space for the long term project will likely be ten gigabytes for the converted materials plus an additional three or four gigabytes for each year of new materials, growing to maybe five or ten gigabytes per year as more content providers take advantage of the new multi-media possibilities.

Mechanisms need to be put in place for the conversion of SIGCHI publications to whatever standardized online format is chosen. In the long term, it will not be possible to rely on having this conversion performed by cheap labor, so as much as possible of the conversion will need to be automated and we will need established mechanisms for paying for whatever manual work remains.

SIGCHI electronic publishing will need to be a financially sound proposition, so we will need to determine funding mechanisms that do not depend on development funds. During the medium term period, viable financial models have to be developed and experiments carried out to determine how and what users are willing to pay. By 1997, there will almost certainly be mechanisms in place on the Internet for various ways of charging users (by the hour, by the megabyte, or as a subscription fee).

During the short-term and medium-term periods, we envision to have all publications published in a paper format in parallel with their electronic versions. The only exception is the multimedia companion materials to the conference proceedings, which might be published on videotape in parallel with the online version, but which most likely will only be published electronically. After 1997 it may be possible to eliminate paper copies of some publications, but most likely, SIGCHI will continue publishing paper versions of most publications until after year 2000. A target date for the elimination of paper publishing might be 2010.