textfiles/magazines/electron.jnl

Newsgroups: bit.listserv.pacs-l
From: John Franks <john@math.nwu.edu>
Date: Thu, 21 Jan 1993 12:10:42 CDT
Sender: Public-Access Computer Systems Forum <PACS-L@UHUPVM1.BITNET>
Message-ID: <9301210219.AA17983@hopf.math.nwu.edu>
Subject: What is an electronic journal? 
Lines: 173

[This contains all 4 parts, concatenated.]

              What is an Electronic Journal?

        by John Franks
        Department of Mathematics
        Northwestern University
        Evanston, IL  60208-2730
        john@math.nwu.edu

        January 1993

There is considerable enthusiasm among scholars for creating purely
electronic journals which can be distributed via the internet.
However, in discussing this with colleagues and other interested
parties, I find that there are widely varying conceptions, many of them
conflicting, of what should constitute an electronic journal.  Most
scholars, when asked, are supportive of the idea of such a journal.
But, often they have only a vague sense of what it should mean --
sometimes little more than the hope that like electronic mail, articles
which interest them will magically appear on their desktop computer.

In this article I would like to explore some alternative possibilities
for an electronic research journal and comment on the strengths and
weaknesses of these alternatives.  My focus will be a narrow one --
restricted to a scholarly research periodical, marketed primarily to
research libraries.  In particular, I want only to address a
publication whose authors and editors are unpaid.  The addition of
royalties paid to author or editor could have a major effect on the
issues considered here.  Likewise, the electronic publication of a
book, even one with a narrow scholarly audience, might entail quite
different considerations.  Moreover, I want only to address the
possibilities for journals distributed via the internet, rather than
say, publication in CD-Rom or magnetic tape formats.

WHY DO WE EVEN NEED A JOURNAL?

The first question for an author in the internet arena is why publish,
in the traditional sense, at all?  Why not simply write articles and
make them freely available on the internet to anyone who is
interested?  After all, there is no direct monetary incentive for the
author.

In fact, journals are not an absolute necessity.  Making articles
freely available via the internet is one way to publish electronically
and some authors will choose this method.  I would call this form of
electronic publishing the ``vanity press model.'' Like all the models
of electronic publishing considered here it has some advantages and
some disadvantages and we will try to enumerate both.

The Vanity Press Model

First, let's look at the drawbacks, and answer the question why have a
journal at all.  There are at least three important functions which a
journal can provide beyond mere distribution of text.

The first of these is certification.  A journal has an editor who
chooses a referee or referees to read a submission and attest to its
scholarly worthiness.  The editor also maintains quality control in
non-content areas such as language and presentation (usually with the
aid of a copy editor).  Different journals have different scholarly
standards.  This process provides a peer review mechanism for
certifying the quality of scholarly work.  Academic institutions rely
on this process when judging the merits of an individual for promotion
or tenure.  While an author may have no direct monetary incentive
to publish in a journal, the indirect one can be significant.

The second important function is  archiving.  An author would like
to know that twenty or thirty years from now, perhaps after she has
retired, her work will still be available to other researchers.
Additionally, scholars in the field would like to have an authoritative
version of the author's text together with, at least, a definitive date
of its creation.  Traditionally, archiving is a function not provided
by the journal, but by libraries which purchase the journal and
maintain its preservation.

The third function which a journal offers is marketing.  If I
simply write an article and make it available from my personal or
departmental computer to  anyone on the internet, how will other
scholars know of its existence?  By contrast, if I publish in a
recognized journal, other scholars are much more likely to be aware of
my work.  This might be because the journal is in their library and
they glance at its contents on a regular basis, or because they consult
a second order table of contents such as  Current Contents.

These three functions, certification, archiving, and marketing
constitute the primary value added for the author who publishes in a
journal rather than using the ``vanity press'' model.  As we discuss
other models of electronic publishing we will want to see how well they
all perform these author support functions.

It is equally important to ask how well an electronic journal supports
subscribers.  This is the area where there are the greatest potential
advantages over traditional paper journals.  Indeed, if an electronic
journal is not substantially better or cheaper than a traditional
journal, its success will be limited.  And if it offers less
functionality than a traditional journal it is difficult to see how it
will be able to survive in the long run.  At an absolute minimum, it
must be possible for the subscriber to an electronic journal to print a
hard copy of an article of interest, which is of the same quality as a
photocopy of an article in a printed journal.  Simply viewing an
article on a computer screen will not be acceptable, nor will a printed
copy in a markup language.

Beyond this minimum, two of the most important criteria by which we
should judge different models of electronic publishing are their ease
of access and and the quality of their user interface.  These are the
areas where an electronic format can surpass the functionality of a
traditional journal. It might, for example, allow the scholar to browse
and search electronically on his desktop computer before printing a
copy, on his own printer, for detailed study.

Despite its seeming weakness in the author support functions, the
vanity press model does quite well in these scholar support areas.
Since the scholar downloads the electronic text to his personal
computer, he has complete freedom and flexibility in the choice of how
he views it, searches it, or prints it.

Another big plus for the vanity press model is speed.  An article can
be made available to the scholarly public, literally the instant it is
completed.  This is such an important asset that many authors already
use this model, in addition to publishing in a traditional journal.
This practice, of posting an article to a so-called ``preprint data
base'' can take different forms.  Typically, an author submits an
abstract of his work to a moderator who periodically distributes a
collection of abstracts, together with information on obtaining the
full text of articles, to an electronic mailing list of interested
scholars.  In all cases of which I am aware, anyone can join the
mailing list without charge and there is little or no editorial control
by the moderator (i.e. the certification function is not provided).
The full text may be kept centrally by the moderator or supplied by the
author either through anonymous ftp (see glossary) or, more commonly,
by electronic mail.

There are several variants of this process and there will surely be
evolutionary changes in the future.  Already some groups in physics are
making preprints available via gopher (see glossary).  This provides a
much better mechanism since it provides a number of features not
available through the e-mail process.  The most important of these
include:

       *  a simpler, easy-to-use user interface
       *  on-line browsing of abstracts or full text,
       *  keyword searching of abstracts or full text,
       *  immediate downloading of desired articles.

If only to meet the need of preprint distribution, the vanity press
model of electronic publishing will be with us for the foreseeable
future, and its use is likely to expand greatly.  There is sufficient
interest that the ease of use and functionality of this model will
likely continue to improve.

The absence of the marketing function in this model is not as big a
problem as it might initially seem.  Also its significance as a
drawback is diminishing and will continue to do so.  The abstract
distribution mailing lists and other preprint distribution channels,
provide an author with an increasingly effective way to provide
electronic visibility for his work.  It seems likely that some authors
who are indifferent to (or actively resent) the certification function
of journals, and are willing to forego the the archiving function, will
opt to publish some of their work only via the vanity press model.

It is worth noting, by the way, that the practice described above of
``double publishing,'' -- first electronically, using the vanity press
model and then traditionally through an established journal -- may
generate some controversy in the near future.  Publishers would like
the electronic availability of preprints to cease as soon as an article
appears.  Some publishers, in their copyright transfer agreement,
explicitly deny the author the right to make his work available on an
electronic data base [1].  I know of no instances of this restriction
being enforced, however, and current practice seems to be for
electronic versions of articles to be available indefinitely.

WHAT SHOULD A SUBSCRIBER TO AN ELECTRONIC
JOURNAL ACTUALLY GET?

Surprisingly many people who are strong proponents of creating an
electronic journal haven't thought a great deal about the answer to
this question.  Those who have seem to offer a wide array of very
divergent answers.  More than anything else it is the answer to this
question which distinguishes the different models of electronic
publishing.  As we characterize some of the different visions of what
should constitute an electronic journal, it is useful to keep the
varying answers to this question in mind.

The Data Base Model

The second model of electronic publishing (and the first which involves
what we could really call a journal) is  the ``data base model.'' In
this model all articles reside on a centralized data base maintained by
the publisher and what the subscriber gets is the right to access that
data base and probably use search software on the central computer to
locate and download articles of interest to him or her.  This is
roughly the way the commercial data services like Lexis/Nexis or Dialog
work.

In practice this might work as follows for the scholar wishing to make
use of the journal.  The subscription to the journal would be purchased
by the library of the scholar's institution.  The library  would
acquire a password allowing access to the journal data base, and would
be responsible for protecting it.  To use the journal the scholar would
typically schedule a time slot with the library and go the library at
the appointed time where a librarian who has access to the password
would login to the central data base.  When the scholar finds an
article of interest, it is probable (though not certain) that he would
be permitted to make a single hard copy of it for personal use.
Because of concern about unauthorized redistribution it is unlikely
that the publisher would allow an article to be downloaded in
electronic format.

The publisher might only charge the library a fixed annual fee for
subscription, but current practice suggests that some publishers are
likely to impose additional charges.  For example, cost may be a
function of the maximum number of simultaneous users.  Some publishers
will also likely want to charge extra for the use of their search
software and perhaps also for connect time.  This may not be entirely
negative.  If the price of a journal depends on the frequency of its
use then libraries would have to pay less for access to infrequently
used journals.  Moreover, publishers of several journals might well
offer package deals enabling libraries greater access to journal
material at less cost.

How well does this model meet our three author support needs of
certification, archiving and marketing?  Certification and marketing
would likely be quite comparable to a traditional paper journal, but
archiving would be dramatically different.  Since the library does not
maintain a copy of the text, it has no archival function in this
model.  There are significant trade offs here, which are difficult to
evaluate.  On the plus side, if a library starts subscribing to such a
journal they presumably have immediate access to all past issues
(though publishers may want to charge extra for this).  On the other
hand, if a library cancels its subscription to such a journal it loses
its access to all issues including those which appeared during the time
it was a subscriber.

More importantly, however, if a publisher should go out of business it
is not clear who, if anyone, would assume the archival responsibility.
This appears to be a major weakness in the archiving function for this
model.

This model is also quite weak in the scholar support criteria: ease of
use and quality of user interface.  It's functionality is roughly
comparable to that of a traditional paper journal and almost identical
to a journal which is traditionally marketed but published only on
CD-Rom.  This model realizes very few of the potential electronic
journal advantages, which have sparked the interest of scholars.  Most
noticeably the scholar must still physically go to the library and with
the aid of a librarian produce a copy for personal use (assuming this
is possible).  In some ways the functionality of this model is less
than that of a traditional paper journal.

The Software Model

One of the most miraculous technological achievements of this century
is the development of economically important goods which are
essentially infinitely reproducible at negligible cost.  The miracle of
the loaves and fishes pales by comparison to the ease with which anyone
with a personal computer can duplicate either software or electronic
documents, or someone with a digital tape recorder can duplicate an
artistic performance.  It must be one of the greatest ironies of our
age that this capability is less often viewed as a boon to mankind than
as an enormous liability to the publication of music, or software, or
even scholarly research.  By now we are all familiar with the downside
of this technological miracle: unauthorized reproduction of
intellectual property deprives its creator of the fruits of his labor.
If the creator has no incentive to create he will not do so.  (For a
fascinating contrarian view of this subject see [2]).

Given the similarities in the nature of this problem for electronic
publishing and software publishing, it is not surprising that one
vision of an electronic journal seeks to leverage the techniques used
in software publishing.

What the subscriber gets in the ``software  model'' is a piece of
software.  It should run on a networked personal computer or
workstation and probably be available in the several standard flavors
of such devices.  Other than the addition of this software this model
is quite similar to the data base model.  Here's how it might work.

A library or individual subscribes and receives in exchange a floppy
disk in the desired flavor.  When the software is run on an internet
connected computer it connects to the data base on the journal's
central computer.  The user can then perform searches, download etc.,
but all downloaded materials will be sent in a proprietary encrypted
form which the software can decrypt and display to the user.  There is
no need for a password, since someone who is not in possession of a
currently valid copy of the software cannot decrypt the text.  The
software might, or might not, allow the user to print a copy of a text
document for personal use (it would be technically difficult to allow
this while disallowing the creation of an electronic copy of the
document).  The software would have an expiration date which at each
use would be compared with the current date on the central server.  The
problem of unauthorized access to the journal is reduced to the problem
of preventing the unauthorized reproduction of the software (a
previously addressed if not totally solved problem).

Since this is really a higher tech version of the data base model it is
comparable to that model in meeting the certification, archiving and
marketing needs of the author.  In particular, it shares the major
archiving weakness noted above.   On the other hand in terms of
functionality for the journal reader it is potentially an improvement.
For example, it is possible that the scholar's library could negotiate
a site license for the software or perhaps a floating license (see
glossary).  In this way the software could run on the scholar's
personal computer and display text there, even though the only
subscription is through the library.

The Subnet Model

The next model of electronic publishing may be the most commonly used
commercially as of today, but it is not as yet used for scholarly
journals.  Instead it is currently used primarily for electronic
journalism.  Here is an example of how it works.

My university subscribes to a daily news service called ClariNet which
provides all UPI syndicated articles. It consists of an enormous amount
of material, including not only world, national and regional news (from
all regions), but also sports, and columns.  There are several hundred
newspaper length articles daily.  The university is licensed to make
this material freely available only to members of the university
community.

It is distributed using software which also simultaneously distributes
USENET (see glossary) articles.  This software, like all client/server
software (see glossary), splits the distribution function into two
parts.  All the text resides on a central server, but a server central
to my university -- the archiving function now resides with us.  This
central server provides the articles via a standard protocol to
``client'' programs running on a variety of platforms.  These include
networked personal computers and workstations, microcomputers in
publicly available labs, and larger computers designed to provide dial
up access to electronic mail and other network services for faculty and
students.  The protocol used is called the Network News Transfer
Protocol, (NNTP),  and the software for both servers and clients is
readily available without cost.  Surprisingly, it seems that, on
average, this software is of higher quality and better supported than
most commercial software.

The restriction that the ClariNet information be distributed only
locally is enforced by the server checking the IP address of the
computer running the client software.  The IP address is that strangely
formatted number, like 129.105.123.456, which is associated with a
networked computer and provides the basis for routing network traffic.
(IP stands for Internet Protocol).  This number has a hierarchical
structure.  For example, all IP addresses at my institution begin with
the two triples of digits 129.105.  This means the the news server
software can simply deny access to any client whose IP address does not
begin with this sequence.  In other words, the service is offered to
anyone on our university ``IP subnet.'' There are a variety of
different software ``clients'' for this server.  These are software
packages designed to run on a particular platform (e.g. Mac or IBM
PC).  They allow the user to browse the available documents on the
server and present selected articles to the user for reading,
downloading or printing.  It is the responsibility of the client
software, not the server, to deal with any display idiosyncrasies of
the the user's computer and to take advantage of any of its features.

The license granted my university permits us to archive these
documents, but, we do not.  Individuals have the right to make copies,
electronic or printed, for their personal use.  Protection against
unauthorized use is afforded by copyright.

The subnet for my university is divided into further subnets by the
additional digits in the IP address.  For example, appropriately
specifying the next three digits designates all those networked
computers in my academic department.  And, of course, specifying all
twelve digits (usually) uniquely determines a single computer.  This
makes it equally feasible for a publisher to provide access to everyone
who has access to a computer on my departmental subnet, or to everyone
who has access to an individual computer.

The particular client/server software and the NNTP protocol used for
news articles is not appropriate for a scholarly journal, but there are
several alternatives which are generally available without cost. In
particular, the National Science Foundation has funded the Clearing
House for Networked Information Discovery and Retrieval (CNIDR), which
will develop and support client/server software using the ISO standard
protocol for electronic text known as Z39.50 (see glossary).  There has
also been substantial development of software appropriate for this use
by Universities wanting to create campus wide information servers.
Most notable in this category is the gopher project.

There are many advantages to a scholarly journal distributed in a way
similar to this.  The utility to the scholar is much greater when he or
she has direct access to documents.  This model would rank quite high
in the scholar support criteria of ease of access and quality of user
interface.  If a journal is made available through a standard protocol,
the user should have substantial choice about the interface which he
uses to view or download the data.  I routinely use three different
clients to read the UPI news described above, the choice depending on
whether I am using my personal computer at home, or a workstation in my
office.  This kind of flexibility is not likely to be possible with the
software or data base models described above.

The mechanism used by gopher or NNTP servers for restricting access to
to certain subnets is much simpler than a password scheme and cheaper
to implement.  It is very much cheaper and simpler to maintain than a
model where the publisher must create and support all client software.
There are substantial economies for the publisher who uses standard
software supported by university computing organizations or
organizations like CNIDR.  It may seem surprising, but the quality of
the client/server software supporting standard protocols and available
without cost is much higher than what a publisher is likely to develop
and generally of at least as high quality as the average of mass market
commercial software.  The level of support for such software is
commensurately high.

In the subnet model the publishers flexibility in charging is somewhat
limited.  Subscriptions can be offered to universities, departments, or
individuals, but since the text is now archived by someone other than
the publisher, it is no longer possible to charge for searching or
connect time.

The Subsidized Model

The three electronic journal models described so far, the data base,
the software, and the subnet, differ primarily in the extent and method
of their efforts to *prevent* the contents of an electronic journal
from being read by those who have not paid for it.  In the first two of
these models the cost of these efforts will represent a substantial
fraction of the cost of publishing the journal.  It is not
inconceivable that the cost of restricting access to the journal will
represent a majority of production costs.  These costs, of course, will
be passed on to the subscriber, but there is another less tangible cost
for the subscriber which may be more significant.  Experience with the
publishing of software has shown that attempts to prevent unauthorized
use, make the use much harder for the authorized user.  This is true to
such an extent that many publishers have abandoned software copy
protection, in response to user demand, and rely instead only on the
protection afforded by copyright.  It is quite possible that the
inconvenience resulting from schemes to protect electronic journals
will be even more obtrusive than in software publishing.  In
particular, any scheme which requires the user to physically go to a
library and perhaps to enlist the aid of a librarian, or to login and
supply a password *each time* a journal is consulted is unlikely to
find favor among subscribers.

All this is especially ironic since the authors and editor derive no
benefit from the attempts to restrict access.  On the contrary, the the
best interests of the authors and editor are served by the widest
possible distribution (even to non-subscribers).

These considerations lead naturally to the consideration of alternative
methods of funding electronic journal production, which would permit
free distribution to any interested user.  Electronic journals
currently in existence are mostly of this type, though, as yet, only a
few could be considered true scholarly journals as opposed to newsletters.

A subsidized journal which provides a good example from the point of
view of technical production and distribution, is  EFFector
Online, the newsletter of the Electronic Frontier Foundation [3].
This publication, which appears approximately monthly, is available to
any interested party through at least four different electronic
protocols.  As issues appear they are posted to the USENET system.  In
addition they are made available for anonymous ftp, they are made
available via a gopher server and they are indexed and available to
WAIS clients (see glossary).  This shotgun approach to distribution
meets the subscriber needs of easy access and quality user interface
better than any other electronic publication of which I am aware.

Not all of these distribution channels would be appropriate for a
scholarly journal, but until such time as a standard emerges for
browsing and downloading electronic documents, it is a wise choice to
make documents available via a variety of mechanisms.  The cost of
duplicating distribution protocols is not high, and is far outweighed
by the benefits to users.

A second electronic publication worthy of mention in this category is
the Ulam Quarterly.  This is a refereed mathematics journal
provided primarily in an electronic format. Issues of the journal are
available by anonymous ftp and are ``offered without charge, courtesy
of Palm Beach Atlantic College Mathematics Department with support from
the University of Florida.[4]''  This provides an example of a journal
in this category where certification is handled in the traditional
manner.  At present this journal is electronically archived at two
sites and marketing is minimal.

Who might underwrite the costs of electronically publishing a journal
if there are no subscription revenues?  There are a number of
possibilities.  A professional society might sponsor such a journal and
pay for it out of members' dues.  Costs might be provided, at least in
part, by government grants.  A journal might be sponsored by a
University, or even a single academic department, as in the case of the
Ulam Quarterly.  An important factor is that with effectively
free distribution via the internet, and the fact that authors and
editors are not paid, the cost of producing an electronic journal can
be quite modest.

AMONG THESE MODELS WHICH WILL EMERGE AS THE DOMINANT ONE?

This is a difficult question to answer.  It is not clear what direction
commercial publishers will take.  At the moment they seem generally
conservative and uninterested in innovating.  But, in addition to
publishers, two other groups, scholars and librarians, will strongly
influence the development of electronic journals.

It is in the interest of scholars, both as producers and consumers of
journal articles, to have the widest possible distribution with the
fewest encumberances.  While a scholar's strongest motivation in
selecting a journal for his work will likely be to place it in the most
prestigious journal which will accept it, it seems likely that other
factors being equal he or she will opt to publish in a subsidized
journal where the article's exposure is likely to be greater.

While the interests of librarians may overlap with those of scholars,
they do not coincide.  A key issue is the state of libraries' readiness
and willingness to archive electronic journals.  On the one hand
librarians have little desire to become computer center managers.  On
the other hand they understand that if they only license access to
information that is owned by a publisher then their role as librarian
is diminished.  They become little more than a conduit to the publisher
for University funds.  For a library to own electronic materials it must
archive them.  This in turn requires  computing facilities and new
expertise.

It is important to understand that the attitudes of many
library staff members towards electronic publishing, or computing in
general, are influenced by their experience and expertise with the
software and computers they use for Online Public Access Catalogs
(OPACs).  These are typically commercial software systems like NOTIS,
which were designed (and often run on computers which were designed) in
an era before personal computers and workstations were widely used.

It is likely that among many librarians there is still an expectation
that systems like NOTIS and the computers on which they run can be
relevant to providing online access to archived electronic journals.
In my opinion, there is very little chance that this expectation can be
realized.  Librarians have already come to realize there traditional
OPAC platform cannot provide access to information in CD-Rom format and
that to provide this access it is necessary to acquire separate
computers and even separate local area networks.

Access to electronic journals, provided using modern protocols, will
likewise require new computing facilities and new expertise.  It is not
completely impossible to provide access using the old software and/or
hardware, but it will be much less cost effective to do so.  Moreover, the
quality of service will be so low that users will find it unacceptable
when compared with similar services provided on modern computers.  It
may be possible to teach an old dog new tricks, but it is very much
cheaper to buy a new dog.

Of course libraries will make the transition.  But it will likely take
time and in the short run libraries will be ill equipped to archive
electronic journals and provide their patrons with access to them.
This lack is even more dramatic for materials which are more
complicated than ASCII text.  For example, in mathematics and some
sciences, it is very common for journal articles to be created in the
TeX text formatting language.  The Ulam Quarterly provides its users
with articles in two formats  -- the TeX ``source'' which is what the
author prepares, and the Postscript output which is obtained from
processing that source, and is suitable for sending to Postscript
capable printers.  Almost no libraries today are prepared to deal
constructively with TeX source.  And relatively few are prepared to
handle Postscript on a substantial scale.

All this, may, for the short term, give libraries a reason to prefer
the data base or software models described above, because these models
will require the least new computer hardware and expertise. On the
other hand, there are strong countervailing forces. There is a desire, I
think, among librarians to continue their role as archivers.  They are
likely to be willing to acquire the new skills necessary for this
purpose.  This argues for an electronic journal model which permits
librarians this role.  Likewise, current intense budget pressures
should make the subsidized model popular among librarians.

This article is, of course, highly speculative.  The track record of
those who try to predict the course of developments in the use of
computers is rather poor.  Nevertheless, for those of us thinking about
the development of new electronic journals, choices have to be made
now.  It is my hope that is article can clarify the array of
possibilities which lie before us.


                          GLOSSARY
                          --------

anonymous ftp: (see ftp)

client/server software:
Software whose use involves two computers connected on a network -- a
``server'', on which some information physically resides, and a
``client'' which provides a user interface and requests information
from the server.  The advantage of this scheme is that the server needs
no information about the user's interface.  The client and server
communicate via a specially designed protocol.  Thus a single server
can communicate with users of many very different kinds of computers
without knowing anything about the screen or terminal characteristics
of those computers.  It is the responsibility of the client (running on
the user's computer) to know about the display characteristics of the
user's interface and to supply the information in a way compatible with
them.  See {\it gopher} for an example.

floating license:
A client/server mechanism for licensing software for use on computers
on a network.  If N licenses are purchased for use on a network with
many more than N computers, the first N client computers who want to
use it are permitted to do so.  Subsequent requests are denied until
fewer than N copies of the software are in use.  This has the advantage
of making it possible to use the software on a very large number of
computers (though not simultaneously) while purchasing a much smaller
number of licenses.

ftp:
File transfer protocol.  A standard protocol for transferring files
between computers on the internet.  Normally, it requires the user to
have an account on both computers.  However, it provides a mechanism
called {\it anonymous ftp} which allows the owner of a file on one
computer to make it freely available for copying by anyone on the
network.  Most ftp clients have no capability of viewing or browsing
the files they transfer.

gopher:
The most widely used electronic information delivery system (not
counting USENET which is really a conferencing system) is called
Gopher.  Initial development on gopher was done at the University of
Minnesota (whence its name), but important parts have been developed at
Illinois, Indiana, Rice, Stanford, Utah, and elsewhere.  Gopher is a
client/server based distributed information delivery system.  (see {\it
client/server}).  At present there are gopher clients for the Apple
Macintosh, IBM PC, IBM mainframe (CMS), NeXT, Dec VMS, Unix (curses),
and X-Windows (including Sun Openwindows).  All the client and server
software is freely available without cost.  A unique feature of this
software is the ability to make links from one server to another so it
appears to the user that the contents of the second server is a subset
of the hierarchy of the first.  Currently the NSF and NIH run gopher
servers as one means of online access to their public documents.
Several hundred colleges and universities use this software as the
basis of campus wide information servers.

NNTP:
Network News Transfer Protocol -- the protocol used for transferring
text on the USENET conferencing system.  It has facilities for
transmitting text documents between servers and between servers and
clients.  (see USENET)

USENET:
This is a large conferencing system with a distributed data base which
exists on literally thousands of ``servers'' world wide. It contains
``articles'' in various ``groups'' organized by subject.  There are
currently in excess of 2,500 groups.  Articles are kept only for a
short time (typically 2 weeks) and then discarded, thought some groups
are archived.  The collection of articles present on a server at any
one time can easily exceed a gigabyte (= 1,000 megabytes) of disk
space.  Groups can be ``moderated'', in which case articles are
submitted to an editor who accepts or rejects them for inclusion, or
``unmoderated'' in which case anyone can ``post'' an article to the
group.  This would be an appropriate mechanism to distribute a
newsletter, and is used to distribute the newsletter of the American
Physical Society.  There are a number of client software programs
available for most major platforms.

WAIS:
WAIS stands for Wide Area Information Service.  It consists of a full
text search program utilizing a client/server model.  WAIS is
complementary to Gopher.  It is useful when one wants to do keyword
searches through a very large number of documents and then browse those
documents with the best matches for the search terms.  It also has some
built in capability for auditing in order to charge for access.  It is
based on an older (1988) version of the ISO standard Z39.50 for full
text search and retrieval.

Z39.50:
An International Standards Organization Standard protocol for full text
search and retrieval.  Public domain servers and clients using an older
version of this protocol are currently available (see WAIS).  It is
expected that similar software supporting the latest version of the
standard will soon be available without cost from the Clearing House
for Networked Information Discovery and Retrieval (CNIDR) which is
receiving NSF support to develop it..

                    REFERENCES

[1] American Math. Soc.,  Transfer of Copyright Agreement

[2] Richard M. Stallman, The GNU Manifesto,
available by anonymous ftp from prep.ai.mit.edu in /pub/gnu/GNUinfo/GNU

[3] EFFector Online, a publication of the Electronic
Frontier Foundation, ISSN 1062-9424, available via gopher at gopher.eff.org

[4] Ulam Quarterly announcement on Amer. Math. Soc. gopher
at e-math.ams.org port 70

Copyright 1993 by John Franks.  Permission is granted to reproduce this
article for any purpose provided the source is cited and the author's
name and affiliation are not removed.