You've been cruising the Net and discovered the perfect article to
add to the lecture notes you're handing out tomorrow. Unfortunately,
there's only one person who knows how to set up the job
specifications for your local network printshop, and he's been home
asleep for hours. If only that system were as easy to use as the laser
printer sitting outside your door!
Project CUPID--which stands for the Consortium for University
Printing and Information Distribution--is pushing the limits of what's
technologically possible in the network publishing business. This
informal collection of colleges and universities created four years ago
by the Xerox University Advisory Panel now operates as a project
under the auspices of the Coalition for Networked Information, with
funding from Kodak, Sun Microsystems and Xerox. Through CUPID,
the goal of creating a system for distributed printing over the
Internet of finished, high-quality, production documents has been
The Current Reality
The new printing systems now available can, in a single unit,
perform operations that previously required multiple machines--or
an entire printshop--and many steps of manual effort. Documents
emerge in finished form, suitable for distribution to customers--
bound, stitched, or stapled, with covers and select paper stock.
Operational costs and output quality make these systems competitive
with--and often superior to--offset presses for a wide range of
These new printing systems also come equipped for connecting
to a variety of data networks, including the Internet. A request for
printed material--and/or the original copy of the material itself--can
thus be a world removed from the location of the printer. This
mutual leveraging of printing technology and network technology
holds the promise of new paradigms for information dissemination,
and can be expected to have an especially large impact on any
business involved in document distribution.
Until now, the benefits of high-end network publishing
remained largely unrealized because of a lack of standards and
common procedures--in short, the lack of a national infrastructure
for network publishing. An individual or business interested in
transmitting a document for remote printing at even a single location
must almost always directly contact some person at that site to
determine everything from the network address of the printer to the
terminology used in the electronic work order. Today's custom
publishing and print-on-demand activities almost always entail
elaborate advance arrangements, significantly undercutting the
ability of network publishing to exploit spontaneous decisions and
targets of opportunity.
The headaches encountered by potential network publishers
today are evident in this representative sample of questions that
might be asked of a candidate printshop:
o What capabilities does your printer have? Can it do side-
stapling or only single-corner? What paper colors are available?
What cover stock?
o What software must I use to send the job to the printer? Since
the software needed to control each type of printer is different,
where and how do I acquire the program needed for your particular
printshop? Once I acquire the software, how long will it take my staff
to learn to use it? How will updates be distributed?
o What type of network protocols does your printshop require?
Will our existing Novell or AppleTalk network work, or is a gateway
o What format must the document be in? If PostScript, what
level? If TIFF, which version? Is there a limit on the size of the file
your printing system can handle?
o How must I specify the various high-end options that make
this process so desirable--cover stock, paper color, finishing options?
In requesting front/back printing, for example, is duplex or two-
sided the appropriate term? For an important variant of this printing
style, is the keyword tumble or end-to-end?
It is useful to contrast this situation with that of the desktop
publishing industry that has grown up over the past 10 years or so.
We routinely route the printed output from our personal-computer
applications to laser printers around our companies or colleges,
without giving a second thought to the indispensable standards of
file format, printer options and commands, and network structure. A
similar--but more comprehensive--set of standards must exist in
order for high-end network publishing to reach its potential.
Apart from concerns about standards, the widespread success
of network publishing also requires agreement on how to address a
variety of procedural issues involving user authentication, copyright,
and billing and accounting.
Some of these areas involve problems that are truly difficult to
solve, but recent developments in commercial uses of the Internet
provide reason to hope that solutions of widespread applicability will
Bridging the Gap
Two years ago, a CUPID subcommittee issued a report describing an
architecture for a network publishing infrastructure. This
architecture specifies a standard for interactions between publishers
and printshops communicating over the Internet, where a publisher
is any individual or business wishing to have a document printed,
and a printshop is a site with one or more printers connected to a
The CUPID architecture includes clients and servers for both
publishers and printshops, each with certain capabilities and
responsibilities in the end-to-end procedure. The CUPID protocol
allows publishers to request services from printshops, independent
of the type of hardware, software, or network connections operated
by each party.
In the CUPID architecture, a typical publisher uses the CUPID
Publisher's Client graphical user interface to select a document for
printing and designate one or more CUPID printshops where the
printing will take place.
For each target printshop, the publisher may specify separately
the number of copies, paper and cover stock, and finishing options,
along with any special instructions that may be needed. The CUPID
printshops are selected from a list automatically maintained by the
CUPID system, and the characteristics and capabilities of each
printshop are advertised by that printshop's CUPID Printshop Server.
The publisher thus is only presented with options of which the
printshop is capable.
Once submitted, the CUPID system adds each print request to a
queue maintained separately for each printshop. Production
managers in the printshops subsequently use the CUPID Printshop
Client graphical user interface to view the contents of their queues,
selecting jobs for processing according to local scheduling
requirements or other considerations. CUPID maintains status
information at all points in the process and makes this information
available to the publisher. In addition, the CUPID architecture defines
and describes the role of agents, who are individuals authorized by
the publisher to certify that certain parts of the printing process
have been satisfactorily completed.
The key to CUPID's approach to the problem of network
publishing is the insertion of an extra level of software intelligence
between the publisher and printshop, mediating the end-to-end
interaction for the benefit of both parties. It is this software, for
example, that translates a standard way of expressing a print option
into the appropriate specifications for the particular printer being
used. This extra layer of processing also facilitates solutions to such
problems as authentication, accounting, security, and encryption.
Fine-Tuning the System
The use of CUPID extends also to other software systems. CUPID has
been integrated within the Cornell Digital Library, to allow patrons to
direct pages of scanned documents to a variety of printers. And an
interface to CUPID has now been developed based on WWW
(WorldWideWeb) protocols, aligning CUPID with the most rapidly
growing application on the Internet.
With regard to copyright, the CUPID architecture spec limits
itself to the push model, where the person ordering the printing is
physically in possession of the file to be printed (he or she pushes
the file over the network to the printer), and it is presumed when
this happens that copyright permission has already been obtained.
The documentation anticipates adding the pull model--where the
person ordering the printing pulls the bits from a third party, along
with copyright clearance, royalty payments, etc. This model will be
implemented in Phase 2 of CUPID.
In addition, the CUPID group will follow up on its efforts to
date by establishing contacts with various standards bodies in the
printing and computer industries, to ensure both that any useful
existing standards (such as, for example, paper color) are adopted by
CUPID, and that the standards which CUPID is itself creating will be
appropriately codified. Other future plans for CUPID--some
tentative--include the following:
o More features will be added to the existing prototype
o Hardened implementations will be produced and used as on-
campus general-purpose print utilities at Cornell and Harvard.
o CUPID clients, which currently exist only in Macintosh versions,
will be created for other environments.
o The link between CUPID and WWW will be vigorously pursued.
o Several experiments are being discussed with commercial
publishers who wish to allow customers to order documents from a
central catalogue for local printing.
o As consensus is reached within the Internet community on
such issues as authentication, accounting, and encryption, the
selected solutions will be incorporated into CUPID.
The Promise of CUPID
Once CUPID's capabilities become ubiquitously available on campuses
around the country, the possibilities for revolutionizing the way
information is obtained and distributed in the academic environment
are many. Following are some of the ways these capabilities could be
put to use:
Replace print-then-distribute with distribute-then-print. Rather than
printing all documents at a single point and then shipping them to
multiple locations, instead do the printing closer to--or perhaps at--
each point of sale. This can lead to lower press runs, less
warehousing and waste, faster response to content changes, reduced
overhead, and lower costs.
Foster print-on-demand. A researcher or student may browse a
digital library electronically and request local printing of the
required book--or of a few pages of the book. More generally, a
publisher might offer a catalog of available documents which would
be delivered on request to a printshop selected by the consumer.
Empower custom publishing. Textbooks and other course material
could be distributed in different versions to different locations. Local
editions of news reports, business briefings, product reviews, and an
endless variety of other documents could be easily created and
disseminated. Highly focused advertising becomes straightforward.
This list gives just a hint of the impact that network publishing
will have on the worlds of academia and commerce. New businesses
will come into existence and old ones will grow and shrink as this
technology becomes integrated with daily life. Few will be totally
Take me to the index