+ Page 64 +
 
---------------------------------------------------------------------------
 #######   ########   ########  ###########
   ###     ###   ##   ###   ##  #   ###   #   Interpersonal Computing and
   ###     ###   ##   ###   ##      ###               Technology:
   ###     ###   ##   ###           ###        An Electronic Journal for
   ###     ########   ###           ###            the 21st Century
   ###     ###        ###           ###
   ###     ###        ###   ##      ###             ISSN: 1064-4326
   ###     ###        ###   ##      ###                July, 1995
 #######   ###        ########      ###     Volume 3, Number 3, pp. 64-90
---------------------------------------------------------------------------
 
   Published by the Center for Teaching and Technology, Academic Computer
           Center, Georgetown University, Washington, DC 20057
     Additional support provided by the Center for Academic Computing,
       The Pennsylvania State University, University Park, PA 16802
 
       This article is archived as MCADAMS IPCTV3N3 on LISTSERV@GUVM
                      (LISTSERV@GUVM.GEORGETOWN.EDU)
--------------------------------------------------------------------------
INVENTING AN ONLINE NEWSPAPER
 
Melinda McAdams,  Content Developer, Digital Ink Co.
 
     I have spent the last sixteen months building an online
version of The Washington Post. In many ways the experience has
been like trekking in a wilderness without a map or even a compass, and all
of us who have worked on the project have learned a great deal along the
way. We have faced most of the hard questions about translating the
newspaper to a new medium, and if we have not arrived at definitive
answers, we have at least explored and grappled with multiple
possibilities.
 
          An online newspaper has to be different from the print
product. Our task was to take a lot of large pages that are covered with
printed text arranged almost haphazardly and that are worthless twenty-four
hours after they appear and translate them into a medium where their
contents will have value indefinitely, be part of a much larger collection
of data, be read on a small screen in a scrolling format, and be searchable
in various ways.
 
+ Page 65 +
 
 
     We began with the content of The Washington Post, assuming that we
would use all or most of it and supplement in whatever ways seemed
suitable.
 
     First we had to decide how much the online newspaper should
resemble the print product. What should the online version provide that the
printed version does not? How should content be organized in the online
environment? What parts of the newspaper's traditional format can be
carried over and what parts must be let go?
 
 
     Consideration of the human issues included decisions about what kind
of staff is required to produce an online newspaper; that is, are
journalists necessary, or is producing an online newspaper just
a production job? We have also had to determine, with the help of our
colleagues in the newsroom, what relationship the newsroom should have to
the online product.
 
          Throughout this long process, we have continually asked
what users will want from an online newspaper. We find out a little more
every day from our thousands of beta testers, but the information is still
hard to quantify.
 
BACKGROUND: DIGITAL INK CO.
         The Washington Post Company created a separate subsidiary
to develop and produce an online version of the newspaper and other
electronic products. That subsidiary, Digital Ink Co., was launched in
October 1993, when I was a copy editor on the Metro desk at The Post. I
submitted my resume within days of the announcement and was hired in March
1994, one of the first ten employees. I had done graduate work in interface
design theory, mostly related to the organization of information online,
and I had worked as a free-lance producer at Prodigy for several months
during grad school.
 
         At the time I was hired, the platform for delivery of the
online product had not been chosen, although we were already
evaluating the platform we eventually selected: AT&T Interchange (then
still owned by Ziff-Davis, which had developed the software).
My understanding is that a number of possible platforms were considered,
including stand-alone BBS software and the major online services.
Interchange was chosen because it would allow us to do many things that the
others would not, such as integrating photographs and other graphics.
 
     Digital Ink is independent from the newsroom, so
that any coordinated effort requires mutual consent. A legal agreement
between Digital Ink and the newspaper (which itself is a subsidiary of The
Washington Post Company) defines Digital Ink's rights to use the content of
the newspaper. However, there is plenty of communication between the top
editors in the newsroom and the people who run Digital Ink, so this
independence does not prevent the newspaper from knowing what the online
service is doing.
 
+ Page 66 +
 
           The head of Digital Ink is the editor and publisher.
Reporting to the publisher are several directors; at the beginning there
were directors of the online service, the audiotext service, marketing,
operations, and product development. Now there are also directors of
graphic design, technology, and finance. In just over a year, the staff has
grown from ten to forty, not counting a number of free-lancers, temporary
and part-time workers, and interns.
 
         The online product is produced by an online manager, five
content developers, nine online producers, and several artists.
 
THE NEWSPAPER METAPHOR
 
         A fundamental principle of this venture has been that the
newspaper metaphor provides a superior structural model for an online
service. The primary reason is that the printed newspaper is very easy to
use; it is so user-friendly that anyone who can read can figure it out.
 
     The importance of a good user metaphor can not be underestimated. A
person's expectations and assumptions about how an online system works and
what it can (and cannot) do come largely from this metaphor. If those who
design a system neglect the user metaphor, the users' assumptions may not
correspond very well to the actual system, and as a result they will find
it confusing and difficult to use. Faced with a computer response that
seems strange, users often make up an explanation, which, right or wrong,
then becomes part of their understanding of how the system works.
 
+ Page 67 +
 
 
     Much of the literature on user interfaces refers to the need for
metaphor in the interface.  Just as a speaker chooses metaphors that will
make his or her meaning clearer to the audience, a designer must choose
metaphors that help the users understand the system.
 
         The newspaper metaphor uses the "front page" as the entry
point to the system; it relies on headlines to tell users what items in the
system are most important; it employs division into sections similar to the
sections of a large metropolitan daily. It does not restrict the content of
the system to the content of a newspaper, but it promotes an assumption
that users can follow a typical newspaper organizational structure to find
any information they want.
 
     While the newspaper offers a good model for browsing, it provides
little in the way of models for searching. Libraries, books and magazines
supply card catalogues, indexes and tables of contents, but a newspaper at
best provides a partial index to its contents on a single day. Features
such as advice columns, the comics and even the classified ads may be found
in different parts of the newspaper on any given day, making the
navigational model of a newspaper unreliable. The newspaper is a relatively
small, manageable set of information presented in a form that can be easily
scanned. An online service must provide access to many times more
information. This is not to
 
 
+ Page 68 +
 
say that the newspaper metaphor is unusable; it is in fact useful in a
number of ways, particularly for an online newspaper's purposes, but it
does not answer all the questions involved in building an online
information system.
 
          With the newspaper supplying our basic metaphor, we set
out to build onto it the additional features that an online service needs.
 
TRANSLATING THE NEWSPAPER
 
     One of the difficulties in creating an online newspaper involved
organizing information along newspaper section lines. The Sports, Business
and Metro sections suggest very practical online sections (although Metro
was not without problems online, as will be discussed). Most weekly
sections of The Post, such as Health, Travel and Book World, also fit
naturally into subject divisions. The A Section (the front section of The
Washington Post) comprises national and world news for the most part, with
editorials and op-ed at the back; Page One may of course carry stories from
any section. The A Section, though, was easy to deconstruct. The Style
section (a daily section somewhat equivalent to what some papers call
Lifestyle or Living) presented the most persistent problems.
 
     From the A Section, we swiftly spun out National, International and
opinion (now called Comment!) sections for our online product. Business and
Sports clearly deserved to remain intact and to receive prominent billing.
Some argued that Style should also stand, especially because it carries
strong brand-name identification in the Washington area.
 
+ Page 69 +
 
 
Giving up the Style section
 
     Style runs profiles, interviews and features about figures from the
entertainment world and the national political scene. It runs stories on
trends and hobbies, gossip, Ann Landers, Miss Manners, articles on families
and aging. It also runs reviews of movies, live and recorded music, books,
theater and dance. For some Post readers, Style is the only section they
care about. What prevented us from putting Style online intact was the
eclectic mix that makes it so appealing in print. Browsing through one
day's printed Style section is very enjoyable, but turning it into a
navigable online section proved impossible.
 
     Compounding the problem, The Post also has a Weekend section, which
comes in the paper each Friday. Weekend runs movie reviews, a large section
on music, theater and dance reviews, but no book reviews. Its features
often focus on activities and how to participate in them (hot-air
ballooning, zydeco dancing, indulging in afternoon tea at a fine hotel).
The overlap with Style is significant but not total.
 
     In an online environment, users cannot be expected to go to two
different sections to get their movie reviews, their articles about
recording artists and so on, just as we would not expect them to go to both
Style and the online Books section to find book reviews. The clear division
of the print product rises partly from the relationship of time to a
printed newspaper: Weekend is a one-day-a-week insert. Online, we would not
make Weekend disappear for six out of seven days.
 
     Although it would be possible to create two sections with overlapping
content (all movie reviews could appear in both Style and Weekend without
taking up additional disk space), that would probably confuse users. From
the beginning, it was clear that one of our assets would be our ability to
organize information better than the newspaper can, to make things easier
to find -- to put all the movie reviews and features and profiles (even
business news from Hollywood) in one place, and make them searchable, and
connect them to the theater and show time information, and keep them
available online for as long as the movie was playing.
 
+ Page 70 +
 
     The result: We have created separate sections online
for Movies, Music & Concerts, Theater & Dance, Books, and so on. However,
we don't always know what to do about a Style feature about the life of a
sixteen-year-old transvestite, or where to put the gossip column (because
it's as likely to mention Hillary Clinton as Michael Jackson) or Ann
Landers. And we find that our users miss Style; they come online looking
for the Style section because we are The Washington Post, and Style is an
important part of our newspaper.
 
     Periodically we reconsider having a "Style Online" section for the
kind of articles that are so distinctly a part of the paper's Style
section. But we know that the minute we have the word Style on an online
section, the users will be asking where the movie reviews are.
 
Expanding local coverage
 
          The Post's circulation area covers several counties in
Maryland and Virginia as well as the District of Columbia, and we have
about a dozen zoned weekly sections, each aimed at one county or locality.
We wanted to integrate the articles from the zoned weeklies into our online
Metro section, and we also hoped to include as much local information as we
could beyond that, recognizing that with such a broad geographical area,
the printed Post cannot possibly cover all of it thoroughly. To give the
online users much more than they get from their newspaper, we planned from
the start to bring local government and school information online.
 
+ Page 71 +
 
     This made it necessary for the online Metro section to become very
large. As we began to add the local information a few months ago, we found
that our original organization plan was less than ideal. We had intended to
put local news that was of broad interest in the top-level Metro section,
where everyone would see it, and we created a subsection called
Neighborhoods within which we meant to put all the county information. In
practice, we learned that this structure forced users to dig down too deep
(from the opening screen to Metro, from there to Neighborhoods, and from
there to the county of choice) to get their local information. Meetings
with beta-test users showed that many people who were interested in that
information had been unable to find it. (Part of the reason is that they
did not expect to find more in the online section titled Metro than the
printed paper provides in its Metro section.) In response, we have
reorganized the local information to make it available directly from the
front screen of the Metro section, and the Neighborhoods subsection will
probably be eliminated.
 
     This structural challenge has come up again and again, because users
tend to assume that the surface, or top level of the service, shows
everything that is in the service. It is not clear why they make this
assumption. Newspaper readers do not look at Page One and assume they can
see all contents of the day's paper there. No online service presents a
complete list of its contents on the first screen. Yet people seem to lack
understanding of tree structures, in which multiple branches lead down from
a single point (or screen). It may be that they expect to find a complete
table of contents.
 
     A valuable lesson we have learned is that we must be flexible; a
design or hierarchy that we think is very good may have to be changed
substantially if we discover that users find it unwieldy or unintuitive.
 
+ Page 72 +
 
 
Eliminating Page One
 
     It seemed inevitable that the newspaper's Page One would disappear.
The screen is so much smaller than a broadsheet page, and although we
discussed making a digital scan of The Post's front page and transferring
it to the screen, the length of time it would take for such a high-
resolution image to come up makes that impractical. Besides, no one wanted
us to be chained to the paper's front page throughout an entire day; our
front screen will always reflect the latest news, not what The Post had set
in type and committed to ink about midnight the night before.
 
     On very rare occasions, every article on our front screen is a Page
One story from today's Post. But we can feature no more than five articles
on our front, while The Post typically puts seven or eight stories on Page
One. More often we feature only one or two Post stories on our front
screen, because we have many other kinds of content to highlight. We change
the front screen at least three times between six a.m. and eleven p.m. If a
big news story breaks, we can put it on our front screen within minutes; we
did so with the Oklahoma bombing story, using reports from AP Online (which
we offer as part of our service).
 
     Our decisions about our front screen are based on different criteria
from the newsroom's, however, partly because "inside" we have separate
screens on which to feature the top national news, the top business news,
and so on. We are as likely to feature a guide to mid-Atlantic beaches on
our front screen (and we have) as we are to report news of an act of
Congress.
 
+ Page 73 +
 
 
A mirror of the print product
 
     We recognized that many users would come online and want to see their
Washington Post in the format that was familiar to them, no matter how many
advantages the reconfigured version offered. They would want to see what
The Washington Post had put on Page One; they would want to find their
Style section. We decided that if we could provide a mirror of the print
product in a way that would not require excessive person-hours and would
not interfere with the online structure, we would, because then we coul
satisfy more users.
 
          We were able to do this earlier this year, in an online
section we call Today's Newspaper, where a very simple set of folders gives
the user lists, by headline and byline, of all the articles in each section
of today's paper. Any article can be opened for reading by clicking on its
headline. The response was immediate and very positive. It seems that many
of our users prefer this format. Some, but not all, are people who moved
away from the Washington area and can no longer get The Post; others get
the printed version but enjoy reading online.
 
     Some users want the online service to be a perfect mirror of the day's
Washington Post, and others want an altered, online-adapted version. We
thought that naming our service The Washington Post Digital Ink would lead
users to expect something other than the newspaper, but we have found that
many people expect and even prefer what we have provided in the Today's
Newspaper section, and some even think it should appear on the first screen
of our service.
 
+ Page 74 +
 
 
NEW CONTENT
 
     For journalists, part of the appeal of an online newspaper comes from
the potential to offer all kinds of information that will not fit in the
print product. A common joke plays off the New York Times slogan: "All the
news that's fit to print"; the alternative version reads, "All the news
that fits, we print." The newspaper economic model demands that news be
delivered in a set proportion to advertising, both measured in column
inches of space. A common ratio is 60-40, meaning sixty percent advertising
and forty percent editorial material. This model keeps the per-copy price
of the paper low, but it means that each day's newsroom decisions depend on
how much news will fit into the space that's available.
 
     In recent years, many newspapers confronting a "shrinking news hole"
have decided to cut out certain information they traditionally provided. A
notable example is stock market tables; some papers have chosen to provide
some market information via audiotext systems instead of printing it. The
newspaper's readers find a phone number in the business section; when they
call it, they use an automated menu to get the quotes they want. Other
papers have eliminated different features, such as particular comic strips.
 
     Online, we have a bottomless news hole. This means that not only can
we carry information that the newspaper may have cut out; we can also offer
information that the newspaper never had, and we can keep it always
available, day after day.
 
+ Page 75 +
 
 
Information from outside sources
 
     Government information is an obvious example: We list the office
address and phone number of every member of Congress, along with e-mail
addresses where available. Local government information is also on our
service. Users can find out what number to phone for information about
recycling, county health services, parks and recreation, and so on. In this
way an online service with a local emphasis can become an enhanced phone
directory, providing descriptions of services along with the name of an
agency and its address, phone number and hours.
 
     The difference between this collection of information and a phone book
is organization and searchability. However, that means we must devote staff
hours to organizing the information and making it searchable, and also to
entering it in the first place and keeping it updated. We have found that
this requires us to develop an inventory of, and a maintenance schedule
for, the information that we are putting online. Unlike information in the
print product, which is likely to end up on the bottom of a bird cage a day
or so after delivery, information in the online service must be looked
after, not merely put there and forgotten. The more of this kind of
information we choose to put online, the larger the maintenance overhead.
This has an impact on staff size and also on the kind of staff hired (e.g.
it is impractical to have an experienced journalist doing data
entry for hours, but it is also inappropriate to rely on a data entry
person to do fact checking).
 
+ Page 76 +
 
     Allowing local governments and organizations to maintain their own
information on our service is an option, but it also requires an investment
of staff time that is not part of the newsroom's regular operation. Members
of our staff must contact and meet with people from any office or
organization that we would like to have online, demonstrate our product,
and set up procedures that will enable those people to enter and maintain
their data. These meetings are not sales calls; they involve only editorial
staff. Yet the situation turns out to be very similar to a sales call--our
group must convince the other group that it will be to their benefit to
come online with us.  This creates a significant divergence from
traditional newspaper practices: In the newspaper, if information is
provided by an outside entity, it is in the form of an advertisement, even
if the newspaper gives the space away as a public service. Online,
sometimes it is to our benefit to give away "space," and we do not consider
it an advertisement. In those cases, we take great care to label all the
information as provided and maintained by the outside entity; we consider
it a matter of responsible journalism to make it clear that such
information has not been collected or verified by The Washington Post.
 
     In addition to government and public service information, we are
exploring other sources of information outside the newspaper, including
restaurant reviews and a guide to inns and resorts within easy driving
distance of Washington. We have had discussions with the local Sierra Club
chapter and other outdoors groups, and with the author of a book on area
bookstores and literary history. The goal is to provide useful and
interesting information to people in the Washington area, which is
certainly in keeping with the goals of The Washington Post. However, as
this overview should indicate, the staff of Digital Ink pursues information
that no reporter at The Post has ever had reason to pursue, much of it
having nothing to do with "news" in the traditional sense.
 
+ Page 77 +
 
 
Archives and linked articles
 
     Another type of content that is not available in the newspaper is our
archives. In-house, The Post has used a searchable database of past
articles for almost nine years. We put that database online, making it
possible for our users to search for articles back to 1986, and they have
responded enthusiastically. The most common criticisms concern our fifty
cents-per-article surcharge and the range of the archives; some users want
to search farther back than 1986.
 
     We are also able to link older articles to articles from today's
paper, or to breaking stories from the wires, providing instant perspective
or background. When the Justice Department decided to look into the
acquisition of a local software company, for example, we were able to
quickly tie in the initial article about the acquisition, an earlier
article about a major change in the company's product strategy, and our
"Washington Post 200" profile of the company.
 
     This functionality affects staffing requirements, in that the people
who put the news online must be adept at conducting online searches and at
rapidly evaluating a set of articles and choosing a few that provide
suitable background information. These searches are performed both on our
own online archives and on public information available on the Internet,
sometimes in conjunction with phone calls to agencies or organizations that
may be able to point us to a specific Internet site. The necessary skills
combine those of researchers, editors, and reporters.
 
+ Page 78 +
 
 
Talking back to users
 
     Discussions present an obvious content type that does not exist in the
newspaper. We knew when we set out that other online services had found
that users spend the majority of their online time in real-time chat rooms
or participating in discussions or forums. (The Interchange platform does
not yet offer chat functionality, so we have not dealt with chat issues.)
We have learned that every discussion on our service should be read every
day by someone on our staff (several people can do this, so long as no
discussions are neglected)--
not to moderate or censor the postings in any way but to answer users'
questions and learn about any problems the users are having or suggestions
they are offering.
 
     We have also learned that users of an online service from The
Washington Post expect to find Washington Post reporters and editors
online. Several reporters and columnists have an online presence now, and
The Post's executive editor and managing editor are online and participate
in one specified discussion. However, it is clear that our users expect
more. We often receive requests for the e-mail addresses of particular
reporters, or for a list of e-mail addresses for all Post reporters,
columnists and editors. Users have so far been very gracious when we
explain that we are still working to get more newsroom people online and
that we don't want to give out email addresses of reporters who have not
agreed to it. However, the implication seems to be that users see a
newspaper's going online as evidence that the paper now wishes to have a
closer relationship with its readers, and they are eager to let their
opinions be known--not just in public discussions, but in personal e-mail
to specific individuals.
 
+ Page 79 +
 
 
      Any online entity that allows discussions must confront
the question of whether to moderate them. We have heard that some people in
the newsroom are uncomfortable with the fact that we have chosen not to
moderate discussions, but the decision seems necessary based on existing
case law, which indicates that if an information service provider edits or
censors discussion postings in any way, that provider may be open to a
lawsuit from someone who feels a particular posting was libelous. The
Interchange platform prevents the editing of any posted note (even by the
person who wrote it, even if that person is an editor), but it does allow
the deletion of entire notes. This leaves the option to remove a posting
and send it back to its writer, asking the person to rewrite and repost the
note without whatever portion may have been offensive.
     Even though we don't moderate discussions, we have found that they
require a great deal of staff attention. Users ask questions about the
service, where to find things and how to use particular functions, that
only members of our staff can answer. But as the number of discussions
grows, it takes more staff time to keep up with them. This marks another
significant difference between an online service and a newspaper: A
newspaper may have one full-time copy aide screening letters and funneling
them to the Editorial desk, and an ombudsman receiving mail from readers,
and columnists receiving their own mail, but all these communications are
isolated from one another and can easily be answered with form letters. In
online discussions, the complaint or other remark is public and so is the
reply--or lack of a reply. In a neglected discussion, users' postings
become more caustic as they begin to feel that no one from the staff of the
online service is paying attention to them.
 
ORGANIZATION OF AN INFORMATION SPACE
 
     Organizing vast amounts of information continues to be a challenge. In
fact, the challenge grows as the scope of our information space grows.
 
     Our basic organization scheme follows the newspaper's, as described
earlier: Metro, National, International, Business and Sports. To
accommodate articles from Style, Weekend, Book World and a number of other
special sections (some originating with the paper and some, such as
Outdoors, unique to the online service), we also have Living and Arts &
Entertainment sections, each with about ten subsections. Our archives, the
CIA World Factbook, the US Government Manual and other reference sources
are in a section called Reference Desk. Advertising can be found in
Marketplace, although it is also linked throughout the service.
 
+ Page 80 +
 
     On its face, this scheme appears to be simple and straightforward. But
it violates a basic principle of user interface design: More than about
five navigational options on one screen are too many and will confuse the
user. The front screen of our service shows ten links to inside material
(one leads to a feature, not a section) in addition to the various featured
links that are headlined in a Page One format. While meetings with beta-
test users indicate that users understand our list of sections (a vertical
column at the right side of the front screen, or directory) to be our table
of contents, they nevertheless asked questions such as "Where can I go to
see what's on Page One?" and "I just want to read the news. Where is that?"
 
     Inside the main sections, we found other organizational problems. One
is that certain of our subsections seem to belong to more than one "parent"
section. Computing, for example, was conceived as a part of the Living
section, because we determined that it should focus on computer use as an
avocation and not on industry news. Users visiting our Business section,
however, wanted to know where to find computer articles, and eventually it
was decided that a link to Computing should appear in both Business and
Living. Then, of course, we found that users are initially surprised to
find what they think are two different Computing sections. We encountered a
similar difficulty with Real Estate, which began as a subsection of
Business but late in our development phase was shifted to a subsection of
Marketplace called Classifieds. Later a link to the Real Estate subsection
wound up appearing in both places.
 
+ Page 81 +
 
 
     In retrospect, our initial structure could have been improved.
Differing opinions on what the structure should be may have resulted in a
less clear design than either side alone would have produced. At several
points after the structure was in place, it was suggested that the entire
structure should be revisited and refined, but a redesign never came about,
and so the existing structure is essentially the one we laid out on paper
in the spring of 1994. Instead, we have made a large number of incremental
changes (such as adding Computing to Business). This is a typical response
when software-based systems show a need for improvement, but one that may
do more harm than good, because incremental adjustments may end up creating
conflicts and inconsistencies and hence confusion.
 
 
Logical paths
 
     The process of organizing information online is very different from
most of the work done in the newsroom. Even the coordination of election-
night coverage, although a huge task, is more a matter of collecting
information than organizing it. It also differs from writing a book,
because while a book is navigated by reading from beginning to end, by
using the index, or by browsing the table of contents, the writer
concentrates on the first form of organization, the logical progression
from start to finish. Online, we start with a screen that, like
the top of a tree diagram, must present the points from which many branches
lead outward. We can lead through several "stops" to a thing that is not on
the first screen, but it must be clear to the users that something on that
first screen would logically lead to the thing below.
 
     We must avoid creating grab bags (like the Style section); when users
find a data set titled "Local Parks and Recreation Areas," they should not
follow a link and come to a description of a national park two thousand
miles away. However, a logical path could lead to a list of national parks
in the local area, and from there to a list titled "National Parks Outside
the D.C. Area." A tree model provides a good way to give more to users who
want more and at the same time avoid giving too much to people who want
only an overview.
 
     Another example is in the way we organize documentation. Instructional
material is broken up into topics, but not every topic needs to appear on
the top screen that presents one set of help documents; it is better if
there is an overview of search techniques, for example, and then from the
overview users can go to more detailed descriptions of every aspect of
searching. If all the documents about searching are listed on the top
screen, how will the user know where to start? On seeing too many documents
about one topic, the user may decide it is all too complicated and not read
any of them.
 
+ Page 82 +
 
 
Selectivity, navigation, searching
 
          One of the harshest criticisms a person can make about
online information is to call it "shovelware."  The primary meaning is that
the information is just a giant heap "shoveled" online with no regard for
its value or meaning. But another aspect of shovelware is lack of
organization. If you took the contents of the Encyclopedia Britannica and
put them online without a search engine, that would be shovelware too.
 
     The first step in avoiding shovelware is to be selective about what we
present online. The next steps are to make it navigable by both browsers
and searchers.
 
         In a newspaper, navigation is as simple as turning the
pages, but online it requires movement from screen to screen, and that
takes time. A person can turn all the pages in a newspaper in a few
minutes, but to open the same number of screens on a computer (particularly
with a graphical interface) would take much longer.  Thus we must provide
ways of movement that will allow users to skip unnecessary sections and
features. We must also avoid making them open too many screens to get to
what they want.
 
     Searching is one way to allow users to move quickly from one area to
another. All online services provide some form of user-initiated search,
but these are not always satisfactory. Trying to get to the airline
reservation system on America Online, I typed "airlines" and "planes" and
"flights" and got nothing. I finally got it when I typed "Eaasy Sabre" (the
name of the section). We cannot assume that users will know the exact name
of the section or feature they want to find. Attaching keywords such as
"planes" and "flights" is a large task, however. If we had sufficient staff
to add keywords to all articles and all features on our service, users
would be able to find just about anything. But we get about 200 new
articles a day just from The Post, and the task is just too large.
 
     Search mechanisms may work well for users who know exactly what they
want, but we also want to preserve the comfortable browsability of the
printed newspaper. Consistency in the organizational scheme enables the
users to move around more efficiently as they become accustomed to the
system. Other aspects of carefree browsing include preventing users from
losing track of where they are, or feeling lost, and making it easy for
them to go back to something they saw earlier.
 
+ Page 83 +
 
           We need to organize large sets of information such as
restaurant reviews, recipes and descriptions of local attractions. Such
sets need to be browsable as well as searchable, should have links to all
possible related information, and should avoid links to peripheral or
unrelated information (misleading or extraneous links). A small set, such
as attractions on the National Mall, turned out to be straightforward:
create a map of the Mall, research each attraction, create a document for
each attraction, and link them all to the map.
 
     A larger set requires more planning. For example, a freelancer
collected information about all the public parks in the Washington area,
put information about each park into a separate document and sorted them
into folders by county or quadrant of the District of Columbia. This
organization does not account for users who want to find all parks in
Virginia that have soccer fields or to find parks with lighted tennis
courts regardless of where they are. Every time we build a set of
information, we must ask whether we need to provide other groupings and
whether the primary grouping is the most useful for presenting this
particular set.
 
 
STAFF REQUIREMENTS
 
          The original staffing plan for our company called for a
few experienced journalists to run the online service and direct several
junior "packagers" whose jobs would consist mostly of getting data into the
system. Choosing the Interchange platform forced a change in that plan,
since Interchange's graphical screen layouts require a fair bit of writing
original headlines and other descriptive copy. It was eventually decided
that a strong journalism background would be required for all new hires for
what has become the editorial staff of the service. It is not easy to train
non-journalists to write good headlines; on the other hand, reporters with
no copy-editing experience are also untrained in the art of the headline.
We found it necessary to bring in experienced newspaper copy editors to
edit screens after they are created, even though the articles that come
from The Post require no additional editing.
 
     While journalists' skills suit many of the requirements of an online
newspaper, a person's previous online experience is not trivial in this
environment. We have interviewed some
people who reveal they have never been online, never used a modem to
connect to anything. Even some recent graduates of journalism school have
never used anything but an online library card catalogue. A smart person
with good journalism experience can, of course, learn a lot about the
online world in a fairly short amount of time, but there are differences
between a person who has "lived online" and someone who has not.
 
+ Page 84 +
 
 
     A journalist with little online experience tends to think in terms of
stories, news value, public service, and things that are good to read.
These are the staples of a one-way medium. But a person with a lot of
online experience thinks more about connections, organization, movement
within and among sets of information, and communication among different
people. Online is bi-directional. This is not to say that journalists
cannot think about the latter or that new media people cannot think about
the former; it merely indicates that neither side alone is likely to
produce an above-average online newspaper. Journalists often complain about
the shallow content on commercial services such as America Online and
Prodigy, but if they have not also thought about the things those services
do well, their vision of an electronic newspaper may be too flat, too
closely tied to the print world to which they are accustomed.
 
          Thinking about an electronic information space requires
that we conceptualize a place where people spend time, a place they return
to again and again, rather than a product they receive, use and discard.
People who have not lived online do not seem to know this. They see a flat
screen and they equate it with flat paper.
 
     I feel we have learned that to produce an appealing online newspaper,
an organization needs good, experienced journalists and good, experienced
online people and some people who are both, and all of them need to consult
closely and frequently, often in small, autonomous groups. If some of the
online people are non-journalists with a strong user-interface background,
it can only help the product. I have been stopped short more than once by
one or another of the non-journalists among us and forced to reconsider
something because he or she said, "Why do you want to do it that way?" and
brought me to the realization that my idea was very newsroom-like and not
likely to be intuitive to the non-journalists who will be our users.
 
 
+ Page 85 +
 
 
There may be a tendency among the journalists to feel that their opinions
should carry more weight because the product is, after all, a newspaper,
but they should remember that the medium is different from the one they are
used to. The journalists need to learn to defer to the new media folks on
issues of navigation and organization. The new media folks will naturally
defer to the journalists on issues of content and news judgment. By
encouraging a good balance between the two kinds of expertise, an online
newspaper may come to embrace the best of both worlds.
 
THE NEWSROOM'S ROLE
 
     The Washington Post newsroom is entirely separate from our
organization. The reporters and editors who come online and participate in
discussions do so voluntarily. Our service is the online service of The
Washington Post, but it is merely a vehicle for the content of the
newspaper; it is not created or shaped by the same people who put out the
newspaper.
 
     The users presumably know this, and none have expressed surprise or
dismay over it. As noted before, they expect to be able to communicate with
the newsroom, but it does not seem that they expect the service to be put
together by the same people who bring them their newspaper.
 
     There are many good reasons to set up a separate staff for the online
product, but one of the drawbacks is curtailed interaction with the
newsroom. A closer relationship with reporters might give us ideas about
new content for the service. If our desks were in or near the newsroom,
reporters and editors could stop by for a moment and perhaps get ideas from
us--if someone was going to a city council meeting, for example, we could
ask him or her to pick up a copy of the
 
+ Page 86 +
 
budget for us to scan and put online. Another thing we miss is advance
warning on special projects and series. The Interchange platform has
restrictions that make it impossible for us to use art or charts from the
newspaper, and if we knew about the newsroom's plans in advance, we might
be able to prepare our own art to accompany certain articles. But our
ability to coordinate with the newsroom is limited by our being in a
separate building. The results are not necessarily negative, but the end
product is certainly a different one than would be produced with more
newsroom involvement.
 
          It is not apparent whether our users place any value on
our news judgment or our ability to sort through information and organize
it for them. However, they do seem to want Washington Post articles
delivered in this format, online; we are often asked (via e-mail) how to
find specific columns and features. The users have also requested
additional information: more details about the federal government and
actions of Congress, more data from the local communities. Some have
complained about our use of AP Online articles, which we depend on for
breaking news, while others have said they are happy to get up-to-the-
minute reports on our service. It is not yet clear whether sometime in the
future we will need to have reporters on the online staff or whether the
newsroom will someday supply breaking news to our service hours ahead of
the print deadline.
 
+ Page 87 +
 
 
ADVERTISING ONLINE
 
     We assume that advertising must help offset the price of the online
newspaper, as it does for the print product. However, it is difficult to
get advertisers online before we have a good number of users for them to
sell to, and so we still have had relatively little experience with
advertising on our service.
 
     We have had many discussions about where to place the advertising and
how to handle it. All of our questions have not yet been resolved.
 
     Outside the Marketplace section, advertising could appear on a section
front, such as Business or Books; in the index of a section, where it would
be contained in a folder labeled, for example, "Mutual Funds Advertising";
or on an article itself, in the left margin, where there is space for links
to other items. Advertising has appeared in the last two places but not yet
on a section front, where it is felt that the available space is too limited.
 
          A "link" in Interchange is a small icon, and an icon by
itself is not very intrusive. But we have sometimes used an ad link in
conjunction with an illustration, putting both in the left-hand margin of
one of our articles. The illustration in these cases was relatively large,
taking up almost half of the white space available in the margin. I am not
aware how the users felt about these ads, but my own opinion is that they
were intrusive, even more so than Prodigy's ads, because they were next to
what I was reading and because the article otherwise occupies its window
alone. Our marketing people argue that having ad art beside the article you
are reading is exactly what happens in the newspaper, and no one ever
complains about that, but I have argued that the screen is a smaller space
and differently defined. I do not object to putting the ad link alone in
the margin, but more than that seems to interfere with the experience of
opening and reading an article.
 
           We have also discussed specific kinds of advertising
placement, the most common example being an ad for a best-selling novel
linked to The Post's review of the book. The editorial people have firmly
objected, saying that would undermine the integrity of The Post, fostering
the appearance that we wrote the review in exchange for the ad. What few
object to, though, is linking an ad for a bookstore to a book review.
Marketing people have sometimes expressed skepticism at this distinction,
and perhaps they are right. One possibility is to segregate all such
advertising in the Books section in an index folder labeled "Books
Advertising," but our marketing folks are very reluctant to do that because
the index folders are small and easily overlooked.
 
+ Page 88 +
 
 
     Similar discussions have been held about our Restaurants section. We
anticipate the construction of a restaurants directory, with names,
addresses and phone numbers of all local restaurants. To each restaurant's
listing we could link The Post's review, if there was one, and the
restaurant's ad, if there was one. But the larger debate concerns cross-
linking the ad and the review. If we placed the ad link on the review, it
would compromise the review. If we placed the review link on the ad, and
the review was negative, the restaurant owner would not be happy. If we
allowed the restaurant to choose whether to link the review or not, then
what would the user think about the ads that did not have a review link
attached?
 
           Some of the unexplored avenues for online advertising
include sponsorship, such as a feature on local rock bands sponsored by a
local music store chain, and explanatory content, such as a series called
"How to Invest for Retirement" authored by an investment brokerage. These
would be multiple element packages created and maintained by the advertiser
and identified as such. So far, however, few advertisers have been eager to
experiment.
 
SUMMARY
 
     At the time I am writing this (June 1995), The Washington Post Digital
Ink is on the verge of launching commercially.  Surely we will learn many
more things afterward. But much of what we have learned to this point could
apply to other online newspapers.
 
     Foremost is that an online newspaper cannot be a strict translation of
the print product. To try to put the newspaper online and stay true to the
print concept would be to severely handicap the online product. Most users
will welcome the enhancements in the online version. But many users--even
those who like the online version very much--will continue to read the
printed version every day (or at least as often as
they ever did) because it has its own appeal, its own superior features.
 
     Second, organizing the breadth of material found in a daily newspaper
is no small task, and organizing the superset of many days' articles and
outside content is even more taxing. Figuring out what to keep from the
print product, what to discard, and what to bring in from outside may
require several practice tries. Integrating the things that online does
best-automated searching and communications between users and also between
users and producers--may take some work before a comfortable mix of these
and the traditional newspaper is discovered.
 
     Last, the people bringing you your online newspaper may be very
different from the people who bring you the print product. It remains to be
seen whether all will uphold the same standards of journalism that
professional journalists have sought to promote and protect in this
century. Given the nature of the online environment, however, it seems
certain that the closed fortresses of media power, from which all news
emanates in one direction, will have to let down drawbridges and allow a
greater degree of two-way communication between inside and outside. This,
more than any aspect of content or structure, may make the greatest
difference.
 
 
 
+ Page 89 +
 
***
SOURCES:
 
Laurel, Brenda, ed. The Art of Human-Computer Interface Design. Reading,
Mass.: Addison-Wesley, 1990.
 
Nielsen, Jakob. Multimedia and Hypertext: The Internet and Beyond. Boston:
AP Professional, 1995.
 
Norman, Donald A. The Design of Everyday Things. New York: Doubleday, 1988
(originally published as The Psychology of Everyday Things).
 
Suchman, Lucy A. Plans and Situated Actions: The Problem of HumanMachine
Communication. Cambridge, England: Cambridge University Press, 1987.
 
Tognazzini, Bruce. Tog on Interface. Reading, Mass.: Addison-Wesley, 1992.
 
Tufte, Edward R. Envisioning Information. Cheshire, Conn.: Graphics Press,
1990.
 
Turkle, Sherry. The Second Self: Computers and the Human
Spirit. New York: Simon & Schuster, 1984.
 
Wurman, Richard Saul. Information Anxiety: What to Do When Information
Doesn't Tell You What You Need to Know. New York: Bantam Books, 1989.
-------------------------------------------------------------
 
BIOGRAPHICAL NOTE:
Melinda McAdams is a content developer at Digital Ink Co., a subsidiary of
The Washington Post Company. She has written several articles about online
newspapers and designing information spaces. For eleven years she worked as
a copy editor at publishing concerns including Time magazine and The
Washington Post.
Her e-mail address is mmcadams@well.com.
 
+ Page 90 +
 
--------------------------------------------------------------------------
Copyright Statement
--------------------------------------------------------------------------
Interpersonal Computing and Technology: An Electronic Journal for the
                             21st Century
 
Copyright 1995 Georgetown University.  Copyright of individual articles in
this publication is retained by the individual authors.  Copyright
of the compilation as a whole is held by Georgetown University.  It is
asked that any republication of this article state that the article was
first published in IPCT-J.
 
Contributions to IPCT-J can be submitted by electronic mail in APA style
to: Susan Barnes, Ph.D. Editor IPCT-J,  SBB3007@IS2.NYU.EDU
_