One dictionary describes mistakes as those decisions made from faulty judgment, inattention, or inadequate knowledge. If we are making mistakes in electronic publishing, it is because of the third reason, inadequate knowledge. I would not even call those mistakes but bumps on the road to discovery. Our job as publishers is to make mistakes based on inadequate knowledge now so that we will gain the knowledge that will help us avoid making those mistakes again.

While creating Academe Today, The Chronicle of Higher Education's online service, I gained a great deal of knowledge, often by making mistakes.

A Brief History of the Times

In 1990, The Chronicle began exploring putting the weekly newspaper into electronic format. With the growing presence of desktop computers on campuses, the emergence of campus-wide networks, and the blossoming u se of e-mail and the nascent Internet, the time seemed right. I had been a reporter at The Chronicle for eight years covering information technology before I was asked to head the project. I accepted eagerly, convinced that it represented the future. With in months I had stopped writing news articles and was working on the art of the memo.

By 1991, we had the full text of The Chronicle up on the University of Southern California's campus-wide network. In 1993 we put a guide to The Chronicle and all the job ads up on the Internet in Gopher format. In 1995 we made the full text of The Chronicle available to subscribers on the World Wide Web, and we were producing a daily higher-education news service delivered by e-mail.

I led that project from the time it was a one-woman shop to its growth to a division that had twenty-three employees dedicated to various aspects of electronic information — nearly 20 percent of the staff at the time.

If you haven't seen the Online Chronicle, visit it at http://www.chronicle.com. Only a few links from the front page work for non-subscribers: Try Information Technology or Colloquy to get a flavor of the rest.

The USC Project

When we first put The Chronicle on line, we did so in a joint project with the University of Southern California. USC had a campus-wide information system, USCInfo, run by the library in conjunction with the computer center. USCInfo offered 14 bibliographic databases plus the online catalog to users both on campus and dialing in from home, so neither the network nor the search methods were novel. For The Chronicle, that gave us a test that would eliminate the Hawthorne effect where the novelty of being able to access information by computer would affect usage.

(n.b. We called it "The USC Project." The folks at USC called it "The Chronicle Project.")

I knew it would be a challenging job to convert our files to a searchable database. Articles were — and still are — written, edited, marked up, and typeset on an Atex system, a high-end newspaper-production system.

The big problem, I thought, was the "modes," the Atex tags that told the printer what typefaces to use, modes like italic, bold, small caps, font faces, and spacing. Like most newspaper publishers, The Chronicle had elaborate rules about where to use those modes and why. Small caps, for instance, were used for acronyms, sans serif for bylines.

However, what turned out to be the problem, and the cause of mistakes (or learning experiences), were the modes of thinking.

Adapting to change means thinking in ways we haven't thought before. Our natural tendency is first to describe a new situation in terms we are familiar with — our current modes. It's only after those models don't quite work (and we know that only by e xperimentation) that we create new modes. Alvin Toffler describes the initial process as "straight-line thinking," the kind of thinking that allows a statistician to look at the decline in Ph.D.s awarded in physics over the last ten years, and to conclude that in another dozen years there will be no Ph.D.s awarded in physics.

So when I started the Online Chronicle project, I thought of the online Chronicle as an extension of the paper Chronicle. I thought our online readers would be just like our paper readers, only somehow more so. I thought that by making The Chronicle electronic and therefore searchable, it would become so much more useful.

I believed that:

• Our biggest problems would be technical.
• We would be in danger of losing our market.
• We would make money.
• We could count on USC's librarians to help us understand electronic information delivery.

The reality was somewhat different.

Those Atex modes we used in markup turned out to be both a big problem and a great asset. They were tough to convert, but they helped me to identify the component parts of the articles that would make formatting, searching, and organizing our information easier. I tracked down all the modes we used for headlines, kickers (the short titles that are sometimes used above the headline), decks (the longer titles that are sometimes used below the headline), bylines, captions, subheads — all the layout components of a Chronicle article. With help from our technical staff, I wrote Atex programs to convert the Atex modes to ASCII tags, and later I hired an outside company to write Perl programs to further convert those ASCII tags to database formats or HTML.

Back in 1990, the conventional wisdom was that we could offer our electronic version to any institution with a campus network, sell a site license, and make money. While I doubted that was true (after all, The Chronicle then had nearly 90,000 subscribers, and there were only 3,200 colleges and universities), I had no evidence to prove otherwise. The Chronicle had to go on line at USC to see what would happen.

Site licenses may not have been the answer, but I hoped that we would make money by making The Chronicle more widely available in higher education. After all, The Chronicle had fewer than 90,000 subscribers, but the potential market was twelve million students and three million faculty and staff. If we could reach just one percent of that market, we'd increase our subscriber base by more than half. And the advertising potential was enormous. We had to try making The Chronicle available electronically to see what new readers we might attract.

Lessons Learned

I worked with the Center for Scholarly Technology at the University of Southern California, the brainchild of Peter Lyman (now the librarian at Berkeley). CST brought together some of the best people from the library and from the computer center to tackle the issues of electronic information delivery. I found that we were all learning together: Librarians and computer people had not cornered the secrets of information delivery.

Some of the problems we had to solve together included:

• Abstracts.

  The librarians raised the question of abstracts, pointing out that their researchers were used to them.

  As none of us wanted to write the abstracts, we decided to use the lead paragraph of each article as an abstract stand-in, letting people search just in the lead if they chose and then see just the lead before deciding to call up the entire article.

  Users hated it.

  They pointed out that Chronicle articles were short enough to dispense with that intermediate step. And I believe that a full-text system that doesn't force people to wait for an article to be delivered, plus articles that are better written (no need to read a dozen pages before getting to the point) obviates the need for abstracts.

• Searching.

  The BRS Search system that USC used offered full Boolean searching, and the librarians suggested we take advantage of it. We put the Boolean instructions at the bottom of each search screen.

  Users ignored them.

  Most users, we discovered, searched for a single word or phrase, looking for an answer rather than the answer. We finally simplified the search options on the screen, keeping them available to those (mostly librarians) who knew the magical incantations that called them up from the bowels of the system. The users were satisfied.

• Corrections.

  The few mistakes that slip into The Chronicle are forthrightly admitted in the first pages of the newspaper. I thought that it would make far more sense to attach each correction or clarification to the original article. The librarians voted for appending the corrections at the beginning. I decided to simply correct the text. If a name was spelled wrong, or a date or fact incorrect, I would change it and not point out that a change had been made. My belief was that we were best serving the interests of our readers that way. I wanted to make it easy for readers to find correct information, and now, for the first time in The Chronicle's existence, we had the opportunity to go back and make it right.

  In one case, The Chronicle re-ran a long table of figures because a respected higher-education organization had given us wrong information. I pulled the incorrect table out of the electronic files and substituted the correct one. Scholars who study mistakes won't find much to work with in The Chronicle archives, but other users won't find mistakes, either. (Clarifications were different. When The Chronicle explains a subtlety that may not have been clear — or may have been left out — "fixing" the mistake actually changes the article. Instead of fixing it, we appended clarifications to the end of articles.)

• Fact Files.

  The Chronicle's lists of facts and figures about higher education — the faculty salaries, the library holdings, the campus-crime statistics, and the minority-enrollment figures, to name a few — are some of the paper's most popular features. Administrators told me that they carefully copied those Fact Files into spreadsheets, to allow them to manipulate the numbers to wring out more information.

  I saw that as an opportunity, and decided to offer the Fact Files in comma-delimited ASCII to make it easier for readers to load the information directly into their spreadsheets.

  That conversion was difficult — and unnecessary. When we put the spreadsheet-ready Fact Files up at USC, no one used them. Our research showed that no one wanted all the data in a Fact File, just the data about comparable institutions: those in the same state, or the consortium, or the same Carnegie classification. It was actually easier to enter the information about that handful of institutions in a spreadsheet than to download information about all institutions and delete the noncomparable institutions.

• Layout.

  I knew we would be losing information by giving up The Chronicle's page layout. Stories and sidebars had to be linked manually in the conversion process, and a reader's interest could not be captured with a compelling photo. Readers familiar with The Chronicle missed the layout cues (not to mention the cartoons!).

  What surprised me was that online Chronicle users who were not familiar with the paper preferred the online organization. They actually said that they found the print version confusing and didn't know how to find information in it.

• Box scores.

  Our research had its lighter moments. At USC a cross-database search often turned up Chronicle articles, leading some students who had never heard of the paper to delve further into its offerings. But they were often disappointed and disillusioned by the difference between what they expected and what they found. One student sent me e-mail complaining that The Chronicle couldn't be a higher-education newspaper because it didn't have the scores of the recent Final Four tournament. In fact, t he student noted indignantly, it didn't have any sports scores. He dismissed The Chronicle as a credible source of higher-education news.

• Copying.

  Like most publishers, we feared for our intellectual-property rights on the Internet. The ease of making exact copies and broadcasting them widely had us sweating. However, I found very little copyright violation, and most of it was by people who were featured in stories who posted the article on their personal or departmental Web sites. When I contacted them to tell them, gently, that they needed to get our permission to post the articles, they quickly took do wn the article until permission came through. (The Chronicle gives non-profit institutions free permission to make reasonable numbers of copies; for-profit companies are asked to pay a small fee.)

• Advertising.

  While I expected the Internet to be a great vehicle for advertisers, I thought we would have to sell them on the idea. Instead, soon after our Gopher version was up several advertisers from the paper approached us asking to post advertising on our site. I was very pleased. "What did you have in mind?" I asked. "We don't know; you tell us," the advertisers answered. Rather than persuade our advertisers, I had to help them design their online campaigns. We had become the information-delivery experts online, even for advertising.

• User categories.

  The Chronicle funded research at USC to help us evaluate use of the system. Using data logs, surveys, and interviews, the researcher gave us a better picture of use and interest than we might have gotten any other way. In one study he identified two distinct groups, Power Users and Information Seekers. The former were usually subscribers, and they read The Chronicle online as soon as it was posted — often before they received their paper copy. They browsed the headlines and dipped into stories of interest. And they returned to the online Chronicle during the week to search for information. The researcher identified fifty Power Users on that campus of some 30,000 students and 12,000 faculty and staff.

  Information Seekers, on the other hand, were often unfamiliar with The Chronicle and came across it in their database searches or by choosing it from the main USCInfo menu. Some never returned to it; others came back after a few weeks, and used it occasionally, usually when looking for information on a specific topic. The researcher identified about one thousand Information Seekers, a user group I had not anticipated, as they were not typical Chronicle subscribers.

• Chronicle use.

  To my surprise, the largest group using The Chronicle on the USC network was Chronicle staffers. Reporters and editors used the system regularly, to get background for stories, to check style, and to confirm names or titles. Moreover, they appreciated the ease of use of a database 3,000 miles away; the Atex system allowed a clunky search, but few people at The Chronicle ever used it because it was so slow and awkward. That heavy Chronicle use continued and grew with the move to the Internet and the World Wide Web.

Mistakes Made

I think we missed some opportunities because of the straight-line thinking that insists The Chronicle is a newspaper, even online.

• The jobs database.

  The Chronicle's classified advertisements are an important database of information about jobs in higher education. Yet The Chronicle replaces its job ads when a new issue comes out; older classified ads are not available. I argued that The Chronicle should build a searchable database of position announcements. Scholars interested in higher-education hiring trends would appreciate access to that rich lode of information. And I believe that Chronicle reporters could write some informative stories about tr ends in higher-education hiring. Yet when I raised the issue I was told that because The Chronicle does not have all job advertisements — some disciplines are more likely to use scholarly societies' publications than The Chronicle to advertise po sitions available — it is not a reliable source of information. I believe that historical comparisons of partial information yields useful results, but I lost that fight.

• Library subscriptions.

  The Chronicle offers online access only to individual subscribers. True library subscriptions, which would allow librarians to offer The Chronicle directly to library patrons are not available. To a large extent that is understandable: The Chronicle's market is higher education, and by selling the online version at a lower price per user as a site license, the company is risking its business.

  However, the USC experiment showed that there is an untapped market for The Chronicle: those who are interested only in the searchable archive, the Information Seekers. The Chronicle's inability to see itself as anything but a newspaper — with the em phasis on news — has blinded it to the value of its information as a historical database.

  By selling access to The Chronicle's archive to libraries interested in making it available on campus information systems, The Chronicle could tap a new market. In return for a delay of six weeks — or even six months — libraries might be willing to pay a (considerably lower) price for access. By refusing to see itself as a historical database, The Chronicle is missing an interesting, and possibly moneymaking, opportunity.

• Lost information.

  I am particularly sad that I didn't find a way to retain all the information in the Atex system. The Chronicle's Atex has a rich complement of foreign and special characters that are used in the newspaper. For the online version I chose transliteration, representing foreign characters with the closest ASCII characters. If I had it to do over, I would have found a way to at least retain indication of the foreign characters in some background mode, in preparation for the time that we could represent them in a networked environment.

• Resource needs.

  The Chronicle was not prepared for the drain on staff resources. Studies show that it takes some 3,000 hours to support 5,000 users with passwords, and that doesn't count the time it takes to help people find things in an unfamiliar database. The Chronicle had to assign two people to handle in-house support when it offered Internet access to subscribers.

  That fact should have been obvious from a research project involving a graduate course in library science. Those library students didn't read the instructions and so missed some obvious features of the system. If library students didn't read the instructions, we should not have expected people less familiar with online searching to do so.

Pioneers Are the Ones with Arrows in their Backs

Three months before we first went online at USC, I wrote in my diary:

"We are exploring what people want to do with full text in ways we never thought necessary when we started this. Oh, if only we had had some pioneers to follow! I feel like Gutenberg, trying to make the technology work while missing obvious features like page numbers. We are constantly trying to make decisions between ease [of use] and power, and attempting in every case to offer both. We are trying to build from both ends: recreating The Chronicle while recreating USCInfo (remember, we thought we'd just have to strip out the formatting instructions and put in a few markers...)"

"So I am disappointed at our slow progress, but at the same time recognize that we at least are not like the pioneers who sometimes find rivers they cannot ford, mountains they cannot climb. We are more like pioneers in the Great Plains, facing prairies much vaster than we ever suspected, with mountains at the end that seem never to get any closer, as we slog on day after day, digging the wagons out of ruts, caring for sick horses, hoping we will find water before dark."

The Chronicle experiment and its extension to the Internet and the World Wide Web taught me many lessons, the most important of which is the need to try. The Chronicle was early in moving to digital publishing and had the advantage of time and resources to learn what was possible and what worked. While we know a great deal more, there is still more to learn. We haven't reached those mountains. We've got to keep working at it, learning from our mistakes, until we truly understand both the nature of ele ctronic information and the marketplace it creates.

Wagons Roll!

Judith Axler Turner was the director of Electronic Projects for The Chronicle of Higher Education, heading the effort that began in 1990 to put The Chronicle on the Internet. For seven years prior to that she covered information technology for The Chronicle. She has had a nationally syndicated weekly computer column, weekly ski column, weekly Washington-life column, and a regular sewing column. She was a political reporter for publications ranging from the New York Daily News to The New York Times to the National Journal. She has B. A. from Russell Sage College, and three wonderful, talented, good-looking children from her husband, Lester. In her spare time she is the editor of The Journal of Electronic Publishing and likes to make speeches and presentations on electronic publishing.