We're a small statistical agency employing fewer than 60 people. About two-thirds of the staff are devoted to analyzing numbers and writing or producing publications. On average we publish a document a week--a book, bulletin, or fact sheet. Our budget goes primarily to collecting data, and it's not a large budget.

And the budget starts and ends my brief story today. In 1993 we incurred a postage bill that had ballooned to nearly the size of Minneapolis. The publication dissemination budget line had to be shored up, and we had to take a hard look at our future.

The mailing costs were the negative side of our success. We had a great, aggressive clearinghouse that shared a commitment to our legislative mandate: We were to provide as much accurate, on-time statistical information about crime and criminal justice as possible and we were to do this at no charge to the user.

The debate student who called got a sheaf of reports about capital punishment. The journalist received personal attention and a thick packet of materials. The neighborhood watch captain went on our victimization mailing list. The thousands of agencies and men and women who contributed to our surveys and censuses received the books that contained the answers they had given us; these were our only legal payoff for cooperation, and it was the least we could do.

But we had to cut back. We killed the glossy quarterly summary of publications that at its height had gone to tens of thousands of homes and offices. We made readers order each of our thick books of tables rather than receive them automatically. We substituted executive summaries for the numerical behemoths we previously shipped to everyone. We scrimped on printing, and we began to think about digital libraries.

The BJS Electronic Library was what our planning group called the public access facility that we wanted to create. The more visually oriented among us nominated Digital Gallery, but they lost. The concept of this BJS place was a walnut paneled reading room. Overstuffed leather chairs. Sherry at 4 o'clock.

When we determined that a couple of gigabytes would store our back list on-line, we recognized at once that our electronic library would not be the Alexandrine national library that Transportation or Patents or Energy must maintain. There were ways we could have dumped loads of numbers onto the Web site, but from the moment of conception the BJS library was to be a place where the user could find the right document right away. Unable to afford or even to find the search engine that we wanted, we were committed to effective design.

That early pledge to ease of access kept us on course, but not all our best-laid plans ended so happily.

In those days, excitement surrounded Internet components named after small burrowing animals or comic book characters. What we wanted for our digital facility, we decided, was a gopher site. Months passed. By the time we actually dug our rodent den, few customers wanted to use it, and it now stands forlorn.

This last example of planning leads us to the puzzling BJS First Rule for Living on the Information Superhighway: What we plan we ultimately don't want to reach so we should plan really carefully. This rule is a kind of variant of the curse: "May you get what you hope for." Lurching ahead without a well-established, thought-out plan is foolish. In today's technological terrain, however, resolutely following a considered plan to its bitter and dead end is equally silly.

Here I must interrupt my story to tell you of some of the advantages that BJS enjoyed. We did not have to build a electronic library fit for the Trump Palace because we helped to support organizations that were also serving our customers. BJS can be seen as a kind of network.

I have already mentioned the clearinghouse managed by Aspen Systems: the National Criminal Justice Reference Service (NCJRS). Besides having a corps of experts responding to calls on 1-800 lines, NCJRS operated a bulletin board for distribution of our reports.

Three other national or international criminal justice professional groups that we support in part were also offering computer bulletin boards. So, for chatty folks and for the technically proficient, willing to set parity bits and be patient with their 14.4 modems, BJS was at the ready.

For the lovers of encyclopedias and for reference librarians everywhere, The Sourcebook of Criminal Justice Statistics joined us on-line from Albany, New York. BJS customers with file readers able to copy a table to the clipboard could import contents directly into their documents.

In the thin atmosphere of Ivies and the Big 10 and major research universities, BJS was providing data files for analysis through the National Archive of Criminal Justice Data, at the University of Michigan. Member universities could acquire our data on 9-track tapes, and their students could swing mainframes to crunch our survey results.

From my little catalog of dispersed resources, you can see that we were still lacking. We wanted to be omnipresent, with the apposite statistic at just the right moment. Instead, in the railroading lingo of the pre-asphalt epoch, we were sitting on a siding far from the trunk line of information.

At the time, all about us, major Federal agencies were throwing what seemed to be terrabytes of information across the telephone system. The BJS task force members could not even access those agencies hurling all those data because the branch of Justice where we reside had not yet gained World Wide Web service. Our capable task group chair, Marianne Zawitz, who has done the most to make the electronic library a reality, would surf the Web at home and then bring us reports, like a spy returning from the unknown territory we were proposing to enter.

But in on-line capacity we at BJS weren't much different from the people whom we try to serve. We have an enforced modesty and frugality. The criminal justice community and people wanting criminal justice statistics do not belong to the Sharper Image set, living fast on the cutting edge. I suspect it's as much a question of costs as desire, but if you live in a house fitted with Shaker furniture, pretty soon you come to appreciate the function and good design of simplicity.

Our director had a colleague who was living in Mexico and dialing in with his 1400 baud modem on his laptop with the graphics turned off. The constant question while we designed the first pages was what would our Pueblan correspondent see if he accessed BJS on-line? Such a consciousness of limitations continues to guide us in how and what we offer our digital users.

About a quarter of the people who ask for our publications are State or local government employees: the policemen, sheriffs' deputies, and budget analysts; another quarter are in school as students or teachers; another quarter are members of organizations--businesses, community groups, foundations; and the last quadrant are a mixed group--the general public, the media, Federal employees.

Interestingly, less than 10% of the total audience--the journalists and our bosses on the Hill and in the White House--captured and still capture a large part of our attention, but the Web has certainly reduced the distance between the President and a third grader preparing a report on crime in his neighborhood.

I try to speak authoritatively about the borrowers from the BJS electronic library, but don't let my tone fool you. We send out about three-quarters of a million paper-and-ink reports a year, and on average nearly a thousand people a day visit our site, but I still have only the dimmest notion of who our readers are and what they want from us.

This is BJS's Rule Two for Living on the Information Superhighway: We admit our ignorance, we try to correct it, but we don't let it prevent us from acting.

One valuable source of information at the pre-Web-site stage was a reader survey sent to the whole mailing list. From about 36,000 forms sent, we received almost 6,000 responses. In fact, three and a half years after the survey, I continue to receive mangled questionnaires.

In sum, our readers told us that they belonged to the fax generation. Two-thirds of them lived at the corner where paper-and-ink and electronic impulses cross, and when we acted on that knowledge, purchasing a fax on-demand service, we made one of our wisest decisions ever in terms of service and workload. And when we substituted the mass-fax for a 32-cent mailing of newly released reports, the savings were considerable. As you know, the fax often smears illustrations, distorts fonts, and renders some documents into murky blurs. But our readers love it. They like the convenience and speed of getting precisely what they want. It's their technology.

Hence, a corollary to Rule Two just cited: Knowledge makes for better decisions than ignorance, always.

Our readers also told us that a healthy majority--about two-thirds--had a computer. But in 1994 there were a lot of lonely computer users: too many of our potential library browsers owned an ancient version of WordPerfect. Fewer than 1 in 10 had a modem. We understood from our survey that our digital library would not replace the paper-and-ink publications. Rather, we would expand our audience.

As progress unfurls we discover that we were right and wrong. We captured new readers, especially from the White House Web site because we provide the crime statistics for the Social Statistics Briefing Room. Our old audience, however, is adding modems to its computers. We know this because, like air traffic controllers, we guide them over the phone to the requested links on the Web. They appreciate Marianne Zawitz's genius at keeping the site simple and direct, with none of the labyrinthine embellishments that are the curse of World Wide Web excesses, but they still get lost. Tell you the truth, so do I sometimes.

Our first metaphor was an Electronic Library. Walnut and leather and sherry.

But there were issues that never belonged in a library with a antique globe and vellum bindings, and we had to address them. How were we to account for documents that we updated on the Web without a paper trail? We decided to mark all of our Web documents where revisions occur and to record in a database each revision made. We anticipate that someday we will be storing all site contents on CD-ROMs.

And how are scholars and journalists to cite what they find on our Web site? We are committed to constant maintenance and improvement. While we will try to preserve URL's to specific documents, sometimes both the equipment and the design conspire to make bookmarks in perpetuity an impossibility. So we are suggesting that citations include the date of a visit, our domain and site names and the site category, omitting the specific document designations that are likely to be sacrificed. We are keeping a record of Web document names for anxious researchers in need.

While working in the library and still wrestling with the budget, we came to a new understanding of what we wanted our Web site to be: a digital library and a digital hothouse. Glass and light and air and the smell of growing things.

According to my best reading of the Web analysis of our site, our visitors want the simplest versions: given the choice between a PDF file and an ASCII file, they choose ASCII. Despite the sacrifice of readability and context, the typical Web user wants the easiest document to snatch, though some are probably checking the ASCII contents before they go through the more elaborate procedure of accessing a PDF.

I don't applaud the preference for ASCII, but I am happy that we offer the choice. I am also happy that our customers have our report with the Department of Justice press release and last year's version of the report and access to the original data at Michigan and dozens of links that can add meaning to our report. All this our customers enjoy as long as we don't stand in their way.

Here I feel another BJS Rule coming on: Number Three for Living on the Information Superhighway: The spirit behind the Henry Ford option for Model T color--any color you want as long as it's black--does not apply to the Internet. Only superlative selections at supermarkets on the superhighway will survive.

That rule brings us to the BJS Web site today. We open our new Electronic Greenhouse in late April or early May, and those familiar with our site will find the same handsome simplicity and deep lists. All the paper and ink publications, as Adobe and ASCII documents, will be offered. But more recent, high-interest information will be placed in HTML on the Web pages. To find key facts, a Web visitor will not have to access a document first but can go directly to the desired statistics.

There will also be the new Internet-only publications. From the start we have had spreadsheets for the larger tables published in paper and ink. Now we have spreadsheets too large to fit on any paper smaller than half a basketball court. The interested user can copy or download the data of interest and make his or her own graphs or tables.

The Sourcebook is on its way to losing the book part of its name as it becomes a site that constantly updates contents as they are available. At paper-and-ink publication time the Source staff will copy the PDF files into a CD-ROM, and both the CD and the paper version will be distributed.

Remember when I noted that the budget would reappear in our story? In the past we published documents of limited interest that contained every detail of a census, spilling like an oil tanker across a thousand pages. We cannot afford to publish such documents anymore, and so their contents are going directly to the Web without the dusty detour to paper. For the readers without Internet access, we are offering to fax or to mail all the numerical tables that they need. It's a case where the saving of a quarter-million dollars inspires us to provide individualized services.

To help readers and archivists keep up with our publication series and the welter of old publications, the new site will offer a common-sense guide that will take the user to the most recent report in a series.

After the greenhouse is constructed, I expect that we will start using more CD-ROMs to record the data and publications for our customers and posterity.

I also anticipate that we will soon have a database in place to allow Web visitors to create unique tables. I believe it's called "on-the-fly."

Soon we will begin to use the Internet to survey the prosecutors' offices and police stations and other data sources. Our Internet site will receive as well as send information.

After all these changes the Greenhouse metaphor will have lost its value, and perhaps a Daedalus image will take its place.

This conclusion leads me to the last BJS Rule for Living on the Information Superhighway: The trip is better than the arrival, so it's best to be always leaving.