The Nexis Nightmare   

Mistakes that won't die. The same old experts. The endless repetition of anecdotes. Will reporters ever master the computerized morgue, or has it already mastered them?

How many members of the French Foreign Legion fought in Operation Desert Storm?

That seems like a natural enough request for a newspaper editor to ask a librarian. But requesting it 10 minutes before deadline won't win you any friends. Elizabeth Haworth, the Newsday researcher who had been asked to find the number, did the best she could: She phoned the French Foreign Legion, the French Embassy in Washington, even the Pentagon.

No luck. Everyone was gone for the day. Haworth glanced anxiously at the clock. She knew she could probably pull a number – any number – from a news database such as Nexis, but how reliable would it be? Despite her misgivings, she logged on and found a single citation that said 2,000 legion troops had been sent to Saudi Arabia.

She gave the editor the figure with a strong word of caution. "I told him I didn't think we should use it," she recalls.

The editor ignored her. The next day, Newsday stated that 2,000 French Legionnaires were in the Persian Gulf.

"On deadline, any newspaper in the country would have done the same thing – everybody operates this way," says Haworth, now a research consultant for Time Inc. "Maybe the figure was right, maybe it wasn't. No one knows."

Welcome to the Misinformation Explosion.

Fueled by the growing popularity of both commercial and in-house computerized news databases, journalists have found it that much easier to repeat errors or rely on the same tired anecdotes and experts.

That's not to say that mistakes never took on a life of their own before the electronic age, or that the same academics and analysts weren't quoted repeatedly. Sportswriters in particular have a rich tradition of turning apocryphal reports into legends. More recently, in a profile of music historian Nicolas Slonimsky, the New Yorker reported that his "horror of horrors was the inadvertent factual errors that, once born into print, refused to die, and indeed spread exponentially from one sourcebook to another, eternally. They haunted his sleep like vengeful wraiths."

Electronic morgues clearly offer major benefits to journalism, and in many ways they have revolutionized the craft. Their influence is so widespread, in fact, that there has been debate about whether journalism students should be required to learn database search methods as part of their coursework.

In the field, meanwhile, some reporters have become so enamored with resources such as Mead Data Central's Nexis that New York magazine recently noted the database "has begun to develop weird, source-like qualities of its own." (Nexis contains some

2,300 news sources and the capability to reduce 3.7 million pages of data to 110 pages of "hits" in less than 10 seconds.) Often activists or journalists will carelessly cite the number of times a certain word appears in proximity to another word to support theories about the abundance or lack of coverage of particular topics.

Databases such as Nexis, Knight-Ridder's Dialog, Oklahoma Publishing's DataTimes and Dow Jones News Retrieval also have encouraged the expansion of what might be called the Golden Modem. A short list of experts has the potential of becoming even shorter as journalists scan background articles pulled up with general search terms such as "health care" or "gun control." Spy once mocked the practice by quoting ubiquitous political analyst Norman Ornstein in every story of one issue, on subjects ranging from ambassadors to taxi surcharges. The original Spy is dead; Ornstein lives on (he's been cited in at least 80 stories during the past year, according to a Nexis search).

Perhaps the most irritating effect that computerized morgues have had on reporting is the life they provide for errors. Consider the case of Richard Lamm, a former governor of Colorado. He was the politician quoted as saying the elderly "have a duty to die and get out of the way." But that wasn't all he said.

In 1984, Lamm was giving a speech on the ethical dilemmas that might soon be created by the rising costs of health care. At one point he stated, "We've got a duty to die and get out of the way with all of our machines and artificial hearts and everything else like that and let the other society, our kids, build a reasonable life."

As Lamm explains it now, he was pointing out that "people talk about having a right to die like we could refuse... I was just trying to say that death is not optional."

That's not the impression the Denver Post gave readers, however. The daily condensed Lamm's comments into a quick lead that quoted him as saying elderly and terminally ill Americans have "a duty to die and get out of the way." The remainder of the article qualified his remarks somewhat, but the damage had been done.

Lamm's press secretary went to the Post to straighten things out. The reporter had taped the speech, and the editors concluded the quote had been taken out of context. The next day, the paper ran a prominent correction.

If computers didn't exist at the time, that may have been the end of it. Unfortunately for Lamm, they did, and do.

"The corrections move by bicycle while the stories move at the speed of light," laments the former governer, now a professor at the University of Denver.

Lamm still sees the error repeated today, 10 years later, despite repeated corrections. (The Denver Post itself repeated the mistake more than once in the years following the original story before a stern memo was distributed in the newsroom.) As recently as last November, the New York Times corrected an article from the month before that had repeated the distorted quote. At the same time, the Times took the opportunity to point out and fix identical errors in eight other stories from 1984 and 1985 that had mentioned Lamm. Readers might not remember, but Nexis does.

øot all mistakes are as significant as a governor supposedly advising the elderly to die sooner rather than later, but editors and reporters at smaller papers know reader response can be just as intense when facts once corrected get bungled again.

At the 40,000-circulation Sun in Bremerton, Washington, Jeff Brody recalls a misspelled name that slipped into the in-house electronic morgue and couldn't be exorcised. "We spelled it wrong, corrected it, then an intern pulled the original story and spelled it wrong," says Brody, an assistant city editor. "We corrected it again. Down the line, someone pulled the name out again. The problem in our case is that we can't – or don't, I'm not sure which – go back into the story in the database to correct it."

A reporter from an Indiana daily who asked not to be identified recalls a similar incident in which a high school principal resigned his position on an obscure state commission to avoid a possible conflict of interest. "Some reporter misinterpreted this to mean he had left his job at the school," the writer says. "That reporter's story was later used for background. Reporters screwed up the resignation paragraph in three succeeding stories."

Elizabeth Haworth, the former Newsday researcher, says a large part of the problem is that journalists don't show the proper skepticism toward news databases. "People think because it came from a computer's mouth, it came down from the mountain on tablets."

ýdds Kathleen Hansen, a University of Minnesota professor who has studied the effects of electronic databases on newsgathering: "Reporters are supposed to be so hard-bitten and cynical. How can you simply swallow what appears on these systems without applying the same criteria they do to everything else?"

The mistakes are there, but who is responsible for them? Ultimately, the press.

Database vendors rely on newspapers and magazines to submit corrections or updated versions of stories, but some aren't as diligent as others. For the most part, if a correction doesn't make it to print (say a misspelled name that no one complains about), it's never going to appear in a database either.

Some say editors need to realize that the computer age has given more weight to errors. "Before electronic databases, if you pulled some number out of the morgue and it was wrong, the only people who saw it were your readers," says Haworth. "Now, it gets sent out to tens of thousands of users."

The media's rush into cyberspace hasn't helped matters, says librarian Sylvia Frisch of Minneapolis' Star Tribune. "There are places that put data [on-line] the first day they get it," she points out. "How you have any sense of quality control is beyond me. Some librarians just give up."

Some newspapers are very persistent about chasing down database mistakes, but they hardly comprise the majority of publications on-line, says Donna Willmann, a regional director at Dialog. Dailies such as the Chicago Tribune, the Memphis Commercial Appeal and Minneapolis' Star Tribune have been attaching corrections to their electronic stories for years. Others, such as the Charlotte Observer, Knoxville News-Sentinel, Orlando Sentinel, and Philadephia Inquirer and Daily News, have gone a step further by correcting minor errors in their electronic stories even if the errors aren't corrected in print.

Willmann believes that despite some shortcomings, databases are still the most efficient way to link corrections to stories. "You can blame electronic databases on the proliferation of errors," she says, "but if you're reading the newspaper and three days later there's some correction to a story you may not have read or even remember, who says that's a good way to do it?"

Experienced researchers ýtress the importance of understanding how databases are organized so searches bring back corrections and follow-ups as well as targeted stories. It's also a good idea to understand how newspapers work, says Nora Paul, library director at the Poynter Institute for Media Studies.

For instance, it's crucial to realize that reporters don't always follow up yesterday's story, or if they do, it might not always make the paper.

Paul recalls one incident in 1990 in which a restaurant manager had been falsely accused of having AIDS. A reporter wanted to know if this had happened anywhere else.

"I ran a search and came up with a single wire story," recalls Paul, who conducts seminars in news research. "The story told of a West Virginia florist who had been falsely accused of having AIDS and had been harassed. He said he was facing bankruptcy."

The story seemed odd, largely because it was so dramatic. Why hadn't any newspapers picked it up?

Paul decided to do a second search in a larger database, this time using the florist's name. Sure enough, a follow-up revealed the florist suffered from mental health problems and had admitted fabricating the harassment claims.

Elizabeth Haworth says journalists should remember that most newspaper stories are reported in increments. "At least once a year at Newsday we'd get a call from someone asking how many Marines died [in 1983] in the bunker in Lebanon," she says. "If you go into a database to get the official number after the attack, you're missing all the residual deaths over the next two years. When you ask for a database search, you have to know what you're asking for."

The bottom line: Your own reporting might be stellar, but what about the next guy? Says Michael Berens, a Columbus Dispatch reporter who used news databases to document the existence of an apparent serial killer: "The databases don't reflect whether the reporter was juggling six stories that day and dashed off a few graphs on a death, or if a small paper simply rewrote the cops' press release without checking it. These services are a jumping off point; they should not be the final destination."

As databases have grown, reporters have discovered they can be great sources for sources. What better way to find media-tested experts to provide pithy sound bites?

Unfortunately, journalists can become addicted. "A lot of times," says Linda Henderson, library director at the Providence Journal-Bulletin, "I have to tell a reporter to go out and make some phone calls."

The Indiana reporter who asked not to be named recounts how, despite a large population of Serbian Americans in the area, "four poor suckers have been called again and again because their names are in our electronic library next to words like 'Serb' and 'war.' "

Jim Petersen, a senior staff writer at Playboy who uses Nexis regularly for background, notes that the database quickly reveals the experts who have become "quote whores" on any given topic – people he tries to avoid. The same can be said for other resources that reporters rely on for profiles or features: Petersen recalls a search he did after being assigned to write about Cancun, Mexico, in which every travel writer seemed to have eaten the same dish at the same restaurant.

Such electronic recycling has become widespread. During his last days as president of NBC News, Michael Gartner recalls asking the library to find all the articles written about him after the 1992 GM-"Dateline NBC" fiasco, in which the newsmagazine admitted rigging a pickup truck to catch fire on impact. Using Nexis, the librarians did a quick search and came up with 600 citations.

Gartner, now editor of the Daily Tribune in Ames, Iowa, says that figure startled him, because he only recalled speaking to five reporters. Looking over a few of the articles, he was struck by two things. First, many had been written by reporters at local newspapers. "Why in God's name were these reporters spending time in Nexis writing about me instead of writing local news?" he asks. Second, most repeated the same specific details, hinting they had relied on a common source. Gartner suspects Nexis played a key role.

Norman Ornstein, the American Enterprise Institute scholar who has been asked to comment on just about everything, says he's actually had fewer calls from reporters since Nexis became a fixture in many large newsrooms. In part, he believes that's becauseódatabases have broadened source lists rather than limiting them. Nowadays, he says, an academic or analyst needs only to be quoted in print once to get his or her name in the pool, ready to be found by the next journalist diving for experts.

Scholarly research backs him up – somewhat. At the University of Minnesota, Kathleen Hansen and three colleagues found that while database technologies can make it easier for reporters to find new voices, "the same types of sources, representing the same institutional and social power structures, keep showing up." The team concluded that other forces, such as journalistic training, the social position of reporters, deadlines and staff shortages "still seem to work against a broadening of the 'news net' that is cast."

Of greater concern to Ornstein is what he sees as the emergence of an electronic activism among journalists. He fears reporters are deciding what a story means, then using databases to identify sources who will tell them what they want to hear.

"It has become a cover for journalists wanting to express their opinion," he says. "It used to be reporters would call and we'd talk for 35 to 40 minutes. We'd talk about an issue, we'd talk about the background, we'd talk about books to read to really understand it. Lately, I get more calls from people who just want a quote."

While electronic morgues clearly have their pitfalls, many journalists are quick to qualify their criticisms with descriptions of news databases as lifesavers.

In Bremerton, Jeff Brody points out that despite occasional errors, the Sun's in-house database "has made our reporting more complete and sophisticated" by providing instant, detailed background. "Before, clips of every local story were stored in packets by date and catalogued on 3-by-5 cards," he says. "Only the librarian found the referencing system logical. "

Mark Stencel, a reporter at the News & Observer in Raleigh, North Carolina, who learned his search techniques at the Washington Post, recalls an instance where columnist David Broder asked him to find the cost of the Pentagon's infamous overpriced toilet seats. A database search showed that the figure had risen wildly over the years, he says, but "most of the evolution occurred in the years before Nexis became such a key newsroom resource." Without the ability to see instantly that the figure was not etched in stone, Stencel says he might have been tempted to give Broder the first number he stumbled across.

As a source for background material, Nexis is a charmer. During President Clinton's first nine months in office, reporters used electronic libraries to check up on his campaign promises – many of which, to the consternation of the White House Press Office, had gone unfulfilled. That prompted George Stephanopoulos to complain that the White House was being held hostageby reporters with access to Nexis. (The other side of that coin: Don't make promises that end up in print – or if you do, keep them.)

Finally, electronic databases can be invaluable in chronicling trends that are easily overlooked during the hustle and bustle of daily news deadlines.

One of the better known investigative series to rely extensively on electronic libraries was completed in 1990 by Michael Berens. Using Dialog to get started, the Columbus Dispatch reporter found similarities among the unsolved deaths of nine known or suspected prostitutes at various Ohio truckstops. Last year, in a second investigation that took eight months, Berens compared federal statistics about fatal crashes during police chases with newspaper accounts and discovered the government had missed hundreds of deaths.

Despite his successes, however, Berens approaches databases with a strong dose of skepticism. In both of his investigations, he discovered major errors in the electronic clips he retrieved – but only after checking the articles with their original sources.

In other words, after he'd finished his reporting. l