the seductive technology

at 8:20 p.m. on May 9, a sleeping 3-year-old boy was whipped from his mother's lap by an exploding tire tread and dragged to his death beneath the wheels of a bus.

The bizarre death of Ramon Prado provided an occasion - an excuse, really - for The Orange County Register to publish results of its seven-month probe of bus safety in California.

The story showed that the California Highway Patrol routinely fails to meet its legal mandate to inspect buses. But as the story also noted, California's buses have racked up an enviable safety record. And even if CHP inspectors had found and inspected the bus on which Ramon Prado later died, they could have done nothing about a floor so thin that a tire tread could breach it; the floor met federal standards.

So much for seven months work. The story, and the endless research that went into it, illustrate what I call seduction by technology. Entranced by the siren song of high tech, I pursued a story long after my instincts told me it wasn't there. Most articles in Uplink tell of triumphs. This is a story of failure and the four journalistic myths that made it possible.

It began in July 1991, with a bus crash near Palm Springs that took the lives of seven people. The Register immediately began casting about for a way to investigate bus safety. We discovered the CHP's Management Information System of Terminal Evaluation Records, aka MISTER.

It sounded perfect: computerized records on 90,000 bus and truck companies, including every traffic ticket, every accident and every CHP inspection of bus or truck terminals. CHP's reaction made the database seem even better: They demanded $2,800. Over the next three months we argued the price down to $81.18.

Enter Myth No. 1: If a bureaucrat hinders or refuses access to public documents, those records must be explosive. That might have been true, sometimes, in the days when governments kept its records in manila folders. Today, when agencies routinely keep gigabyte-sized databases, it's absurd to presume that custodians know what they're hiding.

Under the spell of Myth No. 1, I was easy prey for Myth No. 2: Gather enough facts and you're bound to find a great story. All reporters play hunches, looking under rocks for something odd. With computers comes the temptation to look under every grain of sand on the beach.

That is what I did with MISTER. Had I faced a similar quantity of paper records, I instantly would have turned away for a reality check. I would have called some experts, read some reports and chosen one or two angles to explore. Instead I plunged into the database with all the direction of a near-sighted man on a foggy night.

"How many people die on buses?" I asked the database. Very few. "How often do buses crash?" Not often. "What does 'often' mean?" Good question, since even CHP doubts the mileage numbers in MISTER.

One day I extracted a list of multiple-injury bus accidents. The heaviest toll by far, more than 50 injuries, was for an accident a year earlier within a few miles of the Register. I could not find a word about that accident in the paper. A local CHP flack looked it up for me: The bus was ferrying prisoners between court and jail; most of the "injured" were inmates complaining of whiplash.

Myths No. 3 and 4 made an already bad situation nightmarish. I suspect both myths are common among computer-assisted reporters.

Myth No. 3: Databases are the perfect reflection of the real world. Think about that one for a second. Intellectually, of course, no one believes it. But emotionally we tend to invest databases with a God-like aura of omniscience. They seem so complete.

In fact, however, databases merely record what their keepers meant them to record, to the degree their keepers can get the information.

It's easy to lose sight of those limitations. MISTER listed hundreds of companies that no longer existed. It omitted bus and truck accidents on local roads, unless police went to the trouble of informing the CHP. MISTER's record of hazardous spills abruptly stopped in 1990 when CHP "temporarily" stopped entering the data.

Before I ever used a mainframe, I was a devotee of Myth No. 4: Mainframes are big, fast and unstoppable - the Incredible Hulk of the Information Age. The myth is correct on one point. Mainframes are indeed big. So was Tyrannosaurus Rex. I thought I needed a mainframe to handle MISTER; at 160 megabytes, it would have stretched the capacity of my PC and overwhelmed XDB, the database program I use. It would have taken weeks to get a few simple answers, I thought.

So I used a mainframe instead. And it did not take weeks to get simple answers; it took months.

We lumbered along, the Register's information services department and I, for three months before I framed my first query to the MISTER database. It took time to install the hardware linking my PC with the mainframe, time to teach me how to use the link, time to load MISTER and then to load it again when the first layout didn't work.

Then the real problems began. Contrary to its legend, I found the mainframe balky. I would ask what I thought was a simple question and then wait and wait - until an irritated computer tech would call, asking why I was consuming 90 percent of the central processing unit's capacity.

I would have to ask four or five questions on the mainframe where one would have sufficed on the PC. And after other users began to complain in droves about that crazy new user monopolizing the computer, the high priests of the mainframe restricted my access. As the weeks dragged into months, I began to dread afternoons digging into the database. Few findings excited me, and never for long; if I found an intriguing pattern in the river of printouts, I usually found a flaw an hour or a day later.

Ramon Prado rescued me from all this. Months before his death I had noticed that the CHP was failing its mandate to inspect every bus terminal. Absent a healthy dose of mayhem, this was just one more tiny instance of decaying state service. Prado's death, which no inspector could have prevented, make bus inspections newsworthy. In retrospect, my failure with MISTER offers a couple of easy lessons: First, never use a computer you don't control; the control issue alone makes a newsroom PC superior to a mainframe. Second, before you buy a database, before you spend months probing its secrets, find out its weaknesses.

But the major lesson is one that I am not yet sure I have mastered: A database is not a story; it just might lead to a story, but it is not a story. Lose sight of that lesson, and you too can be seduced by technology.