How document imaging is breaking paperwork logjams

Business Horizons, Nov-Dec, 1997 by Saul W. Gellerman

Fully automated information processing is becoming a reality. Why do in days what can now take seconds?

Forty-plus years into the Information Age, the full advantages of the microprocessor over the pencil have yet to be realized. On any given day, most information still sits on innumerable pieces of paper, each waiting to be dealt with in turn. Information piles up like traffic at an intersection, and for the same reason: more of it is trying to pass through a junction than the junction can handle. Clearing those choke points is the key to making fully automated information processing a reality.

For all practical purposes, two such choke points have already been bypassed. One is the problem of moving information from one place to another. The highway on which data travel these days is usually a fiber-optic cable. Like all highways, it can handle only so much traffic. Recently, however, Fujitsu demonstrated that 1.1 trillion bits of information can be sent through a cable in a single second, at least over distances of under 100 miles.

We may not need that much capacity right away; it would be like transmitting in one, second as much information as could be crammed into a daily newspaper over 300 years. Still, AT&T and Nippon Telephone and Telegraph are already racing with Fujitsu to extend the transmission range of "terabit" (trillion-bit) technology in the hopes of making it commercially feasible.

Meantime, we're stuck (if that's the right word) with transmission speeds of "only" 2.5 billion bits per second--a mere quarter of 1 percent of what we'll have when terabit technologies become available. But even at these once unthinkable speeds, our capacity to move information is often overloaded. When too many users log onto the Internet at the same time, the system responds exactly as do the freeways in Los Angeles: it slows down. The National Science Foundation is now looking for innovative ways to give high-priority messages what amounts to a "right of way" on the Internet.

The other already bypassed choke point is processing speed: how quickly information can be analyzed or classified once it gets into a computer. The simplest way to express this problem is to say that, practically speaking, it isn't a problem. In fact, getting raw data into a computer and processed data out of it still takes more time than running it through the computer's programs, even when those programs involve millions of separate steps.

That leaves two remaining choke points, both at the intersection of paperborne information and its conversion to the bits and bytes of electronic processing. Crossing this intersection still requires human intervention--the third choke point. It still takes a hand to pick up that piece of paper, an eye to examine it, and a mind to decide what should be done about it. The human phase of information processing may take several seconds, or minutes, or longer if more than one human has to be consulted. Work has begun on ways of clearing this logjam, especially through artificial intelligence. Admittedly, however, the work still has a long way to go.

The fourth and most congested choke point of all is simply the waiting time that precedes human intervention. Before humans get involved, that piece of paper has to make its way through a queue of many similar pieces of paper, each waiting for a human hand to finally pick it up. That's why your application for a mortgage or a driver's license or a tax refund may take weeks to be acted upon. It has to work its way through a logjam of similar applications, and may be caught in as many jams along the way as there are decision makers in the process. The old Army saying "Hurry up and wait" applies just as much to the flow of paperwork as it does to soldiering.

Document Imaging

Now this most time-consuming of all choke points is about to be cleared. The key that opens it is a technology called "document imaging," which has been available for about seven years but is only now beginning to hit its stride. Most information processing managers know about it, and most other managers don't. But they should.

Among the leading providers of document imaging hardware and software are Filenet in Costa Mesa, California; Wang Laboratories in Billerica, Massachusetts; and Viewstar in Alameda, California. The imaging process they use is essentially the same as that used in your fax machine, except that it is incomparably faster.

The scanner in your fax machine doesn't have to work very fast because the cable it's connected to can just about cope with the information the fax now pumps into it. The fax usually has to transmit its signal through copper because that's probably how your office was wired when it was built. A signal passing through a copper wire travels at only about 1/48 the speed it can attain through a T-1 cable, which is the oldest (and slowest) of the fiber-optic cables currently in use.

The scanner used in document imaging can capture the information on an ordinary 8-1/2 by 11-inch sheet of paper in about 1/18 of a second. "Information" in this case is more than just the words and figures on the page. The scanner divides the page into 20,000 minuscule squares called "pixels" and records the reflections of dark ink or light paper from each of them. The reflection from each pixel and its location on the page are recorded in a binary code: "Yes, this pixel reflects light" or "No, it doesn't." All that information is transferred by a laser onto an optical disk about the size and shape of an old 33-1/3 rpm phonograph record.