Jon Udell was a senior technical-editor-at-large (he's now Executive Editor, New Media) when he wrote this
March 92
"Stop Bit" column regarding the overload of information available "...at your fingertips." Have things gotten better or worse with the proliferation of web usage?:
Stop Bit: Infoglut At Your Fingertips
Even as more and better information search-
and-retrieval systems become available, the major problem worsens: There is no common way to use them all.
by Jon Udell
When people find out what I do for a living, they often apologize: "Gee, I'm sorry, I don't know anything ahout computers. I really should take a course." Not to worry, I say. You didn't have to go to school to learn how to use your telephone, did you? Computers seem hard now only because they can't quite melt into the woodwork; they aren't yet simply appliances. But that's changing. Global networks, distributed software, and rich interfaces will someday carry information to our offices and homes and place it at our fingertips.
When will this era of universal access arrive? Clearly, we've got some distance to go. When I paint my rosy picture of the future, I conveniently ignore the often ugly reality of network computing today. Spend too much time wrestling with memory managers, configuration files, routers, or interrupt request conflicts, and it's easy t
o lose sight of the big picture. Still, I take for granted that the nuts and bolts will eventually fall into place. I fully expect that we'll see gigabit-per-second data highways and distributed applications that exploit them by the turn of the century. What worries me is whether I'll be able to find what I'm looking for.
A friend who works at Lotus Development told me that he sold a used car in just 2 hours by posting an ad in a Notes database. The scary twist is that he spent the first 90 minutes just trying to figure out which Notes database to use. Here, I think, lies the ultimate challenge of "information at your fingertips." Who's going to put it there, and how?
Information scientists have thought long and hard about this problem. In general, there are just two approaches: Throw everything into the pot and index it all, or carve things up into chunks and index those. Even with storage, networks, and processors gaining size, speed, and intelligence at a breathtaking rate, it's hard for me to
imagine that a planetary database will ever be effectively indexed in its entirety. Most theorists agree that there can be no substitute for the venerable 2000-year-old Aristotelian method: classification. How can we, entering the third millennium, apply that technique to a knowledge base undergoing explosive growth?
Eventually, of course, we'll have to get machines to do some of the classification for us. The Reuters news agency has already taken a step in this direction. Every day, the thousands of news stories transmitted to Reuters from its reporters around the world pass through an expert system that assigns category codes. If you're a precious metals trader, you'll want your Reuters feed to include all the stories about gold and to exclude those that mention a Mr. Gold or the American Express gold card. Reuters's system does in fact attain both high recall (it finds most of the right items) and high precision (it finds few of the wrong items). That's a laudable achievement.
How do the Reuter
s categories relate to the Library of Congress Subject Headings, or to the sets of index terms used by Nexis, ABI/Inform, and the countless other compendia you might consult during a no-holds-barred search? They don't. A bird's-eye view of the world's electronic databases reveals an archipelago, each island having its own passport and dialect. You can't fault the database publishers. No standard controlled vocabulary is viable, and none seems likely to emerge anytime soon. It's not even clear whether such a thing is possible. Certainly the task is so daunting we'd all prefer that legions of Connection Machines or Knowledge Navigators will render it moot. But while we're waiting around for the world's information systems to spontaneously organize themselves, can't we do anything to improve matters?
Think about what happens when a company forms and brings a product to market. It's up to the company to ensure that its own name and the name of its product don't infringe on established names. Why bother? Pur
e self-interest: It's expensive to retool your packaging and advertising when someone challenges a name you've neglected to trademark. Now imagine the same self-interest at work helping to categorize goods, services, and information products so they'll be found when searched for. Nobody would force you to classify your widget, your store, your book, or your TV program, nor would anyone care whether you'd used the correct terminology. But just think of the advertising dollars you'd save if you hit on the right combination.
photo_link (91 Kbytes)
Jon Udell is a BYTE senior technical editor at large. You can reach him
on BIX as "judell."
5 Years Ago in BYTE
