"A company might develop a page layout program that is better than Quark XPress for handling newspapers," Dewar noted, "but can they continue to develop that product and compete? Ultimately Quark will win that battle, maybe not with V-3.3, maybe not with V-4, but by the time V-5 rolls around, they'll win. ..."

Having Steuart Dewar give his opinion on open systems is facile -- after all, that's the concept behind DewarView, so it's unlikely he'll say it's a stupid idea. But asking the same question in the Newsroom Computers section of CompuServe's Journalism Forum yielded a consensus that reflected Dewar's thinking.

One respondent was JForum Assistant Sysop Sue Mosher, a consultant whose company, Slipstick Systems, helps news organizations use information technology more effectively. Mosher held both editorial and technical positions in a 15-year stint with the Associated Press, where she helped to develop computer systems and software both for AP's internal use and for use by member radio and television stations and networks.

She acknowledges that sometimes she misses her first paying media job -- as a disc jockey, playing progressive rock.

"I don't believe that proprietary formats and open systems -- however you want to define them -- are incompatible," Mosher said. "A proprietary format means the proprietor isn't constrained by the rest of the industry when they want to make changes, i.e. to advance, but it does bring an obligation to remain 'open' in the sense that other systems can use it.

"This is not a moral obligation," she said, "but an obligation to the marketplace, which can vote with its feet if it likes."

Three characteristics mark openness, in Mosher's judgement. A product must be extensible, she said, so it can be modified beyond its out-of-the-box configuration; interoperable, so it works with many other applications, "or at least the ones I care most about," and able to run on a standard platform so it "does not require proprietary hardware."

"Proprietary" is not a dead word by any means, as Dan Crough can explain. He's executive vice president of American Computer Innovators of Amherst, Mass. ACI has built its business on open systems; in newspapers, it integrated a DewarView installation at The Record in Troy, N.Y.

"We look at it two different ways: open APIs is one form of proprietary openness. Then there is an open API in terms of industry standard, such as TCP/IP," said Crough. "What you'd like to find is a balance. You don't want proprietary networking. You want Ethernet, FDDI and other true industry standards.

"For operating systems, programs such as Microsoft Windows and Windows NT are proprietary, but made open by the supplier. The point is that they are products that interoperate well. What we look for is a good combination of the two.

"Linux is an operating system, but not a mainstream technology," he said in a telephone interview.

While it might be attractive to have source code, such as to Linux, it's not necessarily a good thing.

"A truly open system is one for which you have the source code and associated specs. ... And, from your [Linux] example above, you see what happens," said the AP's Steve Graham. "What do you suppose the result would be if Microsoft passed out the source code for MS-DOS to anybody that wanted it? Before long you'd have Kumquat DOS running on the Kumquat PC and nothing else."

Graham, whom Mosher describes as "curmudgeon-at-large for the Associated Press and an expert on international news data transmission standards and techniques," also invoked the P-word.

"The other hand is that vendors normally provide various degrees of openness, thus allowing others to provide software and add-on products based on the proprietary underpinnings. The ready availability of details of the IBM PC-AT bus, for example, means that lots of folks can make hardware to fit it. (And you get the joys of resolving conflicting IRQ levels, port addresses and the like.)

"All major operating system providers offer published APIs for programming. That doesn't make it an open system, however. Obviously, the underlying code from Apple, Microsoft, etc., is not for publication -- i.e., it's proprietary."

But proprietary in that sense is not always bad.

"The Mac is rather trouble-free, largely because it isn't an open architecture -- seen many Mac clones lately? On the other hand," Graham said, "you can go crazy installing hardware on the relatively open PC-AT bus."

Graham has hit upon the tradeoff:

Dewar agrees there is such a tradeoff.

"Open systems have their problems too, but a competent systems integrator (and I'll agree there aren't that many of them around!) can go a long way towards solving those problems, especially as they themselves gain experience with the products.

"But the big plus is control and flexibility. These are not easy concepts to grasp because they involve preparing for the future, whilst most system buyers are still trying to address problems of the past ('OK, with your new system, how many keystrokes does it take to lay out a story....')."

Dewar likens the coming changes in the information industry to going forth into battle. Right now, skirmishes are being fought over delivery methods -- "I now have the top news stories delivered four to five times a day on my alphanumeric pager and I find that kind of neat," Dewar said -- and marketing possibilities.

"More and more magazines and newspapers are looking at CD-ROMs and on-line access so they can tap the archive market for additional revenue. And who knows what else is down the pike?

"What we do know is that the best weapon against these changes is a flexible system that can adapt to these new markets."

Acquiring such flexibility relies on a simple test, Dewar said: "A system delivered by Vendor A is open if I can solicit a bid from Vendor B to perform these kinds of services without any contact with Vendor A:

Noting that documentation is also a key issue of openness, Dewar said, "I consider it highly misleading for a vendor to talk about their open system with an SQL database and neglect to tell the end user that all their data in that database is stored via proprietary software, and that the data formats are also considered proprietary."

ACI's Crough agrees that databases are somewhat problematic because storage and retrieval formats make "the physical representation of information" and "the logical representation of data within that physical representation" topics for careful exploration.

"At some point it all comes down to hardware," Crough said. "Right now there are no standard data formats. For example, an object-oriented database stores information in the file structure different than the way that a relational database does.

"What is important that you can get your information out, not just [from] that database, but by doing joins across database engines."

A join is a database term for combining data from more than one table or database. For example, you might want to compare a list of subscribers from your Oracle circulation database with readership data gathered through sampling and focus groups in a Sybase database, Crough said. The point is not which format is used to store your data, he said. The point is whether you can get to your data when you need to.

And that's the crux of the open systems debate. In the end, the question of open support for a system is more important than the theoretical openness of its components.

Getting to use your choice of word processors, pagination software and databases is more important than gaining access to the source code of your operating system.

And the ability to purchase keyboards, monitors, disk drives, memory and terminals on the open market is more important than having a copy of the microcode that runs your processor.

-- Christopher J. Feola

From THE COLE PAPERS, January 1995, Copyright © 1995, All Rights Reserved.