What do people mean by the word, ‘open’, when talking about open systems or open source? Why do they so often treat it as the opposite of the word, ‘proprietary’? Are they right to do so?
You might more productively regard these terms as representing points on two different axes of meaning. On one axis (the horizontal, say) is the range between completely proprietary at one end and completely standard at the other. These concepts are truly distinct—proprietary objects are owned by one person or organization, while standard ones are owned by many.
The other, vertical, axis has at its base the notion of an object's being open or not. If it is genuinely open, its technical specification will have been published. The converse of this, at the top of the axis, is the notion of an object's being closed, or kept secret.
These characteristics and axes are shown below.
Proprietary objects are usually easy to detect, if only by the forest of ®, © and ™ icons you must fight through whenever you read about them. If the object is software, it tends also to come with a lengthy end-user licence agreement (EULA) attached, full of stern warnings and prohibitions. For instance, the EULA for Microsoft Windows XP contains over 5,600 words. (That for Red Hat Linux contains 840 words.) These symbols and documents are all assertions of intellectual property ownership. They are ‘keep off’ signs erected by people fearful of having their cash cows rustled.
Hardware is also often proprietary. An IBM hard drive might have a similar specification to one from Seagate but it would be no good making a warranty return of that drive to Seagate. Only IBM would have a commercial interest in dealing with its faulty disc drives.
However, later this year, you could reasonably expect to be able to send a newly bought Maxtor drive to Seagate under warranty. The latter has launched a takeover bid for Seagate that should be completed after this summer. The design and construction of the drive will have changed not one whit meanwhile but the ownership of its design will have.
Standard products, by contrast, have their design owned by many, usually through the medium of a standards body. For example, a CD-ROM needs to conform to the ISO/IEC 10149 and ECMA 130 standards for data interchange if it is to work in many computers . Although partly made up of makers' representatives, these organizations do not make, service or own the products they standardize. They do, though, exercise a proprietary interest in the standards themselves.
The ownership of standards might be metaphorical rather than legal. For example, the Industry Standard Architecture (ISA) bus for personal computers has no owner. Although introduced in 1984 by IBM (as the PC AT expansion bus), effective ownership has passed out of its hands. There have been so many divergent implementations that no single agreed statement of the specification exists, yet it works as a standard.
Closed products are those whose technical details are kept secret. Only the maker and its licensees have access to the detailed specification of such a product. The formula for a well-known carbonated drink is an example from the general market. Within computing, Apple has long practised this approach. Until recently, a search for hardware compatible with an Apple Macintosh would have turned up only peripherals. Now, you can get disc drives, memory chips and even motherboards. The Mac is less closed than it was, in other words. Although a long way from being open, it is at least slightly ajar.
Open products are the converse; their details are widely and purposely available. ‘Open source’ software is a prime example, having its source code exposed for all to see, adopt, share and change. The widely used GNU Public License [sic] goes so far as to forbid the denial of your rights to do any of those things.
The popularity of this attitude can be seen by looking on the Web site of the estimable SourceForge.net repository. There you can read about and download literally tens of thousands of programs, all available freely (although not necessarily at no charge). The site has the support of IBM and Hewlett Packard, who both offer some of their products through it.
The differentiations suggested above gain some point when you start positioning products on the diagram. To help with that, here it is in a refined form, with quadrants shown. ‘CS’ stands for closed and standard, ‘CP’ for closed and proprietary, and so on.
Closed and standard (upper left) is not an oxymoron. In the same way as Coca-Cola restricts access to its recipe for soda-pop, so do the military, police and ‘intelligence’ services control access to the specifications for certain items of software and hardware they use. These often have more than one supplier, so need some circulation but only to selected people and organizations. Unless one is in that market, you are not allowed to see these standards or, sometimes, even to know of their existence.
Closed and proprietary (upper right). This is where one would have positioned Apple Macintosh hardware until recently, as discussed earlier. Unlike ‘Wintel’ personal computers, the Mac was and still is clone-proof. There has been no equivalent to Compaq or Dell in the Apple world, which perhaps helps explain why over 90% of personal computers conform to the pattern originally laid down by IBM .
Open and proprietary (lower right). This is where the owner of a copyright, patent or design makes it possible for others to access some part of the product, while not abandoning its ownership. Here is where you find the Apple operating system, for instance, or almost any Microsoft program. Both suppliers widely publish the rules for writing software that will work with their products. The flexible straitjacket this has placed around software developers has been one of the main contributors to the success of both companies.
Open and standard (lower left). Here you find such arrangements as IEEE's various 802.x standards for computer networking and the Extensible Markup Language (XML) for text formatting published by the World Wide Web Consortium (W3C). These are available for anyone to use and, like Open Source Software, have their detailed specifications open to view.
Interestingly, both sets involve standardizing previously proprietary methods. IEEE 802.1, for example, derives from the proprietary Ethernet methods patented by Xerox, DEC and Intel in the 1970s . XML is based on the ISO Standard Generalized Markup Language (SGML), which in turn stems from IBM's Generalized Markup Language developed in the 1960s.
Putting It to Use
What now results is a useful tool for mapping suppliers' products, architectures and plans. (Bear in mind that the lines between quadrants are not as clear-cut in reality as they appear on the diagram.)
You can position offerings against the definitions above or against their competitors' products. Supplier's representatives are usually delighted to help with this exercise, particularly when positioning their competitors' products.
They are typically less so when asked to explain why their own company's new ‘open’ products do not occupy the lower left quadrant. The reply will almost invariably be along the lines that completely standard products do not offer “the requisite functionality”. To meet “the customer's real needs”, their company has “added value” to the base standard. Any customer of Microsoft or Dell will be familiar with the ploy. The result, “as any fule kno”, is something non-standard.
Most suppliers are trying to face two ways at once. On the one hand they are proclaiming the true openness of their systems (and possibly even believing this to be so). On the other they are unwilling to let go of the old certainties of proprietary lock-in and high margins on products. How any particular supplier reconciles these opposing forces is fundamental to what you can buy from them.
 ISO is the International Organization for Standardization, IEC is the International Electrotechnical Commission and ECMA is the European Computer Manufacturers Organization.
 And, yes, despite the protestations of its more partisan users, Apple Macintosh is a personal computer (except of course when it is being used as a server).
 IEEE is the Institute of Electrical and Electronics Engineers, an American body.