Page II: Explosion of special-interest XML dialects could mean the standard is a success or could be the start of a new headache.
"There's an incentive to create a language to solve your specific problem," Bray said. "But if there's something out there already that might serve your need, you should consider using it."
Ron Schmelzer, an analyst for research company ZapThink, says industry leaders typically have little trouble agreeing on what data needs to be represented in an XML schema but get hung up on how to do it -- sometimes creating conflicting specifications.
"When you have two different organisations trying to push two different vocabularies for solving the same problem, it doesn't help the supply chain," Schmelzer said. "If you're a small guy, supporting a bunch of different schemas gets difficult."
But proliferating schemas are more often a reflection of the complexity of the data that needs to be described, said Chuck Allen, director of the HR-XML Consortium, a human resources trade group shepherding more than a dozen XML offshoots to standardise data formats in areas such as payroll and stock-incentive plans.
"There has been some concern about hundreds of standards groups duplicating efforts, and there are cases where some of these groups could look over the other's shoulders more closely," Allen said. "But it gets complicated when you're trying to draft metadata standards to capture all this very complex domain knowledge."
Allen said his group employs sensible standards to ensure new XML projects truly serve a purpose. "We need at least three organisational sponsors and 10 participants," he said. "The main criteria are 'Is it in our domain?' and 'Is anybody else doing something about it?'"
Likewise, it would have been easy for the insurance industry to spawn a wealth of standards specialised for everything from boat coverage to reinsurance. But Lloyd Chumbley, assistant vice president of standards for trade group Acord, said the industry had a head start because it had already centralised on common paper forms, mainly to ensure agents could easily share data with insurers.
"When you're trying to do quotes for a policy, the last thing you need is to have to talk several different languages to communicate with several different insurers," he said. "The insurance industry for the most part has been using standardised forms generated by Acord since the 1960s, and that that helped us maintain a single point of reference as everything got digitised."
Chumbley said the main proliferation challenge in the insurance industry is the localised schemas that have emerged to reflect changes in national laws. "We deal with a lot of different organisations internationally to consolidate XML schemas and definitions," he said. "When you're dealing across different cultures and legal systems, it takes time, but we're making progress."
Allen also expects consolidation of XML dialects. "There's been a lot of speculation as to whether there'll be more convergence, and I think that is going to be the case," he said. "I think it'll be because of IP (intellectual property) issues...which are sometimes more costly than the actual development. It takes a lot of resources to review the patent libraries, police the group's IP policies. If you have fewer organisations, there's fewer IP agreements."
John Simpson, author of several XML-related books, said the proliferation of XML dialects to describe similar data sets isn't the chaos machine one might assume, thanks to the ease of translating from one dialect to another.
"The fact there are different standards is immaterial...it's almost trivial to get it from one dialect into another," Simpson said, crediting the simplicity and integrity of the main XML specification.
"They came up with really simple rules for how the XML spec is going to develop, and those have allowed tremendous flexibility," said Simpson, who created his own schema for classifying "B" movies. "People refer to XML as a language, but it's really a grammar for inventing new languages or describing ones that already exist. The XML spec itself is this kind of wonderful chameleon."
Stephen O'Grady, an analyst at researcher RedMonk, agreed that the simplicity of the base XML standard makes it easy to accommodate multiple dialects, but he envisions a kind of Darwinian selection for competing schemas: Multiple approaches to similar problems blossom, the market states a preference and supporting software is tweaked to push data from one XML dialect into another.
"Because XML is the way it is, it's usually not intrinsically difficult to extract information," O'Grady said. "The situation with (Web log formats) RSS and Atom is a good example. I think it's likely the market will end up deciding one is the way to go over another, and then it's a pretty easy task to consolidate."
High-tech chess players, meanwhile, have a bounty of options. With five projects and counting underway to develop an XML-based system for describing chess moves, about the only apparent agreement is that one side has to be white and the other black.
6%
2%
