More than three hundred industry insiders gathered in Madison, Wisconsin last week for Localization World. It was a somnambulistic event, numbed by economic uncertainty and a looming slowdown in the global economy. One topic threatened to wake the sleepers — the recent roadmap announcement of the GlobalSight Open Source Initiative. Welocalize purchased the system as part of its Transware acquisition and now plans to re-bottle the vintage translation management system (TMS) for release in January. The company is busy replacing system components with open-source foundations; for example, by re-tooling the Oracle database layer for MySQL.

Based on audience participants, two arguments underpin the current LSP thinking around open-source TMS. First and foremost is the argument that open source protects the users from predatory M&A by SDL or other power players in the market. Second, open source ostensibly carries an attractive price point: free. We see serious flaws in both arguments, but the flaws by themselves are caveats rather than killers.

• Is open-source software really free? For complex business process management software like TMS, no. License cost is a relatively minor factor compared to the planning, business process re-engineering, infrastructure, implementation, training, support, maintenance, headcount, and eventual scaling up and out of both the platform and the process. Looked at from this perspective of the total cost of ownership (TCO), license fees account for only 10 or 20 percent of the total cost. Picking the wrong software is a far greater concern, because it puts 100 percent of the investment at risk. No company should ever install TMS software because it’s free. If an enterprise picks the wrong package, it could affect a product release and cause permanent damage to global market share. If an LSP implements the wrong software, it could jeopardize the business.
• Does open source protect against predatory M&A? If the efforts gain traction, yes. However, the vast majority of open-source initiatives fail to garner enough development energy to stay competitive with commercial efforts over the long haul. Some do, such as Project Open. In fact, that effort has been so successful in terms of downloads and development hours, it ranks above 99.9 percent of open-source initiatives as measured by SourceForge. It’s the exception: The vast majority of the 135,000 projects on SourceForge are either comatose or cryogenically frozen. Kudos to the Project Open community, but theirs is not the fate seen by most.

We would love to be proven wrong, but we predict that the new open-source TMS initiatives will be unable to marshal and maintain sufficient resources to stay competitive over a three- to five-year period. The real danger here is not predatory M&A but open source itself and inadequate development:

1. The history of open source in the language sector argues against GlobalSight, especially as other initiatives such as FOLT enter the scene and draw resources into separate and competing camps.
2. Insufficient development resources won’t keep pace with fast-moving commercial developers. By way of example, Across currently employs dozens of full-time employees in engineering, software QA, support, and product management. SDL employs a good deal more than that.
3. Expect a tug-of-war between development efforts for LSP needs versus corporate needs. Like two people on a bed pulling the covers toward their own sides, we suspect that it will be difficult for a resource-challenged open-source effort to balance the licensing, technology, and integration requirements of both LSPs and enterprise clients on a single platform.

Two factors put wind in the sails of GlobalSight right now. The first is the disenfranchised users of Idiom WorldServer, who need to move off that system in the coming year. A workflow-centric system like GlobalSight could be a good home for workgroups accustomed to the Idiom paradigm. The second factor is early involvement from the enterprise giants who could ultimately back the system with development resources: AOL, AutoDesk, Cisco, EMC, IBM, Novell, Salesforce.com, Sun, and Tibco all have seats on the GlobalSight Steering Committee. Sitting on a committee won’t be enough to guarantee success. But if half these companies step forward next year with engineering resources to work on the initiative, good things could start to happen.

Source:http://www.globalwatchtower.com/

Benjamin B. Sargent 23 October 2008