Subscribe to our E-Newsletter


Smith's Blog

Posted Oct 11, 2010 at 12:51 PM by Smith Yewell

I want to thank Keith Mills, the CTO at SDL, for participating in the debate. Keith is a sharp guy and great CTO. SDL is lucky to have him. I also want to thank Jaap Van der Meer for hosting and moderating the debate.

I hope Keith enjoyed the debate as much as I. Our contrasting viewpoints were very interesting and thought provoking, and we also had some points in common! Kirti Vashee has written an interesting summary of the debate in his blog.

Let me begin with the views we had in common. We both agreed on the changes happening in the marketplace. The volume of content is growing; the complexity of rich-media is growing; the number of languages requested by clients is growing; clients are demanding ever-faster delivery cycles, and machine translation is creating growth opportunities for all rather than cannibalizing them. I described my “vision” for addressing these challenges, and Keith described his “practical” ideas for doing the same. It was interesting to me that our answers were characterized as “practical” vs. “visionary”, because I believe both are necessary to solve any challenge. A vision describes the shape of things to come, and a mission describes things to be done within the context of the vision in a “practical” sense. Yes, we have a vision at Welocalize, but we are also on a mission to move beyond optimization of the traditional translation supply chain towards truly collaborative innovation.

My vision for the future is one where our products and services enable any content, on any device, in any language – at any time. The “always-on” world in which we live is requiring “on-demand translation” in order to keep up with the increased pace and volume of content demanded by end-users the world over. I believe our industry is still coming up short in creating an easy-to-use multilingual end-user experience, and I believe that a lack of interoperability has created critical breaks in our translation supply chain at two major points which I call the first and last mile.

This interoperability concept is where my views and Keith’s began to diverge, primarily because of our different business models. I described SDL’s business model as a “walled garden”, and the Welocalize business model as an open, interoperable supply chain.

SDL currently owns the first mile in our supply chain. I used a railroad analogy during the debate to describe SDL’s brilliant purchase of Trados. Trados is still the preferred tool of translators and with Trados on the majority of translator’s desktops, SDL effectively owns the railroad tracks into those translators and is able to levy a “tax” on the words transported on the tracks in the form of licensing fees. Absent the purchase of Trados licenses, APIs and/or SDL professional services, there is subsequently no way to interoperate with the station at the end of the tracks (the translator’s workbench). In addition, because translators are typically isolated in a “desktop workbench” world rather than a networked “community”, knowledge is not efficiently shared and quality suffers.

I don’t think we will be able to keep pace with the changes in the world until we bridge this first mile gap in our supply chain. We need to make it easier and cheaper for translators to deliver quality work on time. Translators should be able to select an inexpensive or even free tool of their choice and have it easily connect to any other tool in the supply chain. Translators should be able to openly collaborate on a shared platform – even across competing multilanguage vendors. We need to enable translators to be more efficient and more profitable. Otherwise, the number of high-quality translators in our supply chain will decline.

Interoperability is also limited by the SDL walled garden in the first mile connections between content management systems and various translation management systems. Once again, absent the purchase of SDL licenses, APIs and/or professional services, there is no way to connect the systems and enable efficient interoperability. Both client and vendor should each be able to choose their preferred translation management system and content management system, and those systems should be able to pass data back and forth. A large amount of time and money is lost because there is currently no easy way to link these systems.

However, three companies are endeavoring to change this fact. Welocalize, Andra AG and Kilgray are currently collaborating within the context of the Interoperability Manifesto to enable data exchange across tools and will open source the results. This is in addition to the connections being created between open source content management systems, GlobalSight and the Open TM2 workbench from IBM. A seamless data round trip across tools is the goal, and these reference implementations are demonstrating how it is possible.

The last mile in our supply chain is also broken. The same lack of interoperability in the first mile of our supply chain limits our connectivity in the last mile to end-users. Community technologies are bringing buyers and sellers ever closer together and improving the user experience. However, the traditional translation quality/review process rarely includes end-users. Instead, the process is limited to a closed loop of linguistic review by additional linguists.

This closed loop of translation QA is being perpetuated by the walled garden. Until we can create open and seamless connections from end-users into the translation systems that we use, we will not be able to optimize our supply chain to the needs of the end-users.

During the debate, I described our supply chain as a series of black boxes that are disconnected from each other. I think our industry has reached a state of maturity where it is imperative to open the boxes and create a better user experience for each constituency in the chain. The result will be “practical” time, cost and quality improvements, and all boats will rise on the changing industry tide.

Post a Comment

Copyright 2009 Welocalize