Blog archive

Possible Game Changer: IBM's Open Source Watson Cloud Platform

When IBM announced its decision last month to turn its Watson cognitive computing technology into an open software development platform, complete with APIs and (Big Blue hopes) a partner ecosystem, the news didn't exactly set the world on fire, but maybe it should have.

News of Watson's victory in 2011 over two human contestants on the Jeopardy game show did spark a mainstream media blaze, albeit a brief one, rife with facile quips about IBM's "Frankenstein of trivia," and repeats of übercontestant Ken Jennings' comment: "I, for one, welcome our computer overlords."

But that was Watson the Machine, a system specifically designed to compete on the game show, comprised 90 IBM Power 750 servers, each running 8 POWER7 cores (3.5 GHz) with 4 threads per core. The OS was SUSE Linux Enterprise Server 11; the software was written in Java, C++, and Prolog; and it used Apache's Hadoop and UIMA and IBM's DeepQA software. The resulting system was able to interpret queries in natural language and used statistical analysis, advanced analytics, and all that processing power to search millions of pages in seconds.

Now comes Watson the Platform -- or more precisely, the Watson Developer Cloud -- a cloud environment for the development of cognitive computing applications that use the big data and analytics capabilities that killed on Jeopardy. Developers will be able to embed a Watson capability into either an existing application or a new app. They'll access the cognitive computing mojo via the Watson Experience Manager, a portal app that provides access to a development sandbox. And each app will be defined on the Watson Developer Cloud by partners adding their own content from the Watson Content Store. The result will be apps labeled "Powered by IBM Watson."

For developers, the IBM Platform/Developer Cloud holds the promise of an entry ramp into the world of cognitive computing, which Big Blue reminds us often is likely to be new territory for most software makers. As IBM puts it, Watson brings to developers its ability "to help facilitate a dialogue, put content in context, maintain continuity of discussions, cull through millions of pages of data, return insights, identify patterns difficult to detect, and learn throughout the process." The ability to add these capabilities to applications "can truly be game-changing," the company says. And it's probably right.

IBM concedes that it will take an ecosystem for this ambition undertaking to succeed, and it's already forming strategic partnerships through the Watson Ecosystem Program. Among the first to sign up was Elance, provider of an online work marketplace. IBM has tapped Elance as its "inaugural talent partner," the company said in a statement, to build a Watson Talent Platform.

"Actually, we're the only talent partner," Elance CMO Rich Pearson told ADTmag.

Elance, which is headquartered in Mountain View, Calif., with offices in Oslo, Norway, matches businesses with freelancers online via a public "talent cloud" comprising about 3 million freelances in 170 countries. The system filled 1.2 million jobs this year, Pearson said.

Working with IBM, Elance developed a private talent cloud for the new Watson Ecosystem. The company's infrastructure, which tests skills and collects job performance data, will help to populate the ecosystem with top performers, Pearson said. The initial talent pool for Watson-enabled apps will likely include data scientists, mobile developers, and designers with experience in data visualization, he said. The company has partnered with Skilled Up to provide online courses for developers who want to ramp up their skillsets. Elance is also planning to apply Watson technology to its own business, Pearson said.

"We're in the business of matching talent and businesses," he said, "and we use our own algorithm for that process. The idea of using IBM Watson technology to improve that process is fascinating from our perspective. The Watson secret sauce is its ability to help with decision making, and we're looking at it as a way of providing richer data to help us make smarter and faster decisions."

IBM's nascent plan is still in the making, though the company has promised to start delivering on all of this in 2014. In the meantime, Watson and Watson-like technologies have already spread beyond that flashy game show debut, perhaps most notably to Memorial Sloan-Kettering Cancer Center. The Center and IBM are working together to combine Watson's supercomputing powers with Sloan-Kettering's clinical know-how—along with "existing molecular and genomic data and vast repository of cancer case histories"—to create a diagnostic and treatment system based on updated research.

Keep an eye on the IBM Watson Web site (and this blog) for further developments.

Posted by John K. Waters on December 17, 2013