Crowd Control

Whether it is offshoring, outsourcing or automation, the financial industry is always on the lookout for new ways of getting more done for less. On March 27, Inside Reference Data hosted a webcast about a new operational model called crowd computing, which promises such efficiency.

During the webinar, which was sponsored by WorkFusion, Adam Devine, the company's New York-based vice president, product marketing and strategic partnerships, explained how crowd computing platforms, such as the one his company provides, allow users to bring together a hybrid workforce of subject matter experts, freelance workers and automation tools to work on particular projects. The platform learns from the output of the human workers so that over time, a greater proportion of work can be automated and headcount can be reduced.

Success Story

Crowd computing is in use today in a number of industries, including healthcare, the media and financial services. Among data providers, Thomson Reuters was an early adopter. Peter Marney, now New York-based vice president, global content management at publishing company John Wiley and Sons, explained why Thomson Reuters adopted WorkFusion's platform during his time there as senior vice president, platform and information strategy, which ended in June when he began at Wiley. Marney, who is an industry advisor to WorkFusion, said he had been responsible for a database of corporate entities, which was growing so fast it was difficult to keep the records up to date and add new ones. He said the ability to combine a crowd of outsourced workers with automation tools and subject matter experts met Thomson Reuters' needs exactly.

Hubert Holmes, New York-based managing director, reference data, at Interactive Data, said his company is not using crowd computing at the moment, but has explored how it could help to keep down the cost of collecting more data.

"I find crowd computing to be potentially a very cost-effective methodology of getting at lots of public data," said Holmes. "As we expand our content sets and do linkages across legal entity and security data, and really enrich our data more, that entails bringing in new data sets. To the degree that has been cumbersome—you have to add people and systems—and costly, we haven't done it as much as perhaps we would have liked. If we can get the cost of data collection down and scale it—turn it on and off—or if you have projects to create certain data sets, it would give us that impetus to do it."

Sweet Spot

Keith Broadhead, New York-based head of solution sales, Americas, at SIX Financial Information, was new to the concept of crowd computing, but said he believed it could help data vendors manage the challenge of analyzing, normalizing and categorizing big data.

A poll of listeners found 45% of the audience believed market data products would benefit the most from crowd computing. Broadhead was unsurprised to see market data as the top choice, but said he expected evaluated pricing to be more popular as it is such a hot topic at the moment.

Crowd computing can be used to harness the work of thousands of freelancers, as well as permanent members of staff. With so many people contributing to the management of data via a crowd computing platform, one listener questioned how the security of the data could be guaranteed.

Max Yankelevich, WorkFusion's chief architect, CEO and co-founder, explained that security is ensured because each data management job is broken down into so many separate "micro tasks" that no single worker has access to any sensitive information. "Once you break down the process into micro tasks, you are effectively putting this process through a shredder where each crowd worker sees only a very, very small part," he said. Yankelevich explained that this parsing of work into many tasks means crowd computing has, for example, been used to analyse tax returns without any threat to the security of personal data.

While breaking down a job into a series of micro tasks can take some getting used to, it is essential for the success of a crowd computing project, said Marney. He explained Thomson Reuters considered the filling of an address field to be a single task, but for the purposes of its crowd computing project, it ended up being broken down into 25 micro tasks, including looking up the company website, finding the address page, extracting the address and parsing it.

The lesson? To get macro benefits from crowd computing, focus on the micro details.