Groups welcome Recovery board's mash-up system to uncover fraud

A recent decision by federal stimulus watchdogs to buy a system that can mash up -- or extract and compare -- reports from fund recipients with existing data to prevent fraud is unlikely to overlap with the work of citizen developers, the Washington transparency community says.

The General Services Administration on Aug. 26 issued a solicitation on behalf of the board overseeing Recovery Act funds for a fraud-analysis platform that can search and discover possible cases of suspicious activity to aid federal inspectors general. The tool will be part of a proactive approach to rooting out waste in government, the Recovery Accountability and Transparency Board said in a statement of objectives.

Board members have long said volunteer, third-party developers -- such as transparency activists -- should devise analytical mash-ups to hold the government accountable for how taxpayer funds are spent. Open government advocates told Nextgov they did not view this procurement as a reversal of that stance, but rather an inherently governmental operation that could create opportunities for later public-private collaboration.

"Obviously the government shouldn't count on the [transparency] community to do its job for it, and I think this is an example of that," said Tom Lee, a technology director at the Sunlight Foundation and a member of Sunlight Labs. "Looking for waste, fraud and abuse is a key part of the [board's] mandate." The lab's division of the foundation organizes a group of volunteer developers to build Web services aimed at increasing government accountability.

"If an IG comes to the [board] and says, 'This guy's been running a crooked construction firm, we need to run a report on his business associates,' it makes sense for the board to turn to in-house analysts to answer that request. They can't just post it to a mailing list and hope that someone will do it for them," Lee noted. "So from that perspective this RFP makes perfect sense."

The system is not intended to scrutinize transactions, but rather detect "areas of concern" and refer those cases to agency IGs for further investigation. Today, there is no single system for the IG community capable of identifying repeat instances of fund mismanagement. For example, a vendor could receive funds through one agency while being blacklisted by another, according to the request for proposals.

Projects would be flagged as "of concern" based on predictors of risk, such as high-prices, past criminal records of fund recipients, regional unemployment rates and tips from citizens.

The board expects the winning vendor to offer software that can find relationships across searchable databases, provide predictive analysis software and install graphics programs that are "supported by a robust data extraction and 'mash-up' capability." The solicitation names software products already on the market that would fit its needs, including Regulatory Data Corp.'s GRID, Palantir and Norkom Technologies' Digital Harbor.

Although Earl Devaney, the board's chairman, "has encouraged the creation of citizen IGs, it is not a replacement for the government doing the hard work to eliminate fraud," said Gary Bass, who serves on the steering committee of the Coalition for an Accountable Recovery, a group of about 30 public interest organizations. Board members, who are experienced federal IGs, have a firm grasp of the equipment they need to do their jobs, he added. "This procurement seems to fall in the category of useful tools that will provide the ammunition they need to fight the accountability battles they are entering," said Bass, who also is the executive director of open government group OMB Watch.

The board foresees hiring a so-called data fusion team composed of one to three analysts with experience in white-collar crime to run the platform. The search software would troll relevant media coverage and regulatory Web sites to identify instances of past money laundering, terrorist financing, drug trafficking, labor violations and other risk factors. The platform would then analyze these search hits using statistical algorithms to uncover nonobvious connections. These relationships might reveal facts that indicate improper awards or result in investigations and audits.

Should links emerge, the platform then would be able to chart, graph or map matching criteria to give investigators a compelling understanding of the results. The system would have to be configured to allow the board to visualize -- or pictorially display -- all stimulus financial transactions. The hired personnel would be charged with creating the mash-ups, running searches and producing referral reports for board staff to hand IGs.

Sunlight officials acknowledged that the board needs an in-house system to detect fraud and do predictive analyses, but say visualizations might be beyond the board's needs and best left to the outside community. "The visualization component of the RFP sounds like it may be unnecessary eye candy," Lee said.

The government should focus on looking for waste, but in general, Recovery.gov -- the public stimulus-monitoring site -- should focus on providing data libraries, codes for building applications and bulk downloads to the community, he added. Paying contractors to design pretty graphics for nontechnical board members might be excessive, according to Lee. "The less attractive the graphs wind up being," the better they may be at actually fostering analysis, he noted.

Board spokesman Ed Pound responded that Congress clearly directed the Recovery Board to provide accountability and to prevent fraud and waste. "This is an inherently governmental function performed by IGs for decades," he said.

Bass said he doesn't expect this operation to duplicate work done in the public or private sectors. "The board likely has access to information that most other agencies and the public do not," such as internal federal reports, he said. Bass suggested it might be a good idea to grant the system access to an existing federal database that rates contractor performance, called the Contractor Performance Assessment Reporting System. The site is restricted to federal officials, who often use its ratings to make contract award decisions.

Furthermore, only an in-house system would have access to the organizational identification numbers that aid recipients entered to register for reporting, Bass added. The identifiers could help investigators avoid mistaking one company for another with a similar name and ensure investigators are dealing with relevant subsidiaries, should a parent company sell off an arm.

"They have to be sure they're wrestling with the same companies," Bass said. "This procurement, if successfully implemented, could be an excellent advertisement for the concept of data mash-ups and provide a stronger to push to get more government data published so that citizen IGs can eventually build better fraud detection tools as well as identify ways to improve the quality of government programs."