As a data scientist, I of course love the police dramas that use analytics to derive key evidence that put the crooks in jail for a long, long time. Well, a few weeks ago, as a part of our ongoing series, Conversations on Big Data, I had a window into just such an operation when I sat down with Steve Beltz, Assistant Director of the Recovery Operations Center of the Recovery, Accountability, and Transparency Board (fondly known as RAT). The RAT Board was originally created by the American Recovery and Reinvestment Act of 2009 (ARRA) to provide transparency of ARRA-related funds and detect and prevent fraud, waste, and mismanagement of those funds. Later the RAT Board authority was expanded to include oversight of all federal funding.
Steve has been working for the public sector for over three decades, mostly in law enforcement as a “detective and reconstructionist”. Steve told us that he “was the first person in the Washington State Patrol that started the computer forensic unit.” “I would get a 100 MB disk.” “DOS was my tool … we’d walk sector by sector through a disk and analyze it.”
While technology has a come a long way, Steve says “you can have the best data in the world and the best tools in the world, but if you don’t have the right person that knows how to ask the right questions, you’ll get nothing.” He goes on to say that “somebody has to know how to understand what the answer is, and then dig deeper.” It takes a special type of skill to look beyond the numbers and the technology and take that extra step to dig deeper into the evidence.
Steve shared a story with us that illustrated his “detection through analytics” approach. He was teaching a class to the Federal Law Enforcement Training Center, when one of the agents in his class reminded him that they had met in the field. The agent recounted that “I was with the prosecutor … interviewing the defendant, and the defense attorney … sitting across the table, the typical thing. They let the defendant talk and explain … and [he] was way off in right field, with his conversation and not even close to what … we’ve already dug up. So it finally got to the point where the prosecutor took out a chart that we had made, the link chart of all his relationships and who he’s working with, and where the money is going, and kind of slid it across the table and said, well you’ve said this, but this is what we found out. And they took an immediate recess for about 10 minutes and came back and decided … to plead guilty.”
In describing the most important ingredients for a successful analytics program, Steve used a 3 legged stool to describe his approach. He feels that is is necessary to have:
1. Good Data
2. Good Tools
3. Good Dedicated Analysts
Steve says these are all critical components without which, “you’re going to sell yourself very short.” He goes on to say that “acquiring good data is a common problem that will never go away so long as shared data is in use.” Steve’s experience has been that most agencies struggle with obtaining good tools, and even better analysts.” On the other hand, “tools are getting very good, and conversely so, very complicated. The analysis process itself … is a full-time job and a lot of agencies are having … people wear multiple hats. So you’ve got an agent that’s not only out investigating but also … trying to come back and do the analysis as well.”
While the RAT Board will be winding down in the near future, Steve acknowledges that the results of its work are still pending. The Recovery Ops Center is currently supporting over 200 cases of alleged mismanagement of funds annually. Steve’s analysts have yet to testify in a court, but he is training them for that event. In Steve’s view, “That’s not really what we’re there for.” He feels that his job is to “build intelligence packets.” He is then much happier if his intelligence or “prosecution packets” result in a guilty plea rather than needing to be used in a trial.
To learn more about Steve’s approach, listen to my complete interview with Steve at http://www.ourpublicservice.org/bigdata/. And keep an eye out for more about the RAT Board in future blogs.
Thanks Brian for sharing such a great case study – great information. The good data, good tools and good analyst is a great framing for others starting up their analytics program.
Balanced framework, good overview for developing units. Thanks for the article.