 Live from Cambridge, Massachusetts, it's theCUBE at the MIT Chief Data Officer and Information Quality Symposium with hosts Dave Vellante and Paul Gillan. Hi, we're back at MIT in Cambridge. This is Dave Vellante with Paul Gillan. Welcome. This is the MIT Chief Data Officer Conference, MIT IQ, stands for Information Quality, the hashtag MIT IQ on the event. Actually, Bert Latimore is running a crowd chat under CDO IQ, so you go to crowdchat.net slash CDO IQ and he's summarizing the CUBE interviews and some of the other conversation on the Twitter stream. Mort Anbury is here. He's the Director of Programs and Strategy at the U.S. Army. Mort, welcome to the CUBE. Thank you very much, Dave and Paul. So Mort, tell us about your role at the U.S. Army and how it relates to data quality. As a Director of Programs and Strategy responsible for cost and economic at Army financial management, we are responsible for implementation of cost culture and cost management into financial management processes of the Army. Cost culture, you might imagine, private industry is much easier because everybody is cost conscious. In a government, particularly during the war, that mission is driving everything and most of our officers are only concerned about how much money they need, how they spend it. Insertion of cost culture was a big deal. We had to look at it from the people's perspective, which that includes convincing leadership that attention to cost doesn't create a public bad image for the country because initially they were concerned that parents of soldiers, they would not like the discussion of cost when it comes to the safety of the son and daughter. Right, money is no object. But that, we were able to turn that dollar log around and convince that they don't have to be against each other. You can be a cost conscious, you can do more with your resources, you can be more efficient with it and still care about safety. You can care about the soldiers. So that discussion particularly was timely because Congress has started cutting the DOD budget so our dollar log kind of found its own, I guess, foundation to move forward. So we developed convincing leaders, developed a series of training, convinced leadership for many actions that, or almost all actions that requires more than $10 million, requires cost-benefit analysis. So we developed a course for cost-benefit analysis, trained over 2,000 army analysts, which is they call it ORSA operation research analysts, to perform and validate cost-benefit analysis and make sure that every commander is when they make a decision, there is considered additional course of action and they consider cost and benefit of that. So that kind of guiding training, we create tools for them to do that. So we, this general fund enterprise business system is a transactional army activities. This transactional activities is just like taking all the old system of financial management and turning them upside down and creating ERP and asking every organization what is your data need, identifying data, not data need, information need. In fact, when we are talking to commanders and leadership, it's what is your information need, what you can have an effect on or be able to turn the knob up and down, what's your span of control. And based on that, develop data need for that organization. Then when we define the data need for organization, we find that the data that is needed by a particular organization doesn't come from the same organization, it comes from someone else. Now, communicating that data need from organization A to organization B, telling organization B, you need to provide this data, it's not for you, it's for someone else. That was a big culture shock because, first of all, it was imposing additional constraint and resources to the people. The level of education that requires and awareness that requires that I'm providing something that is not for me and has nothing to do with me was a big shock. And that, we were able to pass that. And then within that ERP, we created cost module. This cost module is kind of looking at the, I use the example of upside down tree, that funding comes from the end and goes to all the branches. We call it fund center. These are the organization that they have money. The cost centers are the one that are using this money. It could be them or it could be others. And it's just like a truly neural network of information that many to many relationship in many aspects. So creating cost model to define their cost centers, work building structure, and then creating this process of what we do, we do cost planning. Cost planning after we do, as during the execution time, Army has some process, they call it planning, programming, budgeting, execution. That's kind of a financial system process. We write this cost management, kind of a overlay this cost management over planning, programming, budgeting, execution, that during the planning, they do the cost planning. They do cost data planning. And when they do cost planning, we go through the whole review of who needs it, why needs it, what's the format, what's the definition, what's the taxonomy, who prepares it, communicated with the organization, provided, feed it, and that information is available. So that's the cost benefit phase. That's what the CBA occurs. No, the CBA is in support of any decision. This is a data part of activities, this is a data need that we are putting in a general G-Fib. Independent of the cost benefit analysis. Yes, independent cost benefit analysis for any decision. For any initiative. For any issue. You said any initiative over 10 million. Over 10 million dollars. Under 10 million. No CBA. No, under 10 million, under 10 million, at the headquarter we don't need CBA. The cost culture has made a search that the commanders, that the smaller group, they want to do that because they are practicing to be able to sell their 10 million dollars so they are doing it for themselves. But it doesn't get reviewed at the headquarter. This involves standardizing data fields and data categories, not only for your own records but also for those of your suppliers. Oh no. Because the suppliers, army, particular army and duty, their main user or main customer of data is in support of decision that they make themselves. It's not to prepare data for suppliers. In fact we are collecting data for suppliers to find out if our contract, if our management of our contract with them is on target or not. Like a Boeing, I mean the Lockheed Martin type of contractors that they are providing major system to the army, they have data providing requirement. In fact we are reviewing those to see what data that providing is sufficient or we have to get raw data from their accounting system because initially we created a particular template, we told them that and we need that data. So as far as answering your question not that much we are focused on providing data to supplier unless that decision of supplier is part of our decision chain and that requirement becomes the data requirement for the army to provide. Have you encountered any surprises in this process or digging into how data is in the government? Many, many. The organizational resistance is one of them. Organizational resistance. Because most of them they don't want someone to use the term digging in someone else's backyard. Accountability makes people nervous. Nervous, right. You know when I think in fact creation of chief data officer is giving a license to someone to dig in the neighbor's backyard. But that's one thing. The other thing is particularly in the cost and in the financial system, definition no matter how normalize and define and standardize your definition is different. If you ask a contractor what is the cost of this engine, a contractor says how much I bought this, how much I put labor and overhead and profit, that's my cost. You ask the prime contractors how it is, put all wraps all those things and make it material and add to it. When you ask what the cost is, with the same definition that should be manufacturing plus profit plus this. So in fact when you look at the taxonomy of definition, it's the same definition. Provider data, if I don't know if it comes from this particular organization, I have to adjust it for if it comes from another layer. Is this one of the things that contributes to the popular misconceptions or popular perceptions of waste in government, $7,000 hammers and that sort of thing? Is that really a data quality issue or a data definition issue? That is to some degree it's accounting, practice and analysis issue. What it is when you are dealing with the analysis of a particular contractor, you want to see how do I allocate the overhead or how do I allocate the, sometimes it gets to the definition. We in allocation process particularly in past would say okay we have a three project with these contractors and we spread the overhead equally. So that type of analysis for dealing with the contractor even though it was useful, it could create the potential for this data being used. And in reality, I think data is just like a living creature. It's aging, it gets disease, it gets contaminated, it tends to fall. And we have made a lot of decision based on the data that later on we found out did not have rigour, sufficient rigour. And I believe in industry is much more particularly when they are mixing each other's data and assuming that creating this big data answer all the question. But the industry is not responsible to answer to anyone. In a government we have so many watchdog from GAO, from the DODIG, from audit agencies, from Congress, from people. We are responsible because we are using tax for your money and we are responsible. So it is our book is open under Freedom of Information Act. You can ask everything that is in my computer is not classified to see it. So this type of visibility creates a little bit critical perspective toward government because you have that visibility. If you ask the same question from some of our major contractors, you will see that there are worse stories than what we have. Because interpretation of data sometimes cause that type of problem. So you were talking off camera a little bit about your passion for transparency in the financial side of the equation. So where are we with regard to that vision? What is the issue of the views? How many views we would choose to look at particular project process organization? We can have contractors, we can have customers, we can have fund centers, source of the fund, system, I guess, time. You know, different types, I mean you can have all these things. When you look at it, in fact, you create polyhydron or multi-dimensional data model. That creation of, that these things make sense altogether. So if I want to have a, let's say 10 dimension, I need 45 table, consistent table to give the integrity and transparency of data to me. When you map that as a data standardization, it gets so huge that no human being in a transaction form can enter such a data to be useful for that many views. So what we do, we say, okay, what are the questions? This data needs, we go scoop data based on need in order not to create too much standardization at the lower data layer, which is quite costly. Most of our data transaction data entry in the government are low-level government employee that they might not have even education of understanding that quality of concept. So I went travel, they tell me, oh, you travel was for what? It was for training. Was it for ERP as well? I have to identify how much it was ERP. Was it helping the airline industry? Yes, I mean, all of a sudden, he said, oh my God, leave me alone. I just had the $200 in travel. Get the humans out. Let me enter these $200. But someone comment says, what did you do with these $200? And so the problem is- Slice it up into the little pieces of it. The pieces, because of the information, and that piece is needed. They are asking for it. The Congress is asking for it. People are asking for it. When we are talking about audibility in the government, people think it's just like balancing the book and having a general ledger work out and everything. That's not what we are talking about in accountability in the government. We want to know if $200 more than I've spent for the airline coming from Washington, why, what happened, what was the result of it, who gained from it, who lost from it. When you want to do that, then you can take my $200, explore it to the $120 billion, which is our normal budget, plus OCO, that's an emergency fund, becomes $160 billion. So it's $160 billion. You cannot explain it like this. So that is our challenge. So we are trying to find the balance on a data need and a view that we need. Identify the view, write the view, identify the source, and take data for need. If the new information need comes, then we go, it says, okay, who needs it? Who has it? It's structured, they're cost model as such to provide that data. Other than that, it would be impossible for us to create- When we are talking big data, it's huge data in our financial system of army. Every day, few million transaction records happen. And in the course of a year, even with the fastest system it takes, like a few days to get a very meaningful report from it, if you want to tag several things. So that's the challenge we are working on. One, and I think we have a lot of progress. The cost management is on its feet and it's working. Our G-FAP is producing a series of data. We can ask a pretty relevant question from it and be able to answer. And we are learning and growing and moving forward. Excellent. All right, Mark, well, we'll leave it there. Thanks very much for coming on The Cube. Thank you. It's great to meet you. Thank you, nice to meet you. Keep it right there, everybody. We'll be back with our next guest right after this. This is The Cube, we're live from MIT. Right back.