Interview: Jeffrey C. Babcock
As public-sector vice president for the North American region of SAS Institute Inc. of Cary, N.C., Jeffrey C. Babcock brings his Environmental Protection Agency experience to the development of the SAS statistical and data warehousing software used by many agencies.
| ||||
GCN:
BABCOCK: The biggest change is the use of technology. I used to do risk analysis by hand. Every now and then I could dial into the EPA mainframe, but that was rare and costly. A lot of the analysis was by punching numbers into a calculator over and over to make sure I got it right.
GCN:'SAS Institute Inc. grew out of an Agriculture Department grant for analyzing crop data. How did that shape the software?
BABCOCK: It was a solid foundation. We submit an annual questionnaire to key government accounts that we call the SASware ballot, which gives our R&D folks a road map of what directions to take. We're driven by that feedback about what is needed in current versions and also in the future.
We put software through a series of tests in the development cycle. Several government clients'USDA being one'are on the early test list to provide active feedback on whether we're meeting their needs.
The Treasury Department is a large customer, pretty much top to bottom. Virtually all its bureaus use our software in some manner. The Customs Service uses SAS for a Government Performance and Results Act application that helps with performance measures and monitoring.
The Secret Service is licensing software to do fraud detection. The Bureau of Alcohol, Tobacco and Firearms uses our software for data mining and warehousing. The IRS is a large customer for financial management and consolidation, and also data warehousing and mining. We just finalized an add-on to a multiyear arrangement with the Comptroller of the Currency for monitoring financial organizations. Fraud detection is a big issue.
The Census Bureau, like USDA, is one of our longest-standing customers. It has had SAS for more than 20 years and is one of our early testers. It receives all our software and not only does a lot of testing but a lot of internal training. Our trainers here get good feedback from Census' teaching. It's what you'd call a true partner, not only in R&D feedback, but in service feedback.
Our five-year contract with Census that expires at the end of next year won us an award for contracting excellence from the Commerce Department. Census told us that future large, multiyear contracts with other vendors will use the SAS contract as a model.
The Army's health care education studies use SAS for fraud detection under Medicaid and Medicare. They saved more than $28 million in fraud and waste over three years.
EPA is near and dear to my heart and is a long-standing customer for risk modeling, and air and water pollution. Lately, it's been using SAS to publish information through the Web and has received an award for its Web work. EPA is an early tester of our software and provides good feedback. One of its headquarters is at Research Triangle Park, N.C., and we spend a great deal of time there helping with applications and receiving feedback.
The Federal Deposit Insurance Corp. received an award last year for an application it wrote with SAS that helps analyze economic and financial data about insured institutions. Jon Wisnieski, the director of research and statistics, provides good feedback to SAS about software directions.
GCN:'Data cleansing for a warehouse is a tough job. What's the best way to do it?
BABCOCK: The cleansing is unique to each application. The methodology we suggest is to take a backwards-in approach. Instead of cleansing first and then trying to figure out what kind of information to get out of it, we spend a lot of time figuring out the kind of information first, then going in and structuring the data.
Too often, organizations take the opposite approach. They do data scrubbing to get rid of duplicate records, misspellings, different abbreviations for the same word and that sort of thing. Any database administrator goes through that anyway for any kind of reporting, but it doesn't really benefit the front end.
We coach our accounts to take that additional step and figure out what they want first. It shortens the implementation time and increases the chances of success.
GCN:'How do you gauge the user friendliness of your software?
BABCOCK: People mistake user friendliness for familiarity. If somebody is familiar with something, they tend to think it's easy to use. What I ask is, When was the last time you completed a job using technology and got it right the first time through? Good question, isn't it. More often than not, the answer is never.
What they're saying is that, each time they do something using technology, it's an iterative process. User friendliness should be the ability of the solution provider to reduce the number of iterative steps and the amount of time it takes to get the job done.
GCN:'How different is your work with state and local governments from what you do with federal agencies?
BABCOCK: There's a lot of trickle-down. The states use the federal government as a lighthouse. I think the federal government does a lot of really good things from a legislative perspective and a direction perspective. The states take their direction in response, both legally and because they are operating in the same environment with the same missions. They model their organizations after the broad federal organizations'treasury, welfare, health care, environmental programs.
From a technology perspective, they're similar. The issue they have is not enough money. Most state governments have far fewer resources to implement things, so they rely more heavily on third parties to do things for them.
GCN:'Outsourcing is greater in state and local governments?
BABCOCK: It's much hotter. The federal government has used systems integrators forever. It has done data warehousing forever, but it just decided to give it a good name in the '90s. People got away from terms like decision support in the late '80s because it was too technical, so they called it data mining. There's nothing new, it just has new clothes.
GCN:'What do you see as the future of Web-enabling enterprise applications?
BABCOCK: It's completely affecting the direction of our development. Version 7 of our software can make just about every tool provide information on the Web. Version 8, which will be released in October or November, completes the circle. We develop in waves, for certain platforms and certain products first because they're our bread and butter. Then we catch the others up in subsequent releases so they're all at the same level. Version 8 will make every solution publish information to the Web simply and easily.'
We're developing portals, which are the wave of the future for interfaces. They're more structured toward the individual who is viewing information. Every organization wants a series of portals for various applications and users. That will be our next big development cycle.
We support the Common Gateway Interface, Hypertext Markup Language and Java. Back in the old days, five or six years ago, if it took as long to process a job as it does now, people would scream and yell. But the public is used to the Web being slow.
There's a development rule: If a Web application is going to take more than 30 seconds to process a request, you're going to lose the user. There's not much that takes more than 30 seconds with anything you would create with Java.
GCN:'What's the biggest challenge of getting into data warehousing and mining?
BABCOCK: Consensus. You have to figure out what information you want. People over time have collected output data about completing tasks. What they might not be prepared to collect is what GPRA wants: the outcome'the reducing of time from point A to point B.
GCN:'How is the government different in your experience from the days when you worked at the Environmental Protection Agency?
BABCOCK: The biggest change is the use of technology. I used to do risk analysis by hand. Every now and then I could dial into the EPA mainframe, but that was rare and costly. A lot of the analysis was by punching numbers into a calculator over and over to make sure I got it right.
GCN:'SAS Institute Inc. grew out of an Agriculture Department grant for analyzing crop data. How did that shape the software?
BABCOCK: It was a solid foundation. We submit an annual questionnaire to key government accounts that we call the SASware ballot, which gives our R&D folks a road map of what directions to take. We're driven by that feedback about what is needed in current versions and also in the future.
We put software through a series of tests in the development cycle. Several government clients'USDA being one'are on the early test list to provide active feedback on whether we're meeting their needs.
The Treasury Department is a large customer, pretty much top to bottom. Virtually all its bureaus use our software in some manner. The Customs Service uses SAS for a Government Performance and Results Act application that helps with performance measures and monitoring.
The Secret Service is licensing software to do fraud detection. The Bureau of Alcohol, Tobacco and Firearms uses our software for data mining and warehousing. The IRS is a large customer for financial management and consolidation, and also data warehousing and mining. We just finalized an add-on to a multiyear arrangement with the Comptroller of the Currency for monitoring financial organizations. Fraud detection is a big issue.
The Census Bureau, like USDA, is one of our longest-standing customers. It has had SAS for more than 20 years and is one of our early testers. It receives all our software and not only does a lot of testing but a lot of internal training. Our trainers here get good feedback from Census' teaching. It's what you'd call a true partner, not only in R&D feedback, but in service feedback.
Our five-year contract with Census that expires at the end of next year won us an award for contracting excellence from the Commerce Department. Census told us that future large, multiyear contracts with other vendors will use the SAS contract as a model.
The Army's health care education studies use SAS for fraud detection under Medicaid and Medicare. They saved more than $28 million in fraud and waste over three years.
EPA is near and dear to my heart and is a long-standing customer for risk modeling, and air and water pollution. Lately, it's been using SAS to publish information through the Web and has received an award for its Web work. EPA is an early tester of our software and provides good feedback. One of its headquarters is at Research Triangle Park, N.C., and we spend a great deal of time there helping with applications and receiving feedback.
The Federal Deposit Insurance Corp. received an award last year for an application it wrote with SAS that helps analyze economic and financial data about insured institutions. Jon Wisnieski, the director of research and statistics, provides good feedback to SAS about software directions.
GCN:'Data cleansing for a warehouse is a tough job. What's the best way to do it?
BABCOCK: The cleansing is unique to each application. The methodology we suggest is to take a backwards-in approach. Instead of cleansing first and then trying to figure out what kind of information to get out of it, we spend a lot of time figuring out the kind of information first, then going in and structuring the data.
Too often, organizations take the opposite approach. They do data scrubbing to get rid of duplicate records, misspellings, different abbreviations for the same word and that sort of thing. Any database administrator goes through that anyway for any kind of reporting, but it doesn't really benefit the front end.
We coach our accounts to take that additional step and figure out what they want first. It shortens the implementation time and increases the chances of success.
GCN:'How do you gauge the user friendliness of your software?
BABCOCK: People mistake user friendliness for familiarity. If somebody is familiar with something, they tend to think it's easy to use. What I ask is, When was the last time you completed a job using technology and got it right the first time through? Good question, isn't it. More often than not, the answer is never.
What they're saying is that, each time they do something using technology, it's an iterative process. User friendliness should be the ability of the solution provider to reduce the number of iterative steps and the amount of time it takes to get the job done.
GCN:'How different is your work with state and local governments from what you do with federal agencies?
BABCOCK: There's a lot of trickle-down. The states use the federal government as a lighthouse. I think the federal government does a lot of really good things from a legislative perspective and a direction perspective. The states take their direction in response, both legally and because they are operating in the same environment with the same missions. They model their organizations after the broad federal organizations'treasury, welfare, health care, environmental programs.
From a technology perspective, they're similar. The issue they have is not enough money. Most state governments have far fewer resources to implement things, so they rely more heavily on third parties to do things for them.
GCN:'Outsourcing is greater in state and local governments?
BABCOCK: It's much hotter. The federal government has used systems integrators forever. It has done data warehousing forever, but it just decided to give it a good name in the '90s. People got away from terms like decision support in the late '80s because it was too technical, so they called it data mining. There's nothing new, it just has new clothes.
GCN:'What do you see as the future of Web-enabling enterprise applications?
BABCOCK: It's completely affecting the direction of our development. Version 7 of our software can make just about every tool provide information on the Web. Version 8, which will be released in October or November, completes the circle. We develop in waves, for certain platforms and certain products first because they're our bread and butter. Then we catch the others up in subsequent releases so they're all at the same level. Version 8 will make every solution publish information to the Web simply and easily.
We're developing portals, which are the wave of the future for interfaces. They're more structured toward the individual who is viewing information. Every organization wants a series of portals for various applications and users. That will be our next big development cycle.
We support the Common Gateway Interface, Hypertext Markup Language and Java. Back in the old days, five or six years ago, if it took as long to process a job as it does now, people would scream and yell. But the public is used to the Web being slow.
There's a development rule: If a Web application is going to take more than 30 seconds to process a request, you're going to lose the user. There's not much that takes more than 30 seconds with anything you would create with Java.
GCN:'What's the biggest challenge of getting into data warehousing and mining?
BABCOCK: Consensus. You have to figure out what information you want. People over time have collected output data about completing tasks. What they might not be prepared to collect is what GPRA wants: the outcome'the reducing of time from point A to point B.
What's more
- Age: '29 for the rest of my life.'
- Family: Wife of 18 years, Holly, and teen-age daughters Christine and Dianne
- Pets: Miniature pinscher named Rusty, Labrador mix named Toby, orange tabby cat named Tigger, Maine coon cat named Fur Ball and calico cat named Cally
- Sports and leisure activities: Golf, landscaping, cooking and reading
- Worst job: 'No such thing.'
- Motto: 'All my life's a circle.'