Speaking at a Washington conference on government technology, the Gov 2.0 Summit, this morning, federal Chief Information Officer Vivek Kundra acknowledged that to achieve the transparency the Obama administration has promised there would need to be upgrades to many agencies' antiquated IT infrastructures. Delivering those upgrades on the other hand, may be a task that outlasts both Kundra and the administration.
During his conversation with Sunlight Foundation Executive Director Ellen Miller, Kundra touted the administration's efforts to post government data online via sites like Data.gov and the IT Dashboard on USASpending.gov. However, when Miller pointed out that USASpending is often slow to deliver data, Kundra responded that agencies have been held back by their aging IT infrastructures. He argued that systems based on decades-old technology would be unable to handle the increased demand that would result from posting data online.
Of course USASpending's troubles started well before Kundra began his job. The site still does not comply with the 2006 Federal Funding Accountability and Transparency Act, which mandated its creation. Bush administration officials told Nextgov last year that the main barrier to posting contract data online was the uneven quality of data submitted by agencies and cultural resistance within agencies to openness. Kundra reiterated those points today:
We want to get data out there on as real-time a basis as possible. When you think of the data supply chain, though, there are far too many people that want to clean up the data before it goes out there.
While it's refreshing to hear Kundra acknowledge some of the roadblocks to his ambitious plans, finding a solution will be significantly more difficult. He is correct that many agencies still rely on aging technology such as COBOL, but in most cases the reason they do so is because those systems continue to function after decades of use and agencies lack the necessary resources to replace them.
Given the current budgetary climate, it will be difficult to convince an agency to allocate a significant amount of funds to replace a working system with a newer, more expensive system whose main benefit would be to increase transparency.
The Farm Service Agency, for example, has well-documented deficiencies in the system it uses to process payments to farmers. The agency relies on COBOL-based, green screen technology and has even experienced outages because of the excessive demands on the system. But FSA had to fight tooth and nail just to obtain sufficient stimulus funds to stabilize its system earlier this year. A long proposed upgrade is underway, but will take several years to complete.
Judging by the dozens of interviews I've conducted with federal IT officials and employees over the past couple years, this type of situation is hardly unique. Additionally, the types of massive infrastructure upgrades Vivek is talking about would require years to complete and involve enormous risk. Like many things in the government, IT systems generally only receive attention when something negative has happened. So as long as a system continues to function as intended, agencies are unlikely to seek upgrades in favor of more pressing budgetary concerns.
One example is the Defense Finance and Accounting Service, which processes approximately two-thirds of the federal payroll. According to former Veterans Affairs deputy CIO Ed Meagher, DFAS relies mostly on COBOL-based applications, meaning the agency is running a 25-year old platform and a 30-year old application. Meagher said the DFAS system has remained in place because it works and it would take hundreds of millions of dollars to replace it, without taking into account the potential disruptions caused by migrating thousands of federal employees to a different payroll system.
So unless Congress -- with OMB's blessing -- is planning to deliver the dollars necessary to upgrade aging IT systems at agencies like DFAS and FSA while ensuring the vital services they provide won't be interrupted, we're not going to hold our breath and wait for any large-scale improvements in the federal IT infrastructure. Hopefully as agencies update systems individually, transparency and accountability will increase. But as I've been told many times, federal IT is an aircraft carrier not a speed boat. Turning it around will take a lot longer than one year, one term or even one administration.
COMMENTS
I disagree with the statement that the infrastructure that many finance and accounting systems run on is 25 years old. Like any other computer based environment, mainframe platforms have modernized with the times and seen an exponential growth in both capapability and function. Successfully combining a well maintained and modernized mainframe application with today's processing capabilities (e.g. web enabling, dynamic database updates from online transactions, etc.) is key to cost effective modernization efforts. As I understand it, this has been accomplished by companies such as UPS, Progressive Insurance and Bank of America. DFAS & DoD on the other hand have more typically attempted to create a complete replacement of finance and accounting applications based on some 'out of the box' software product. These attempts to create a common environment for disparate, non-standard applications have largely failed over the past nearly 20 years. The hundreds of millions of dollars and untold man-hours expended on these failed modernizations could likely have provided the necessary support to upgrade all of the finance and accounting 'legacy' applications. So yes, it will take more money and time, but we also need better direction, and we need to stop telling Congress what we think they would 'like' to hear and tell Congress what they 'need' to hear.
Robert Pritchard 09/15/09 11:45 am ET
Aging IT infrastructure is no excuse for delays in providing relevant and timely data online. When I was the CAO of the House of Representatives my team took handwritten ledgers and turned them into online finance reports within eighteen months. Simultaneously, they completed a total upgrade of the worst IT infrastructure in the federal government and turned it into a highly workable state-of-the-art system.
Scot Faulkner 09/11/09 01:47 pm ET
How does Vivek's explanation account for the discrepancy between the USASpending.gov and FedSpending.org data? USASpending.gov was inspired by and largely based on technology developed by OMB Watch's FedSpending.org. The two services continue to operate in parallel, and it is my understanding that 1) the data in their databases diverge, and 2) FedSpending.org's data is more comprehensive and accurate. The latter point may be controversial or even incorrect. But, in any case, we need to get a clear explanation for the discrepancy--and I'm not convinced that the tech-based excuses highlighted in this article (when gov. doesn't want to do something it knows is popular, it loves to blame technology) can explain it, however plausible they might initially sound. I'm more inclined to go with the secondary narrative not fully developed in the article, which is that agency officials would prefere not to release this information regardless of cost or technology.
--J.H. Snider, President
iSolon.org
J.H. Snider 09/11/09 09:43 am ET