GovernmentExecutive.com

The Census Bureau's program to develop handheld computers to collect data on Americans during the 2010 census is in "serious trouble," and the bureau should immediately develop plans to use paper forms, according to an analysis conducted by a government research firm and presented to top bureau managers in late 2007.

The bureau plans to issue more than 500,000 handhelds to temporary employees to collect personal data on Americans who do not return census forms in the mail. The handhelds are being developed under a $600 million contract awarded to Harris Corp. in 2006.

But Mitre Corp. officials who met with Jay Waite, deputy director of the Census Bureau, said the agency has experienced so many delays and cost overruns in developing the handhelds that it should immediately develop a contingency plan to use paper forms.

Mitre, which the bureau hired in 2004 to periodically advise it on information technology programs it is developing for the 2010 census, met with Waite Nov. 29, 2007, at Census Bureau headquarters in Suitland, Md., to discuss the progress of the handheld computer contract, called the Field Data Collection Automation contract.

Mitre officials told Waite that "FDCA is in serious trouble," according to a talking points document Mitre prepared for the meeting that Government Executive obtained. "It is not clear that the system will meet Census' operational needs and quality goals. The final cost [of the contract] is unpredictable. Immediate, significant changes are required to rescue the program. However, the risks are so large considering the available time that we recommend immediate development of contingency plans to revert to paper operations."

The Mitre document heightens already serious concerns about the bureau's preparations for the 2010 census. In reports over the past couple of years, the Government Accountability Office has warned Census officials about the consequences of falling behind schedule in developing the handheld computers and has questioned efforts to manage the project's risks.

The Mitre document addressed the more grave scenario that the handhelds will not work as planned. It indicated that the bureau now has reached a point at which it cannot develop all its requirements for the handhelds in time for the 2010 census -- even if it could increase the program's funding. "Money cannot trump time remaining," the document stated.

"This is extremely strong language that cannot be ignored," said Robert Charette, a risk management expert who consults with federal agencies. "The document lays out that [the Census Bureau has] run out of time. It would be professionally foolish not to have a contingency plan in place to meet [Mitre's] concerns and the resources allocated to make that happen."

Census officials said via e-mail that they "have no plans to revert to paper. Also, at this stage of the decade, such a major change in plans for key operations would pose its own set of significant challenges and risks."

They acknowledged that managing the handheld computer contract presents "significant challenges," and said they are monitoring and reviewing "all aspects of the contract in order to ensure a successful census."

The problems Mitre cited in its document included "significant" inconsistencies between the bureau and Harris about when certain deliverables for the handhelds were due. It also noted that requirements for the program had not been fulfilled, and questioned whether remaining requirements could be completed in time. Mitre also claimed the FDCA program "lacks a leader with the experience, stature and passion to make FDCA successful," adding the "Census Bureau has a lack of personnel with large-scale IT program management experience."

Mitre officials could not be reached for comment.

David Powner, director of IT management issues at GAO who has been auditing IT systems the bureau is developing for the 2010 census, said he had seen the Mitre document and agreed with its conclusions. He noted that GAO has warned Census that it had not clearly defined its requirements or developed a defined test schedule, including setting up a trial to test data collection and processing using the handhelds from end to end.

"There's not a real solid plan for what is going to be tested when in the 2009 and 2010 time frame," he said.

But Powner said GAO was not ready to advise the agency that it should consider going back to a paper and pencil method. "Clearly, you want to have contingency plans," he said. "That's not a surprise. But recommending a complete paper census? I do not know that we're there yet."

Mitre recommended the bureau work with Harris to create a baseline of what it can accomplish before the 2010 census to determine the full requirements for the handheld computers and their cost. The document also recommended that Census appoint a manager to "ruthlessly implement" the program's schedule, complete a plan to test the system from end to end (and not squeeze the schedule to save time and money); and streamline bureau management reviews of the contract.

If the Census Bureau cannot use the handheld computers, it is unclear how much more the 2010 census would cost. Its budget is now estimated at about $12 billion. In 2001, when Waite first conceived the idea of using handheld computers, the bureau argued that they would save billions of dollars that otherwise would be spent on printing, storing and transporting millions of maps and hundreds of millions of census forms. (See "On the Brink," Government Executive, July 15, 2007.)

COMMENTS

• I think that the previous comments are based upon clichés and generalities and a lack of knowledge about decennial censuses. They do not credit the actualities of the decennial census projects. I have a great deal of respect for the Census Bureau and its senior staff. For several decades they have succeeded in meeting a Constitutionally required deadline, accomplishing what has, in each decade, been at the time perhaps the largest, most complex information processing project in history. Furthermore they have done so under a governmental system that, in many ways, undermines their work. In particular, they are funded by Congress on an annual basis, by Senators and Representatives on six or two year election cycles. A decennial census is about a 12 year project. Among the most fundamental principles we have learned over the past 60 years of the information systems history is that success in large projects depends very strongly on early planning, analysis and design. Congress, of course, quite justifiably prefers to postpone funding until after the next election. The difficulty of large information systems projects has been documented over and over during the past 60 years. The available technology, including hardware, software and to a lesser extent the methodology of system development, has changed more rapidly than any in history. They have worked in a field where exponential change is enshrined in "Moore's Law". During each decennial census project that comes out at about a 16-fold growth. Imagine building a dam on a twelve year project where the height of the dam increased 16-fold during the project! Meanwhile they have led the nation in the effective use of new technology, from punched cards to GPS navigation. During the past 8 years the Census Bureau has, via the American Community Surveys, eliminated the need for the “Long Form” part of the the 2010 Census. The “handhelds” will eliminate the production, storage and distribution of domicile level paper maps of the entire United States. All of this while suffering under budgets reduced by 9/11 and the funding of the Iraq war. They have, of course, made mistakes. But, they have succeeded in accomplishing their goals and doing so on schedule each time. I will bet that they successfully, perhaps not perfectly, but successfully use their “handhelds” to carry out the 2010 Census. C. A. ( Posted March 4, 2008 3:29 PM
• I found Mitre's recommendations a bit self serving (need thorough testing without regard to cost). This type of Analysis/paralysis is often what leads to failure, preventing the agency from focusing on delivering. I wonder how much Census paid for Mitre to state the obvious? FBI's Virtual Case File followed Mitre's recommendations to a "T" and it failed miserably. Same thing happened with the Navy Literal Ship Program, and DHS Deepwater. I think it is time for our FFRDCs to keep within their charter and not be able to make recommendations that lead to needless analysis/paralysis. Honest Broker Posted January 31, 2008 7:13 PM
• It sounds like the Census Bureau is being typical of someone not used to specifying requirements and then changing requirements half-way through. MITRE should have been on during requirements definition as an independent party. As usual, the contractor will get it done to save their reputation, but the bureau may have to chuck more money at it to fix their bungled process. Pdog Posted January 8, 2008 4:50 PM