Billion-dollar IT failure at Census Bureau
Summary: The US Census Bureau faces cost overruns up to $2 billion on an IT initiative replacing paper-based data collection methods with specialized handheld devices for the upcoming 2010 census. The Bureau has not implemented longstanding Government Accountability Office (GAO) recommendations and may therefore be forced to scrap the program. Harris Corp., the contractor associated with this incompetently managed initiative, was awarded a $600 million contract to develop the handhelds and related software.
The US Census Bureau faces cost overruns up to $2 billion on an IT initiative replacing paper-based data collection methods with specialized handheld devices for the upcoming 2010 census. The Bureau has not implemented longstanding Government Accountability Office (GAO) recommendations and may therefore be forced to scrap the program. Harris Corp., the contractor associated with this incompetently managed initiative, was awarded a $600 million contract to develop the handhelds and related software.
In March 5, 2008 testimony before the Senate, Commerce Secretary Carlos M. Gutierrez said: "There is no question that both the Census Bureau and Harris could have done things differently and better over the past couple of years."
On the same date, Census Bureau Director, Steve H. Murdock, added:
I cannot over-emphasize the seriousness of this problem. My colleagues and I recognize that we must move quickly to address this problem, and implement solutions. While we still have an enormous challenge in front of us, I am confident that we are close to defining and implementing a strategy that will ensure a successful 2010 Census.
The GAO characterized the handheld initiative, known as the Field Data Collection Automation (FDCA) program, as follows:
Of the $11 billion total estimated cost of the 2010 Census, the Census Bureau planned (as of 2007) to spend about $3 billion on automation and information technology in order to improve census coverage, accuracy, and efficiency. Among other things, the Bureau is planning to automate many of its planned field data collection activities as a way to reduce costs and improve data quality and operational efficiency.
The GAO report, dated March 8, 2008, added:
In October 2007, GAO concluded that without effective management of key risks, the Field Data Collection Automation (FDCA) program responsible for the devices faced an increased probability that the system would not be delivered on schedule and within budget or perform as expected. The magnitude of these problems is not clear.... [T]he Bureau has not performed recommended analysis or provided sufficient information to provide a level of confidence in its $11.5 billion life-cycle cost estimate of the decennial census. The Bureau has not itemized the estimated costs of each component operation, conducted sensitivity analysis on cost drivers, or provided an explanation of significant changes in the assumptions on which these costs are based. Together, these weaknesses and actions raise serious questions about the Bureau’s preparations for conducting the 2010 Census.
Computer World blogger, Frank Hayes, summarized the situation succinctly, "The fancy custom handhelds might work. But if they don't, the Census Bureau will use paper instead."
THE IT PROJECT FAILURES ANALYSIS
Managing an $11 billion initiative is a daunting task and unforeseen problems are inevitable. Nonetheless, the GAO, going back to January, 2005, repeatedly identified significant procurement, management, and operational risks associated with this project. For reasons unknown, the Census Bureau chose not to follow these recommendations.
The following table summarizes significant project issues identified by the GAO:
How does a failure of this magnitude arise? Clearly, Census Bureau management is ineffective at properly and efficiently executing the organization's basic mandate. A detailed analysis would probably reveal hidden agendas; conflicts of interest; good intentions gone bad; inexperienced, lazy, and incompetent management; lack of controls; and plain old poor judgment. I believe these deeply ingrained issues are symptomatic of fundamental problems shared by both Bureau leadership and line management.
My recommendation: The GAO must conduct a formal inquiry into two specific areas:
- It should investigate and analyze the management policies and procedures that allowed this situation to develop and persist over the course of several years. We must understand why program controls didn't prevent this huge waste of dollars.
- It should perform a detailed (and I mean exhaustive) investigation of Harris Corp.'s role. Let an unbiased panel determine what percentage of the billion-dollar waste Harris caused and force the company to pay direct restitution for that amount.
Until the government holds contractors and their agency sponsors accountable, massive failures will continue and more money will be flushed down the drain.
Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.
Talkback
Amateurs
The FBI SAIC project was another one. The electronic case files. ROFL! Maybe someday.
There are more. IRS was struggling with a bad project for a while, not sure if they got that one together yet.
Those are just the big ones. Burning 20-30 million on a busted project is nothing these days. No accountability. I've seen the people wasting that kind of coin actually get promoted. They know the IT buzzwords of the week and are well connected. They'll work a project for a couple years and before it becomes overly obvious they're incompetent, they move on.
There's going to be waste in any big system but what we have today in government IT is massive incompetence coupled with a corrupt system with no accountability.
The final point touched on....
The solution is in the govt.s hands. It is called "fixed price contacts". If the "big boys" don't like the price then I'm sure many smaller firms will step up to the mark.
Whilst there will be capacity and performance issues many large or multinational companies implement equally large or performance critical systems for a fraction of the cost of these govt. IT systems.
I worked as an independent contractor alongside a well known top accountancy firm and one of the big US software and systems companies. Their staff asked *me* for technical advice and spent eight months writing reports and attending meeting whilst charging well over ?1200 per day per "consultant" (was in 2001). No [i]productive[/i] work came out of that period, just a refined requirements based on my initial survey and feasibility (that's what I was doing) Next they wanted to move to system design for which they budgeted 12 months and on that scale (using common rule-of-thumb metrics) they were digging in for another 3-5 years.
The system in question in no way merited that level of spend, but it kept plenty of staff employed and generated fat profits. Many of the "consultants" were fresh out of University, the ink on their degrees still wet and each small team of 5 or 6 had a senior manager (a few years experience) and a project leader (typically 5-15 years experience).
Actually, that's not quite the case.
It's hard to stay on budget when you don't know what they want.
I won't say that contractors never try to milk a project, but I am going to have to fall more into the "hard to hit a moving target" camp. It is nausiating how often we have been given vauge, hand-wavy directions from government customers. This leaves us having to spend extra time to develop the requirements for them. Likely due to their not taking ownership of them, they see no problem with changing requirements, heck, even [i]concepts[/i], part, or even most, of the way through a project. One fellow is notorious with engineers I have worked with for forcing design changes faster than [u]Hitchhiker's Guide[/u]'s Heart of Gold coming out of Infinite Improbability Drive. Is it any wonder when projects for him ran late and over budget?
That is why I don't see fixed price contracts as being the panacea to government contracting. Unless you can get the government to be [b]absolutely clear[/b] about what is being contracted, up front, [i]and don't allow them to change their minds afterwards[/i], there is no way to negotiate a fixed price. The contractor either has to WAY overbudget to cover their behind for un-planned changes, go back to the government--hat in hand--whenever the requirements change (taking the [i]"fixed"[/i] out of fixed price), or close up shop and walk away from a failed project when they've spent the last dollar. This would be true, no matter how big or small the company.
I will admit to being concerned about government contractors "wasting" the tax dollars they yank out of my paycheck every week, but a lot of that concern is directed towards the government managers who don't bother to manage, be it risks, costs, requirements, schedules, you name it. Let's be honest, a government contractor, like any company, is trying to make a profit. If there is no oversight, no concrete metrics to be overseen, and no penalties when the non-existent metrics are underperformed (and no reward when they are outperformed, either), is it any surprise that some companies are willing to let projects flounder and keep accepting the government's check week after week? Don't take this as an excuse, as it disgusts me, too, but I do understand it.
If Harris agreed to something--AND that something didn't then morph into something else--AND they failed to meet the agreement, then sure, throw the book at them. I suspect, however, that it will be very hard to prove, because the "something" was either too ill-defined, [i]evolved[/i], or simply because nobody bothered to check that the agreement was being kept. Given the magnitude of this debacle, I doubt Harris will walk away unscathed, but I suspect little, to nothing will be done to the (mis)managers at the Census bureau who let the problem get this bad, and that is what troubles me most of all.
Deskilling of the civil service
The corollary of this folly is that the government technical ranks eventually migrate to greener pastures as they find advancement blocked within the civil service. The vacuum at the top is then filled with "professional managers" (you know - the ones "who can manage anything") who have no substantive system lifecyle experience. Since there is no one left who can define the system requirements, the key CI of an SDLC, the prospects are not good for taxpayers to get a reasonable return on their inflated investment.
The only way out of this mess that I can see is a "reskilling of the civil service", something that would take at least a generation, if a president were to come along who had the vision to undertake it.
Project Management & Scope Control
Your Tax Dollars at Work
Granted that while the requirements are not listed here, it certainly seems plausible that an app could be written for a smartphone or the symbol device to do the collection. Given the ability of those devices to use cellular systems to deliver the data to main collection systems, the process as a whole should be almost efficient in providing real-time monitoring, and reporting. Developing this system would certainly be more effective (cost and time) than trying to develop a new device.
But hey, the US government is rolling in dough, so this situation is perfectly acceptable, right?
Sounds like a great app for the iPhone! <NT>
RE: It sounds like a great app for the iPhone
Repeat of a Canadian episode
You know what they say about governments: You spend a billion here and another billion there, and eventually, it adds up to real money.
Moving Target?
Right!
Face it - the jobs that should be on the line are government jobs. The folks in charger there at the census bureau apparently aren't qualified to do the job. At least, not this job.
Here's a solution:
1) Define the requirements. This needs to be done regardless.
2) Fund an open source project to fulfill those requirements. NOW - you've given a paperless system to the world. AND - it won't cost billions to develop.
(It *may* cost billions to deploy, for hardware and personnel, as the census takers are paid, but those *could* be able-bodied welfare recipients going door-to-door!)
-Mike
RE: Billion-dollar IT failure at Census Bureau
Good point!
-Mike
RE: Billion-dollar IT failure at Census Bureau
RE: Billion-dollar IT failure at Census Bureau
RE: Billion-dollar IT failure at Census Bureau
RE: Billion-dollar IT failure at Census Bureau
Sorry, that dog doesn't hunt
How many of these failures...
You get what you pay for.