My blog on a federal computer project that went six times overbudget prompted an expert on information technology (IT) to send me an interesting email.

In this study on government failure and this study on cost overruns, I discussed some of the reasons why $100 million projects end up costing $200 million.

Bob Charette, who is a contributing editor to IEEE Spectrum, has related views based on his decades of federal and private technology experience. He pointed me to some very useful articles he wrote on costly IT projects, including federal projects. He’s also penned numerous articles on chronic problems in the military acquisition system.

In an email, Bob provided some thoughts on why we often see delays and cost overruns on federal technology projects. If I am summarizing accurately, he observed:

• Government IT projects are often harder than commercial projects, so government projects may have more opportunities to fail than commercial ones.
• It is much, much harder to get an IT project approved in government because of all the hoops and stakeholders involved. So once it is, keeping the program alive is “Job One” for everyone involved. That also helps motivate vendors to not only bid low, but also government acquisition folks to accept unrealistic low bids, as well as put out unrealistic requirements because they believe shortcomings can always be fixed later.
• Part of the problem is “Hubble Psychology,” which the NASA IG defined as an “expectation among NASA personnel that projects that fail to meet cost and schedule goals will receive additional funding and that subsequent scientific and technological success will overshadow any budgetary and schedule problems.”
• For federal employees, to proactively identify risks (i.e., potential problems) is the kiss of death not only for individuals, but for the projects themselves. You don’t want to publicly admit there is uncertainty associated with your project without getting hammered by Congress and much of the press. So risks are actively hidden until not only do they become problems, but full-blown crises.
  

• Another issue is that success in government IT projects is often amorphous. In the private sector, profit and loss are binary and easy to measure. But in government, success is measured against one or more (sometimes competing) values that are rarely objectively measureable.
• Since government IT projects are typically long-lived, objectives morph so it is not uncommon for no one to remember what the project was originally supposed to actually do, or the original sponsors to even be around.
• Folks on Capitol Hill like to play chief software designer/architect and specify government IT systems that are difficult, if not impossible, for the dollars and time allocated (e.g., Real ID was a real doozy in this regard).  
• Research has found that if an IT project encounters a cost overrun of, say, 20 percent halfway through the schedule, then the cost overrun at the project’s end won’t just be double (40 percent), but far more than that.

In June, the Government Accountability Office reported on the government’s $80 billion annual investment in IT, and found, “investments frequently fail, incur cost overruns and schedule slippages, or contribute little to mission-related outcomes.”

Bob’s observations point to some reasons why that is the case.