Productivity Paradox Logo
The IT Productivity Paradox

Ethan Dreyfuss, Andrew Gadson, Tyler Riding, Arthur Wang

"Failing" at IT

To err is human, but to really foul things up requires a computer. -Anonymous

One genuine cause of some of the apparently paradoxical numbers behind IT's performance is the ease with which IT investment can be done wrong. While recent number make it clear that IT does, on the whole, increase productivity, there are plenty of mistakes that can turn IT into an epic failure of a type that is difficult to accomplish with more traditional captial investments. There are a number of factors that combine to make IT investments much more prone to failure than other types of capital investment:

Lack of Transparency

This is the same problem which plagues electronic voting systems. IT systems are complicated, and it is not always immediately apparent how, or even if, they work. Subtle things, such as the corruption of backup copies of files, can go without notice until the backup is called upon in an emergency. This can result in either a lack of trust in IT systems or on the other extreme a sort of blind faith in IT systems.

Easy to Mislead Self and Others

Particularly with customized IT systems, it is often difficult to gauge real progress because a full test of the system may be prohibative and because good metrics often require actual use. One example of this in practice was the FBI's Virtual Case File system. Hundreds of millions were spent on its development, but due to project mismanagement, deceptive practices by a contractor, and a host of other software process flaws a usable system was never produced.

The Statistics

Attempts to do a by-the-numbers accounting for IT failures have resulting in widely different numbers. One major study, The Chaos Report from 1994 found that 31.1% of projects will be canceled before they are completed and that average cost overruns are about 189%. A more recent review of that report finds that their numbers are likely to significantly overstate results for typical cost overruns both due to criticisms of the methods used in the original report, and due to the high variance present in software projects.