Materiel Command and the Materiality of Commands: An Historical Examination of the US Air Force, Control Data Corporation, and the Advanced Logistics System - History of Computing: Learning from the Past - page 98

not surprising that the AFLC sought to be a first-mover in applying a massive unified
database, and a network of computers operating in real-time, to centralize the task.
Like many other organizations of the period, it learned the hard way the great diffi-
culty of managing massive computer programming and system integration projects.
This paper also stands as an all too rare study of technological failure, combating
the “progress talk” that frequently dominates histories of technology—a problem ar-
ticulated by historian John Staudenmaier more than two decades ago [34]. “Progress
talk,” or presenting technology's history as continual and unwavering progress, holds
an even greater stronghold in the history of computing where hardware progress has
been quantified as if it were a scientific law (Moore's law—chip capacity/processing
capability doubling every year or 18 months) and software, growing to fill ever
cheaper memory and add functionality, is frequently perceived as following suit. 19
ALS is a case where project design extended too far out on the technological frontier,
and software (with the central control system in COBOL rather than machine code)
presented limitations on system integration that could not readily be overcome.
Failures in oversize, massively complex, networked computing systems, such as
the FBI's abandonment of a $170 million dollar system in early 2005, continue to
persist. At the time of the September 11, 2001 terrorist attacks, FBI computing and
networking were horribly antiquated. Using a mainframe operating system three dec-
ades old, some field offices were without network connections and unable to transmit
digital images of terrorist suspects. A major computer system project was quickly
launched to create a networked system, Virtual Case File, to replace the FBI's paper
files and aid with tracking criminal cases [35]. Primary contractor, computer services
firm Science Applications International Corporation (SAIC), wrote more than
730,000 lines of code and received more than $100 million as the project grew in-
creasingly complicated. SAIC continued to “meet the bureau's requests despite clear
signs that the FBI's approach to the project was badly flawed,” according to individu-
als involved with the project and those who later reviewed it for the government [35].
With serious security problems and high error rates, the FBI shut the project down. In
a response eerily similar to statements by General Catton and Donald Rumsfeld's in
the mid-1970s in aftermath of the ALS failure, FBI Director Robert Mueller took re-
sponsibility for “not having put appropriate persons in a position to review… [the]…
contract and assure that it was on track.”[36] Further, he indicated that “the FBI
would now start from scratch, and look for a more updated, flexible system using off-
the-shelf software.”[36] Similar difficulties have occurred with recent large-scale
networked computer systems for air-traffic control, electric energy grid management,
and other major IT projects both within and outside of government. These develop-
ments clearly suggest that lessons from failures with the design, development, imple-
mentation, and oversight of massive computer networked management systems are
yet to be fully learned. 20
19 Moore's Law, initially a speculative prediction by Gordon Moore in 1965, evolved into a
company (Intel) and industry-wide (trade association) benchmark. As such, it became a
managerial tool shaping investments and outcomes.
20 One potential lesson is to forego new, path breaking, massively complex, custom systems in
favor of existing state-of-the-art systems with proven performance.