At I.R.S., a Systems Update Gone Awry

["Paul Tiger" <tigerp@xxxxxxxxxxxx>]

My point in sending this to the CVV list is that this is another example of
what happens when the government mandates computer projects. Its also going
to the Boulder LP list, and that does not need explanation.
Paul Tiger
-----
December 11, 2003

At I.R.S., a Systems Update Gone Awry
By DAVID CAY JOHNSTON for the New York Times

After five years, a project to replace the Internal Revenue Service's aging
file-keeping computer system with modern technology is so far behind
schedule that the I.R.S. has told the prime contractor that unless it
improves its performance by the end of the month, the government may have no
choice but to fire it.

The project, which was expected to cost $8 billion when completed, has spent
less than $1 billion so far, but it is already 40 percent over budget for
what it has done, according to the I.R.S. Oversight Board, an independent
watchdog body that Congress created in 1998.

Most taxpayers are younger than the computer system that the I.R.S. relies
on to maintain its master files on individuals and businesses - all the
records of who they are, where they are, their income, taxes paid, and the
amounts they still owe or are owed as refunds.

The I.R.S. says it can still process returns and send out refunds on time,
but its dependence on the 1960's-era Assembler and Cobol computer languages
makes it difficult to investigate and resolve taxpayers' problems. Finding a
record using the existing system can take a week; the new system is supposed
to do the job in seconds.

"This is not about a one-time delay," said Larry Levitan, chairman of the
Oversight Board. "Every single major project under way experienced a
significant delay in time and overrun in budget - not two or three out of
five, but five out of five. What we have here is a five-year track record of
absolute consistency of cost overruns and delayed deliveries."

Big computer modernization projects often run late and cost more than
anticipated. But even given the size of a system for the I.R.S. - one that
must keep track of 200 million taxpayers and an increasingly complex tax
code - the project is not succeeding, according to the board and to senior
I.R.S. executives. The contractor, the Computer Sciences Corporation of El
Segundo, Calif., must show improvement before the end of the year or face
losing the contract, they said.

"If they don't produce we will make a change," Mark W. Everson, the I.R.S.
commissioner, said of the contractor, even though experts at Carnegie Mellon
University in Pittsburgh said that starting over with a new company would
"probably result in different but no fewer problems along the way" - and
delay the new system, which is called the Customer Account Data Engine, by
two or three years.

"I would not enter lightly into rupturing the relationship," Mr. Everson
said. "It is not a desirable outcome to abandon the relationship, but that
does not mean we won't do that if we have to."

Paul M. Cofoni, president of the Computer Sciences unit running the project,
CSC Federal Sector, said "in the early part of the program we did a poor job
of defining" what needed to be done. But that was in large measure because
the I.R.S. had no records of many changes to its old system, he said, and
was reluctant to approve specifications for the new system until it could be
sure the system would be able to find and display all the old information.

Mr. Cofoni said that many of those problems were being addressed.

"I can actually see daylight now," he said in a telephone interview. "We
were given an action list of 46 items to be done in 30 days, and 85 percent
of them were. We're at the point where we are starting to deliver, and when
we're done people are going to say this is an outstanding, award-winning
system."

In a report being distributed to the Bush administration and Congress, the
Oversight Board said that it had not seen improvement in three years, and
added that Computer Sciences' performance "must be monitored very closely
and if significant improvements are not demonstrated quickly a change should
and must be made."

Mr. Levitan of the Oversight Board said that the project was "losing
credibility with Treasury, with the Office of Management and Budget and with
Congress."

Five years into the project, some aspects are as much as 27 months behind
schedule.

While the project to modernize the main file-keeping computer has
encountered serious problems, other technology projects have worked,
including a system developed by Computer Sciences that tracks the status of
refunds and quickly routes calls from taxpayers to appropriate people to
answer questions. Mr. Everson said this had allowed him to put more I.R.S.
executives on the troubled project, although, as a result, the agency had to
set aside ancillary modernization projects.

Mr. Levitan and others said that Congress needed to let the I.R.S. hire more
executives who understand computers. Mr. Levitan said the agency relied too
heavily on a single executive, Fred L. Forman, for computer management
expertise. Dr. Forman, a former executive with American Management Systems,
joined the I.R.S. in the middle of 2001 as an adviser to the modernization
project and now serves as an associate commissioner.

"The I.R.S. needs 10 to 15 Fred Formans," Mr. Levitan said. "They have got
some good people, but they don't have nearly enough to manage the program."

Major corporations often upgrade their systems as technology improves. The
I.R.S. went four decades with the same system because two previous
modernization attempts, the most recent in the mid-1990's, failed, costing
taxpayers $4 billion. Much of the problem involves the risks associated with
moving from one system to another. The current plan, begun in 1998, was to
build the new system, import data and then turn off the old system.

But Charles O. Rossotti, the tax commissioner from 1997 through 2002, found
that approach fraught with danger. Mr. Rossotti, the founder of American
Management Systems, who was brought into the agency after the earlier
modernization efforts failed, wanted to keep the old system going as data
was moved to the new system in segments, beginning with the simplest tax
returns, the one-page Form 1040EZ's, to insure reliable access to taxpayer
records.

Mr. Levitan said that Mr. Rossotti brought technological coherence that has
averted disaster. But he also says a collapse is inevitable without a new
system, because the few people who could keep the old system functioning are
close to retiring.