We've been hearing about this problem for some time now, but like most
people we have been ignoring it. As with many problems, we clip
articles about it, then file them for later reference. It's the Y2K
problem. To a scientist, Y means Year and K means 1000, so Y2K refers
to the year 2000 problem. It's a computer problem with possibly-serious
environment and health implications.
Like most people, we are very suspicious of alarming predictions about
the year 2000. What finally focused our attention on the Y2K problem
was a small item in the back pages of the NEW YORK TIMES Saturday June
13th. It began, "The nation's utilities told a Senate panel today
[June 12] that they were working to solve expected computer problems
when 1999 ends but that they could not guarantee that the lights would
not go out on Jan. 1, 2000."
The utilities say the lights may go out. This seems like a problem
The TIMES went on, "An informal survey by a Senate panel of 10 of the
nation's largest utilities serving 50 million people found none had a
complete plan in case its computers failed because of the problem." The
TIMES explained, "Many electrical plants use date-sensitive software to
run built-in clocks that monitor and control the flow of power. These
could fail if not updated."
The utilities say the lights may go out, yet none of them has a full
contingency plan. How serious could this problem become?
As we examined the items in our "Y2K" file, we found opinions ranging
all over the place. Some people said, "This is a fake problem invented
by people who want to sell fixes." Others said, "This is going to be
the end of civilization as we know it." Where does the truth lie?
I worked 5 years in the Computing Center at Princeton University, so
have more than a passing familiarity with computers. My crystal ball is
as hazy as any one else's, but here is an attempt to offer a realistic
look at the nature of this Y2K problem.
Unlike most problems, we know when this one is going to hit us: on
January 1, 2000, just a little over 500 days from now.
Here is the crux: Many computers only recognize dates by two digits. In
these computers, 67 is 1967 and 98 is 1998. In these computers, a 00
date will mean 1900, not 2000, unless their software is re-written.
When such computers start calculating or comparing dates after 1999
they won't work right --they may simply shut down, or they may seem to
run fine but produce incorrect information that is very hard to detect.
Computers that have this Y2K problem are called "noncompliant"
computers, and it turns out there are quite a few of them.
Many noncompliant computers are the really big "mainframe" machines
that serve as the central nervous systems of financial institutions
(banks, savings & loans, credit unions), stock exchanges, air traffic
control systems, missile defense systems, government tax agencies, the
Social Security Administration, the Medicare program, the insurance
industry, and all of the Fortune 1000 multinational corporations. (And
of course this problem is not limited to the U.S. Every industrialized
country depends heavily upon large mainframe computers.)
A report published by Merrill Lynch, the financial management company,
says flatly, "When the millenium arrives, many computer systems and
global networks will fail because of an inability to properly interpret
dates beyond 1999."
Mainframes will not be the only computers to fail on January 1, 2000 if
they are still noncompliant by then. Many industrial machines
contain "embedded systems" --computer chips that are literally embedded
within some larger piece of equipment, such as power stations, oil
refineries, telephone switches, burglar alarms, emergency room
equipment, air traffic control systems, military defense gear, and
chemical plants, among others.
By the year 2000, there will be an estimated 25 billion embedded
systems, according to the Gartner Group, which advertises itself as the
world's foremost authority on information technology. By Gartner
Group's estimate, two-tenths of one percent of these 25 billion
embedded systems will be noncompliant. Two-tenths of one percent of
25 billion is 50 million. Therefore, the problem, according to Gartner
Group, is to identify and replace those 50 million noncompliant
embedded systems in the next 500 days. To solve this problem, someone
would have to identify, replace, and test about 100,000 chips each day
between now and December 31, 1999. Does the U.S. have enough
technicians to identify, replace and test 100,000 chips each day? It
These embedded systems tend to be embedded in the nation's core
infrastructure --in the water, sewage, and electrical utilities, in
railroads and other transportation systems, in hospitals, in police and
fire services, in the defense infrastructure, and in petrochemical (and
other manufacturing) plants.
BYTE magazine, a technical computer journal, wrote recently, "One
commonly cited problem is associated with gadgets that monitor periodic
maintenance. When the clock strikes twelve on New Year's Eve, 2000,
these devices might think it's been 99 years since their last
maintenance, realize that's too long for safe operation, and shut
Virginia Hick, who writes a column called "Technology and You" for the
ST. LOUIS POST DISPATCH recently interviewed Peter de Jager, a well-
known Y2K consultant to industry. (See www.year2000.com). Here is what
".... De Jager talked recently with an executive of a company that
makes a volatile gas --he would not identify the company more
specifically --who told de Jager how his plant discovered the
seriousness of faulty embedded chips.
"The plant found a chip that failed when the date was moved forward.
When the chip failed, it shut off a valve that would have shut down the
cooling system. A cooling system shutdown, the executive said, would
have caused an explosion.
"That was great news," de Jager said. "Because they checked --there
will be no explosion. They're replacing the chips."
"De Jager worries about the companies that are not checking," Hick
Conclusion No. 1: If we lived in a community with one or more chemical
plants, we would be asking our local government to hold public hearings
on the Y2K problem, seeking public assurances from local plant managers
that they really have this problem under control. What written plans do
they have for assessing these problems, and how large a budget have
they committed to solving them? What progress can they demonstrate?
Does the plant manager have sufficient confidence in the plant's safety
systems to be at the plant with his or her family at midnight December
31, 1999, to celebrate the new year?
Now let's return to the mainframe problem. Because non-compliant
computers could harm a company's financial picture (up to and including
bankruptcy), on January 12, 1998, the federal Securities and Exchange
Commission (SEC) issued SEC Staff Legal Bulletin No. 5, which requires
publicly-held companies to report their progress toward solving their
Y2K problems. On June 10, 1998, Steve Hock, president of Triaxsys
Research in Missoula, Montana, testified before the Senate Banking,
Housing and Urban Affairs Committee that his company had examined the
SEC filings of America's 250 largest corporations. Mr. Hock told the
Senate that 114 of the 250 companies had filed no Y2K information with
the SEC. Of the 136 companies that HAVE filed Y2K information, 101
reported their progress on the assessment phase of the problem. Of
these 101, 60% revealed that they have not yet completed their
assessments of the Y2K problem.
Mr. Hock testified that 36 companies reported their estimated Y2K
project costs and how much they had so far spent. The average company
reported having spent 21% of the expected total costs of Y2K fixes. Mr.
Hock concluded, "[The] data shows remarkably little progress by the
largest US companies in addressing the Year 2000 problem. Most of the
work has been compressed into an extremely tight window of time. Given
the information technology industry's long history of failure to
complete large scale system conversion projects on time, this is cause
for serious concern."
The New York Federal Reserve Bank has said that it will take more than
a year for a large corporation to test its computers for Y2K compliance
AFTER all their software has been fixed. This means all fixes must
be completed by September or October of 1998 so testing can begin in
time. But many large corporations are still at the stage of assessing
the problem, and it's now late June.
How big is the task for a complex corporation? State Farm Insurance --a
company that believes it is on top of the Y2K problem --began working
on the problem in 1989 and found that it had 70 million lines of
computer code to convert, 475,000 data processing items, more than 2000
third-party software programs, 900 shared electronic files, plus
miscellaneous telephone and business equipment in 1550 corporate and
regional service facilities. State Farm still has 100 employees
working "around the clock" on nothing but Y2K.
But even a forward-looking company like State Farm could be harmed by
this problem if its customers, suppliers, partners, bankers, and
regulators aren't compliant by the year 2000. As Merrill Lynch
says, "Even institutions that have fixed their own internal problem
will feel the ripple effects from problems occurring externally."
A survey of small businesses by the National Federation of Independent
Businesses (NFIB) reported June 1 that 75% of small businesses have
done nothing about the Y2K problem. The NFIB estimated that 330,000
small businesses will go bankrupt and another 370,000 will
be "temporarily crippled" by the Y2K problem.
Conclusion No. 2: Portions of the nation's basic infrastructure
(utilities, transportation, defense, manufacturing) seem likely to be
disrupted by the Y2K problem. Furthermore, parts of the world's core
commercial institutions, such as banking and insurance, seem likely to
be disrupted by the Y2K problem.
Therefore, in our opinion, we each would do well to ask ourselves: if
the electric utilities may not be reliable, the petrochemical industry
(which delivers our gasoline) may have difficulties of its own, the
trains may not run well, and the world banking system may be plagued by
errors and glitches, how can we be sure that our employers will be able
to pay us so that we can put food on the table? It even seems as if we
should be asking, how can we be sure there will be food in the grocery
stores? Given what we know, these seem to be reasonable questions.
More next week.
 "National News Briefs; Utilities Say Outages Are Possible in 2000,"
NEW YORK TIMES June 13, 1998, pg. 16.
 Thanks to Roleigh Martin for the Gartner Group estimate. See
most comprehensive --and most pessimistic --web page on Y2K is that of
historian Gary North: www.garynorth.com
 Edmund X. DeJesus, "Year 2000 Survival Guide," BYTE (July 1998),
 Virginia Hick, "Expert Warns Computer World is Running Out of Time
to Meet 2000; Code is Broken and Needs to Be Fixed Fast, He Says," ST.
LOUIS POST-DISPATCH Nov. 19, 1997, pg. C8.
 Mr. Hock's testimony is available at
Descriptor terms: computers; accidents; chemical plant safety;
explosions; fires; y2k problem; chiliasm; millenarianism; merrill
lynch; embedded systems; gartner group; sec; gary north; roleigh martin;