Teaching Computers to Tell Time

At first glance, it might seem a bewilderingly fast way to spend $640,000 in scarce school funds, especially because the results won’t put an extra penny into Los Angeles classrooms.

Nevertheless, the Los Angeles Board of Education first anted up $90,000 for a four-week study. When that was done, it plunked down $550,000 more for an eight-week assessment.

What it has gotten for its money so far is a set of illustrations showing what still must be done and a shocking bottom line: In the next two years the Los Angeles Unified School District will have to spend between $30 million and $52 million.

For what? Simply so that its computers will know what year it is when the clock turns at midnight from Dec. 31, 1999 to Jan. 1, 2000.

And that’s no joking matter. A computer that doesn’t know what date it is could pay 30,000 teachers too much, or too little, or not at all. It could mess up student test scores and blow critical deadlines, such as the due dates of vendor bills. The reprogramming of computers so that they recognize a new millennium is a crisis costing billions of dollars to business and government around the world.

But the urgency that compelled the board last week to approve a consultant’s breathtaking fees--without uttering a word about time-consuming competitive bidding--is testimony to how much more vexing the problem can be to a cash-strapped and sluggish bureaucracy.

The Los Angeles Unified School District is not alone in the challenge. From the IRS to Los Angeles County, the slowly dawning enormity of what is called the Year 2000 Problem has caused costs to escalate beyond comprehension.

The state of California’s preliminary estimate of $50 million just 18 months ago had doubled by early this year and has now nearly doubled again at $187 million, said Rich Halberg, assistant director of the department of information technology.

In Los Angeles, both city and county officials say they have potential cures in sight, though they remain a good year behind similar-sized businesses.

But the school system has hardly budged, hampered by recent deprivation and a culture averse to tossing out tangible assets.

“We’re still repairing Apple II’s,” said John Nagata, assistant superintendent for data processing, referring to computer systems from the 1980s that many businesses consider out of date.

Having fretted two years over what to do, unable to budget a penny to do it, Nagata was at last forced this summer to toss up his hands in surrender as business leaders stepped in to avert what they said was an impending disaster.

The roadblock for government agencies is that there seems to be so little return for the money, said Bob Abeles, chief financial officer for Transamerica Life Insurance and an advisor to the district who pushed the alarm button about the problem earlier this summer.

“It doesn’t do anything for you but keep you running. So you sit there hoping there is some magic solution that will keep you from having to spend this money,” he said.

In the long run, the delay only made the job more expensive as computer programmers got snapped up by business and other institutions that took the lead, creating a seller’s market.

For district officials who grasped the problem early on--those in the politically scrawny information processing departments--the challenge was to persuade their superiors that the problem deserved priority. Unfortunately, their story didn’t ring with the drama of students who can’t read and teachers denied pay raises for seven years.

Essentially, it boils down to a trick that programmers used to conserve precious memory in the not-so-distant days before low-cost hard drives with gigabytes of storage. Instead of representing the year with four digits, such as “1999,” they pared it down to the minimal “99,” just as people do in conversation.

But when the counter trips to “00,” like a car’s odometer, the computer has no way of knowing whether it’s 2000 or 1900.

*

The fix involved a combination of buying new equipment to replace the most badly outdated stock and a tedious process of reviewing tomes of programming code--about 17 millions lines, it was believed at the time--to find every command that added, subtracted or multiplied a date, and rewrite it to differentiate between the centuries. To get started, Nagata presented a modest request for $3.7 million to then-Supt. Sidney Thompson for inclusion in the 1996 budget.

“At first, like everyone else, he got sticker shock,” Nagata said.

Nagata made his case well enough to get Thompson’s preliminary approval, but in the later haggling over competing demands, the proposal was dropped.

Thwarted again in the 1997-98 budget year, Nagata said, he conceived an end run by taking his case to a blue-ribbon committee set up to monitor management reforms urged in a critical 1993 audit by the business consulting firm of Arthur Andersen.

The committee included two board members and three chief financial officers from the private sector, including William R. Isinger of the Los Angeles Times.

But, early this year, that committee was deeply enmeshed in another subject--a 1997 follow-up audit that was highly critical of Nagata’s Information Technology Division, concluding that political infighting has exacerbated the miscommunication and mistrust among the district’s senior management.

Evidence of the infighting was abundant in an exchange of icy memos this spring in which chief finance officer Henry Jones and budget director Marty Varon criticized Nagata for having no plan for the Year 2000 Problem while he held them accountable for not funding the work.

Nagata’s July 30 report to the committee touched off an uproar.

Minutes of the meeting show that Abeles, the Transamerica chief financial officer, judged the district 18 months behind most private industry and thought Nagata’s $9-million estimate three to four times too low, partly because it focused only on the district’s four main information systems, ignoring about 100 other noncritical databases.

Pressing for an outside evaluation, he recommended Andersen, one of only three firms with the expertise, because of its knowledge of the district.

In a blur of spending unusual even for a $5-billion institution, the board approved the $90,000 Andersen study, and, scarcely three weeks later, was confronted with an urgent recommendation from the audit committee to pay six times that much for a deeper analysis to break the job down into small projects that could be put out to bid.

Andersen had found that considerably more code needed review--the current figure is 24.4 million lines--but also said the district should spend more replacing equipment, rather than trying to update it.

Alarmed by the absence of competitive bidding for Andersen’s contract and the firm’s lofty fees, averaging $167 per hour for a team of consultants, board member David Tokofsky got a week’s postponement, but ultimately had no counter to the argument that every week lost increased the risk of an information catastrophe.

*

Under the rushed circumstances, Abeles said he considers the consultant’s fees a good deal when he sees other firms charging up to $200 an hour for similar work.

In trying to put the best light on the crisis, Nagata pointed to the district’s shortage of staff programmers, a mere 65 compared to hundreds for the county.

Still, despite the awkwardness of the political process, the district’s advisors don’t see all that dire an outcome.

“It would always be nice to have started a month earlier, or six months,” said Al Shepetuk, a partner in Andersen. “But there is enough time to get it done.”