[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: y2k woes



PureBytes Links

Trading Reference Links

Michael Guess wrote:
>  When computers were first being developed, they were not powerful enough
>  to easily process dates with a four digit year.  Early programmers used a
>  two-digit year and saved millions of dollars.  The early programmers,
>  working in the '60s assumed that their programs would be obsolete by the
>  year 2000 and that later programmers would use four digit years.  The new
>  programmers learned from copying the old programmers, and nobody ever
>  started using four-digit years.

Alright, time for yet another Y2K flame.  It's tough to keep silent
with such garbage flying around.  If you believe the last sentence
above, you should believe the whole note.

Anecdotes are fun, here's another one:

On January 1, 1988, there was a leap second.  A leap second is the
atomic-clock-keeper's idea of a joke.  Some bright guy at Sun
Microsystems decided to implement a clever algorithm to handle the
leap second.  The code wasn't tested very extensively (if at all).
On Jan 1, most Sun computers went bonkers.  Time flowed forwards and
backwards at random intervals and with random velocities.  It was an
interesting experiment in breaking a fundamental software assumption,
i.e. time always flows forwards.  Not many people work on New Year's
Day, but within a few hours of the chaos hitting New York, someone at
Columbia University fixed the bug and posted a patch to the SunOS
kernel.  The problem was fixed around the world before Sun tech
support woke up, if I remember correctly.  My company was hit
particularly hard as we were selling a real-time forex forecasting
system--a 7x24 market.  EVERY engineer was in the office working on a
fix until the patch came in.  The patch was distributed over the
Internet which didn't fail even though most of the Internet computers
were Suns.

We will not be able to fix every embedded chip before 2000, but not
every embedded chip will fail catastrophically.  We will be able to
react to most critical problems within a matter of hours.  The largest
companies are working very diligently on this problem and have been
for years.  There are MANY companies whose sole purpose is to fix Y2K
problems.  It's a lucrative business.  Demand is high which has
increased the supply of programmers willing to work on COBOL programs.
People have come out of retirement just to "help out".  There are 153 
domain names containing y2k.  Inform yourself, type y2k into yahoo or 
metacrawler.

Some interesting facts: Jan 1, 2000 is a Saturday.  This gives us two
days before the real "go-live" Y2K for most businesses.  Just about
every technologist is assuming something is going to happen.  The
Internet protocols and telephone switches aren't written in COBOL.
(I am guessing, but I suspect the power grid isn't either.)  Computer
systems are not perfect.  Every year, there are many failures due to
daylight savings bugs.  Large portions of the phone system, the power
grid, and the Internet have failed catastrophically more than once.
In some countries, people are lucky if these systems work at all.
Catastrophic failures can actually have a positive economic effect,
e.g. nine months after major blackouts there are more births which is
generally a good thing for an economy.  Your car will continue to
work, but it's clock may break--then again most car clocks are broken
already.  Car systems don't rely on time-of-day clocks, because
computer clocks are notoriously inaccurate--honey, did you check the
time on the ABS today?  Y2K is boosting the economy because of the
increased spending--$5K/year on domain names alone!  Computer systems
have been failing around the globe for years due to time-synchronized
viruses.

Therefore, it is highly likely the economy and society will continue to
function as normal.  Although there probably will be a better than
average Christmas season in 1999 due to the hoarders.