Archived Edition
Although the Y2K problem came and went in January of 2000, we have saved this article as an archived editon of HowStuffWorks because of its historical value. Published at the beginning of 1999 at the height of the Y2K panic in the media, this article is noteworthy for the sentence, "In reality, nothing will happen." In retrospect, that sentence was completely correct, but in January 1999 that was definitely not the picture the mainstream media was painting. HowStuffWorks received quite a bit of flaming email for making this simple prediction.
|
You will be hearing about the "Year 2000" problem constantly in the news this year. And
you will hear a lot of conflicting information in the process. There is also a good
bit of "end of the world" rhetoric floating around on the Internet. What should you believe?
In this edition of How Stuff Works we will discuss
the Year 2000 problem (also known as the Y2K problem) so that you understand exactly what is happening and
what is being done about it. You can also explore a variety of
links. From this information you draw your own informed conclusions.
What Is the Y2K Problem?
The cause of the Y2K problem is pretty simple. Until recently, computer programmers
have been in the habit of using two digit placeholders for
the year portion of the date in their software. For example, the expiration date
for a typical insurance policy or credit card is stored in a computer file in MM/DD/YY format (e.g. - 08/31/99).
Programmers have done this for a variety of reasons, including:
- That's how everyone does it in their normal lives. When you write a check by
hand and you use the "slash" format for the date, you write it like that.
- It takes less space to store 2 digits instead of 4 (not a big deal now because
hard disks are so cheap, but it was once a big deal on older machines).
- Standards agencies did not recommend a 4-digit date format until recently.
- No one expected a lot of this software to have such a long lifetime. People
writing software in 1970 had no reason to believe the software would still be
in use 30 years later.
The 2-digit year format creates a problem for most
programs when "00" is entered for the year. The software does not
know whether to interpret "00" as "1900" or "2000". Most programs therefore default to 1900.
That is, the code that most programmer's wrote either prepends "19"
to the front of the two-digit date, or it makes no assumption about the century and
therefore, by default, it is "19". This wouldn't be a problem except that programs
perform lots of calculations on dates. For example, to calculate how old you are
a program will take today's date and subtract your birthdate from it. That subtraction
works fine on two-digit year dates until today's date and your birthdate are in
different centuries. Then the calculation no longer works. For example, if
the program thinks that today's date is 1/1/00 and your birthday is 1/1/65,
then it may calculate that you are -65 years old rather than 35 years old.
As a result, date calculations give erroneous output and
software crashes or produces the wrong results.
The important thing to recognize is that that's it. That is the whole Year 2000
problem. Many programmers used a 2-digit format for the year in their programs,
and as a result their date calculations won't produce the right answers on 1/1/2000.
There is nothing more to it than that.
The solution, obviously, is to fix the programs so that they work properly. There
are a couple of standard solutions:
- Recode the software so that it understands that years
like 00, 01, 02, etc. really mean 2000, 2001, 2002, etc.
- "Truly fix the problem" by using 4-digit placeholders for years and recoding
all the software to deal with 4-digit dates. [Interesting thought question - why use
4 digits for the year? Why not use 5, or even 6? Because most people assume that no
one will be using this software 8,000 years from now, and that seems like a
reasonable assumption. Now you can see how we got ourselves into the Y2K problem...]
Either of these fixes is easy to do at the conceptual level - you go into the code,
find every date calculation and change them to handle things properly. It's just that
there are millions of places in software that have to be fixed, and each fix
has to be done by hand and then tested. For example, an insurance company might
have 20 or 30 million lines of code that performs its insurance calculations.
Inside the code there might be 100,000 or 200,000 date calculations. Depending on how the code
was written, it may be that programmers have to go in by hand and modify each point
in the program that uses a date. Then they have to test each change.
The testing is the hard part in most cases - it can take a lot of time.
If you figure
it takes one day to make and test each change, and there's 100,000 changes to make, and a person
works 200 days a year, then that means it will take 500 people a year to make all the changes.
If you also figure that most companies don't have 500 idle programmers sitting around
for a year to do it and they have to go hire those people, you can see why this can become
a pretty expensive problem. If you figure that a programmer costs something like
$150,000 per year (once you include everything like the programmer's salary, benefits,
office space, equipment, management, training, etc.), you can see that it can cost a company tens of millions
of dollars to fix all of the date calculations in a large program.