Ruminations

2020 Hindsight

Twenty years ago tonight - if you're reading this on the date it was posted - it was a different world. Prince was partying like, you know. People were debating whether the new millennium started on January 1st, 2000 or January 1st, 2001. Some were wondering why we didn't have the flying cars yet that we were promised at the 1964/65 World's Fair. And those of us in IT were crossing our fingers that we had done enough to avert a potential technological catastrophe known as Y2K.

The Y2K bug wasn't really a bug at all. It was a system design decision, initially based on cost and then on system compatibility. In the end, resolving the Y2K issue cost hundreds of billions of dollars. Looking back, that was more in adjusted dollars than it would have cost to prevent it from happening in the first place.

For those wondering what the Y2K problem was, here is the gist of it: dates were stored in computers using only two digits for the year. When 1999 changed to 2000, the two-digit years in computer dates would rollover from '99' to '00'. This would wreak havoc on anything that used dates for calculations (e.g., billing and due dates), and anything that involved birthdays. I'm sure you can imagine many more. There were also systems that calculated positions, distances or trajectories based on where things are expected to be at a given date and time, and the Earth's rotation.

So why were dates stored only using two-digit years? Cost. The cost of computer storage and computer memory were so high that every effort was made to keep costs down. So not storing two extra digits per date - the amount of space an emoji takes up in a text - saved a lot of money.

Back in the 1960's computer storage (cards, then tape, then disk) was expensive. For example, the cost of storing the two extra digits in a date ranged from $0.01 to $0.10 for each date. That doesn't sound like a lot; but imagine if you had to store thousands of dates for millions of customers (e.g., a call record for each and every call for each customer of the phone company).

If your home computer's 500GB hard drive cost that much, it would have cost between $5.3 billion and $53 billion (in 1960's dollars). Fortunately, your cell phone's 64GB storage would only be $67 million to $6.7 billion.

RAM was even more expensive because it was hard to manufacture, resulting in higher demand and prices (Econ 101). Early core memory could cost as much as $1 for each bit. That may not sound like a lot, but the cost of using just one emoji would be $16 of RAM. For comparison, your home computer's 4GB of RAM would cost over $34 billion (in 1960's dollars). Your cell phone's RAM would cost $17 billion to $25 billion, or more if you have a high-end phone.

Fortunately for accountants and budget directors, computers were available with as little as 2 kilobytes of memory ($16,384 just for 2k of RAM, in 1960's dollars). That kept the costs a little more palatable. 2k of RAM doesn't sound like much, and it isn't, but programmers were able to work magic back then (with some assistance from swapping algorithms).

You may be wondering how so many companies were taken by surprise by Y2K. Not all of them were, but most companies that had computers were very dependent on their computer vendors. So, most Y2K solutions were developed and distributed by the computer companies themselves.

The first person credited with raising the Y2K issue was Bob Bemer, back in 1958. He spent 20 years trying to get the major players - including the US government - to pay attention to the looming problem. He had very little success. For perspective, Bemer spent the same number of years trying to alert the computer industry as the number of years that have passed since Y2K arrived.

I'd also like to give an honorable mention to Erik Naggum. In 1989, Naggum's efforts to ensure that a blossoming technology called "email" used four-digit years prevented email from falling victim to what the rest of the computing world would be battling a decade later.

Some industries were more aware of the Y2K problem than others. For example, banks and insurance companies had a head start when they encountered end dates of mortgages and life insurance policies that were beyond the year 2000.

Of course, Y2K wasn't the first time a two-digit year rolling over from '99' to '00' was a problem. Norway and Finland encountered a similar issue in their government records going from 1899 to 1900. This was known as the Year 1900 problem.

I hate to ruin it for anyone who hasn't finished their research yet, but Prince was wrong. Contrary to many predictions the world did not end on January 1st, 2000. In the race up to it a niche industry emerged to help companies identify and prevent the approaching problems. All in all, it was a lot of work and very expensive.

Some businesses, and even countries, disagreed that all the time and money put into trying to prevent the Y2K issue was worth the trouble. Rather than trying to prevent it they took a "fix on failure" approach. But proponents point out that that software vendors made and distributed the Y2K fixes to all customers and countries, including those making little or no effort to prevent a Y2K problem on their own.

There were different approaches taken to avert the Y2K problem, and they varied in scope, effort and longevity. Fortunately, there were very few major problems. The most significant Y2K related issues occurred in Germany and Belgium. In Germany, around 20 million bank cards failed. In Belgium, Digipass customer identification chips stopped working.

Other less impactful issues cropped up, such as:

• Many HSBC credit card machines didn't work for the three days preceding January 1st, 2000.
• Some EFTPOS terminals failed.
• PlayStation 3 models, except fo the Slim model, had problems.

In the months after Y2K some problems surfaced, too. In the UK, some Quickfare railway self-service ticket machines issued invalid tickets. In Japan, weather data was corrupted and about 5% of post office cash machines failed. In Bulgaria, police documents were issued with invalid dates. In the US, the Coast Guard and an Air Force base encountered issues.

How times have changed. If you have a cell phone released within the last few years your phone probably has more storage space and more RAM than all of the computers in existence in the 1960's combined.

Y2K isn't the only date problem we've faced. January 4th, 1975 brought us the 12-bit date overflow on Decsystem 10 operating systems. And there was an issue in some computers on September 9th, 1999 due to system design shortsightedness. And who can forget the Year 2010 problem, where binary-coded decimals were used for data storage rather than hexadecimal numbers. This affected SMS message dates on some cellphones and Windows Mobile.

The future isn't free of problems either. We're facing a 32-bit date overflow issue on January 19th, 2038. This will affect hundreds of millions of embedded chips in everything from appliances, to entertainment systems, to automobiles, to airplanes and military equipment (including missiles). These days almost everything has a chip in it, and almost all of them are 32-bit.

So happy New Year! Tonight, I'll be celebrating that we're still 18 years away from 2038, and that I can still remember 2000 (even if I can't party like it's, you know).

---