Kryder's Law




A quick look over my site will reveal that it's about tracking price and capacity trends for computer storage. The figures are updated every 12 months and then analyzed to check for any variations or interesting new trends. It's a site that only a real computer nerd could enjoy.

One chart produced from the data is the annual improvement in bang per buck graph, shown here:


An analysis of the long term trend shows a doubling of the hard disk storage purchasable for a dollar each year for some 15 years. More disturbing for those of us that count on this fabulous trend to solve our storage woes is that for the last 3 years the trend has fallen away dramatically. In the last 3 years its only improved, on average, 35% a year. Thats way short of the 100% per annum long term average and if this new trend continues, well, er, its a new trend. People ought to know about it. And I should be the one to tell them. I'll write an article.

Even though I have collected my own data for the long term price trend for hard disk storage, in researching for this piece, I wanted to know what experts had to say on the matter before I started making bold claims about the state of the hard disk industry.

A google of "Moores law for hard disks" quickly brings up the Wikipedia entry on Moore's law as the "I'm feeling Lucky" entry. A good place to start. I quickly learned that "An even stronger law has held for hard disk storage cost per unit of information" than Moore's Law. "The current rate of increase in hard drive capacity is much faster than the rate of increase in transistor count and has been dubbed Kryder's Law."

The Wikipedia entry on Kryder's law says that "the density of information on hard drives has been growing at an even faster rate, increasing by a factor of 1000 in 10.5 years, which corresponds to a doubling roughly every 13 months".

This was all great news for me. The rate of increase in bit density almost exactly matched my data on the improvement of $ per Meg, implying there is strong correlation between the two. If I could show that bit densities are not going to improve anywhere near the required ~90% annually for the next few years then I would be able to say, with a great deal of certainty, that Kryders law is broken.

Well, I can. In a recent interview at Hardwarezone.com, James M. Chirico, the Senior VP of Global Operations at Seagate , said that the annual bit density improvement for the next 3 years is expected to be a mere 40% per annum.

Kryder's Law is broken.

Or it would be if it ever existed in the first place. Here's the thing. Wikipedia cites the source for the "factor of 1000 in 10.5 years" statement as this Scientific American article. The statement is almost certainly taken from this sentence: "Inside of a decade and a half, hard disks had increased their capacity 1,000-fold, a rate that Intel founder Gordon Moore himself has called "flabbergasting". A decade and a half isn't 10.5 years, it's 15 years.

Some simple calculations show that there are 10 separate 18 month periods in 15 years and any computer nerd should be able to tell you that 2 to the power of 10 is 1024. This means the rate of improvement in areal density for hard disks is a doubling every 18 months.

This means Kryder's Law isn't any stronger than Moore's Law. In fact, Kryder's is exactly the same rate as Moore's Law. Its a wonder there is a separate name for the hard disk observation at all. A little checking shows the concept called Kryder's Law probably wouldn't even exist if it weren't for Wikipedia. The only time "Kryder's Law" is mentioned anywhere on the googleable Internet apart from works derived from the Wikipedia article is in the headline of the Scientific American article, which was senationalised, like most headlines, to attract interest. It seems an editor has read the article and used it as inspiration to create the wikipedia entry. A second editor has then misquoted the article when including the 10.5 year figure. Wikipedia's strong reputation did the rest.

Moore's Law is usually expressed as a doubling every 18 months which equates to a 54% per annum increase. Drives have failed to meet the 54% level for the last 3 years and they are also not expected to meet it for the next 3 years. If you have been counting on the fantastic trend to solve your storage problems then you may have to adjust your expectations.




Home