Kryder's Law
"Disk storage density grows exponentially, once even faster than transistor counts."
"Disk storage density grows exponentially, once even faster than transistor counts."
The areal density of magnetic hard drives, and so the cost per bit of disk storage, improved exponentially for decades. At its naming the trend was running even faster than Moore's law for transistors, which is what made cheap mass storage possible.
Rated Empirical: grounded in studies or measurable evidence. The other 608.
Successive magnetic-recording innovations drove the density gains: giant-magnetoresistance read heads, then perpendicular recording replacing longitudinal, and later heat-assisted magnetic recording. Each let engineers shrink the magnetic grain and bit-cell size while keeping the data readable and thermally stable, packing more bits onto every square inch of platter.
The semiconductor analogue Kryder's law was explicitly measured against, transistor counts doubling roughly every two years, and the trend that Kryder's disk-density gains briefly outran.
A sibling in the family of exponential-technology trends: the energy efficiency of computing, computations per joule, doubling over time (another 'law' that is really a fitted empirical curve).
It named and popularised the exponential fall in the cost of disk storage that underwrote the data-abundant era: cheap backups, media libraries, and cloud storage all rested on it. Its later slowdown is equally instructive, a clean case of an exponential 'law' that was really a trend and duly bent.
Kryder's law slowed markedly after about 2010. Actual annual areal-density growth fell toward roughly fifteen percent by 2014, against the forty percent that projections assumed, and Kryder's 2009 forecast of a cheap 40-terabyte 2.5-inch drive by 2020 did not materialise. The causes were physical and practical: the approach to the superparamagnetic limit, where grains become thermally unstable, and the repeatedly delayed, costly transition to heat-assisted recording, alongside competition from flash. One analysis estimates disk ended up about seven times more expensive than the pre-2010 trend implied. It is an empirical, era-bound trend, not a physical law, and the curve has bent.
It is not a law of nature but a fitted historical trend that has demonstrably decelerated, so it should not be extrapolated forward. The often-quoted '13-month doubling' is a commonly cited figure rather than the 2005 article's own headline phrasing, which emphasised a thousandfold rise over fifteen years and 'faster than Moore's law'.
Yes, as far as the evidence goes. Kryder's Law is rated Empirical here: it rests on studies or measurements rather than on a saying.
Rated from three sources. 609 of the 1,116 entries carry this rating. How the scale works.
The best-known laws are the least likely to rest on measurement.
Named for Mark Kryder, who directed Carnegie Mellon's Data Storage Systems Center and later served as Seagate's chief technology officer. The term was coined in a Scientific American article, 'Kryder's Law' by Chip Walter, published in August 2005; the trend slowed sharply after about 2010.
Everything above traces to scientificamerican.com, en.wikipedia.org and theregister.com, three sources in all, one of them the original publication rather than an account of it. Nothing is written from memory, and the method is written down. Spot an error or a better source? Suggest a fix.