ATOM Award of the Month, April 2019

For this month's ATOM AotM, we return to the familiar, but in the process, we want to recognize an ATOM trend that has not gotten as much credit as it has deserved.  

We all know what Moore's Law is, and what has been enabled by it.  But what has always been amazing to me is how little recognition a similar law has received.  Storage capacity has risen at a rate equal to (or slightly higher than) Moore's Law.  It is not a technological byproduct of Moore's Law, as it has always been worked on by different people in different companies with different technical talents.  

If storage capacity were not improving at the same rate as Moore's Law, most computer-type products would not have continued to produce decades of viable new iterations.  From PCs to Smartphones to Video Game consoles, all have a storage requirement that has to match up to the size and number of files downloaded and processed.  Correspondingly, data transfer speeds have also had to rise (USB 1.0 to 2.0 to 3.0, Ethernet to Fast Ethernet to Gigabit Ethernet, etc.).  A 2019-era PC could not have a 2006-era hard drive and be very useful.  

Like Moore's Law, the exponential doubling has spanned a sequence of technologies that all sustain the underlying megatrend, from internal spinning hard drives, to flash storage, to storage in the cloud.  Greater density has been matched by shrinking weight per unit storage and less power consumption.  This also means that has computing decouples from Moore's Law and moves into different (and probably faster) forms of exponential growth, storage will almost certainly also follow suit.  DNA-based storage is a prospective technology that has many attributes comparable to expected future computing technologies such as Quantum Computing.  

Unlike Moore's Law, storage has not always advanced at a steady rate.  There are times when it advanced much faster than Moore's Law, and times when it advanced much slower (such as in recent years).  The 40-year average, however, does appear to match Moore's Law's doubling rate rather closely, and hence what one dollar purchases today is the same as what one billion dollars could purchase then, which itself would have been the size of a house.  

Also unlike Moore's Law, there is not a universally-accepted name associated with this trend.  Mark Kryder is sometimes given this attribute, but he did not put forth a prediction early enough for it to be a prediction by any measure (Kryder officially spoke of this in 2005 whereas Moore made his prediction in 1965), his name is not mentioned in any of Ray Kurzweil's writings or other publications, and since he was not the founder of a major storage company, he is not analogous to Gordon Moore.  

As rising storage efficiency is crucial towards the productization of any other form of computing product (including Smartphones), it deserves recognition for its contribution to the technological age, despite often being overlooked in favor of Moore's Law.  

 

Related ATOM Chapters :

3.  Technological Disruption is Pervasive and Deepening

 

Comments

[Geoman]

Good choice. Not to mention the world wide web would not exist without substantial memory improvements.

It is interesting to see that we are now a world where nothing is forgotten. It is not all to the good, especially as I see people damned for minor sins committed decades ago. No wonder teens today seem so neurotic.

[Kartik Gada]

Many capabilities of great value happened without any fanfare.

For example, anyone can send an email with a 10 MB attachment to, say, 10 recipients. The amount of storage needed suddenly rose 10x, but there is so much now that this capability is not a network and hard drive saturating practice.

But no one remembers when attachments become possible for emails. What was the exact day? Or even the exact year?

Similarly, if a .pdf is posted on a website and available for download (such as my AI report), if 1000 people download it, that is a huge amount of aggregate storage occupied in a one-to-many system that only became commoditized very recently.

[Devang]

Does the emergence of the cloud mean effectively unlimited storage? Or at the very least, no wastage of storage?

Many people can't even fill 10% of their desktop PC hard drives, while others buy expensive extra storage. This supply-demand mismatch, while not a huge deal, always seemed a bit wasteful.

[HB]

There are at least a few potential significant disruptors in the horizon regarding storage. One is three dimensional storage using current materials or similar. Another might be DNA. A few pounds of DNA can store more information than all currently available worldwide storage. The caveat when it comes to more specific predictions is, as always, that a true disruptor often does not appear in the horizon. It appears suddenly in ways virtually nobody expected.

[Fatcat]

A bit topic but this article maintained that the economic doubling will not be
accelerating. In the other hand it does similar analysis to the tends. https://slatestarcodex.com/2019/04/22/1960-the-year-the-singularity-was-cancelled/

[Kartik Gada]

Fatcat,

Thanks for that. I will read that in coming days.

But in general, issues like the problems between Nominal and Real GDP, as well as the greater accuracy of using asset value growth instead of GDP, appear to be relevant in this article as well.

[Kartik Gada]

Fatcat,

Ok, I have read it.

The problems with his conclusion are many of the items discussed in Chapters 2 and 4 of the ATOM publication.

i) Not seeing that too-low inflation crimps 'real' GDP, and that Nominal GDP would be more revealing.
ii) Not using per-capita numbers (and hence attributing economic growth to population growth).
iii) Not seeing that GDP captures technological change inaccurately ('more with more' rather than 'the same with less').
iv) Not looking at asset value growth in addition. Asset value growth, while not perfect, is a better reflection of economic progress than GDP (even Nominal GDP). Assets are, of course, even more concentrated at the top than income, but that, too, would bring more accuracy to that discussion.

[Kartik Gada]

HB,

Yes, DNA storage has high potential. Interestingly, 15 years ago, DNA was cited as a medium for computing, rather than storage. Perhaps, it might eventually serve as both.