Publishing's Amazon Powered Future

This is what I heard an Amazon executive say today at AWS Summit 2012:

"The boundaries are clear in EC2. Amazon is responsible for the dirt and the hypervisor; the customer is responsible for everything else".

If you want to understand the future of publishing, you really need to understand that statement, because very likely it reflects the corporate mindset of Amazon, and Amazon didn't get to be Amazon by having a schizophrenic mindset.

EC2 stands for "Elastic Compute Cloud" which is part of Amazon Web Services (AWS). EC2 allows companies to rent server space in Amazon's global data centers. Amazon has built a set of management services that allows anyone to scale their computing resources up or down with a few clicks of a button. The value proposition is staggering in both directions. Cycle Computing, which presented today described how they had assembled a 50,000 core supercomputer (representing $20 million worth of hardware) at a cost of $4829 per hour for a 3 hour drug discovery computation. The bottom end is equally staggering. A "micro instance" can be had for a quarter cent per hour. That's $23 per year.

The reason that Amazon does this is not to put IBM out of business. It's to make their website operations more efficient. Jeff Bezos saw that Amazon was putting lots of money into computing infrastructure, and he wanted to make sure that the operation was competitive with anything outside Amazon. By opening up the data center to outside users, he forced the operation to be more efficient.

Amazon also benefits from economies of scale. It can get better prices on components, cheaper bandwidth, lower electricity rates and better utilization of the data centers it's sprinkled around the world. Perhaps even more important is the creative energy from users that innovates around Amazon's technical infrastructure. Amazon was able to easily turn on video streaming to the Kindle Fire partly because Netflix was already using AWS to stream movies at a massive scale.

Dirt was a new term for me. It means physical infrastructure, what we used to call hardware and the data center. A hypervisor is software that enables virtualization- the ability to share physical computer hardware to power multiple virtual server systems. My company, Gluejar, owns no physical web servers. To run Unglue.it, we have 5 servers in EC2, for which we pay about a thousand dollars a month, all extras included. We could probably pay half that with a bit of work, and I spent an hour this afternoon listening to an Amazon software evangelist teach hundreds of techies how to spend LESS money with AWS.

What the dirt and the hypervisor share is that they benefit from economies of scale. If 10,000 companies share data centers, then not only is the data center cheaper for everyone, but together the companies can afford to have multiple, redundant data centers spread around the world. 10,000 companies sharing a pool of servers can fund development of scaling software, tools that make fault tolerance and high availability easy, load balancers, automated deployment systems, performance monitoring, security systems, and on and on.

Amazon is fundamentally a company about scale. The common thread between AWS and the internet book seller of 1995 is the identification of markets with large inefficiencies that could be eliminated by using the internet to amass scale. Amazon has alway been willing to lose money to achieve that scale. But this isn't predatory in the sense that having achieved market dominance, they raise prices. Instead, it's ruthless in that once scale is achieved, the resulting efficiencies can't be matched by anyone else.

It seems clear that Amazon has identified the publishing industry as a target ripe for further forcible efficiency improvements. But the nightmare narrative being spun by the publishing echo chamber is tragically unaware of how Amazon works. Maybe it's because publishers imagine that Amazon will do what they would do if they had Amazon's market power. But Amazon won't extort huge sums of money from powerless consumers. Instead, they will ruthlessly bring efficiency to every process involved in publishing. And then they'll invite everyone to use their ruthlessly efficient services.

Just like Amazon Web Services have enabled thousands of scaleable web startups and has made thousands of established companies more efficient, I predict that Amazon Publishing Services will enable thousands of new publishers and make thousands of established publishers more efficient. Amazon's Editing Turk will connect thousands of writers to pools of editors who will work harder at a lower cost. Amazon Creative Services will provide illustrations and graphic design. Amazon Elastic Curation will match groups of consumers with new authors writing the books they want to read. And Amazon Creative Capital will help visionaries invest in promising projects and writers. Amazon will spend to achieve scale wherever scale drives economies, and everywhere else, Amazon will provide hypervisors to match talent with tasks.

So the future of publishing will be dominated by publishers who take advantage of the various efficiencies of scale forged by Amazon (and Amazon competitors, we hope). The rest won't be worth talking about, except as an expression of nostalgia. The efficiencies will have many unfortunate effects, which professors will write books about. But they won't sell well. There are some things that even Amazon can't do.