There has been some recent press regarding Jeff Bezos’s latest directions. Some (relatively) positive posts, and some quite negative posts.

Jeff’s new services referenced in these articles are S3, EC2, and MTurk, Amazon offers many more services than just these three. A complete list is available.

A personal favorite of mine is Alexa Site Thumbnail. As a producer in my old job, I wrote a spec that required the ability to generate site thumbnails.  Dev killed the spec because it wasn’t feasible.  If Alexa Site Thumbnail had been around at the time, I could have saved my spec.

Amazon made a number of intelligent choices when designing these services:

For example, all the services are completely modular. In other words you can use as few or as many of Amazon’s services as you desire. Using S3 for storage doesn’t lock you into using EC2 for computation. It’s a great choice because Amazon won’t have any false successes. None of the services will buoy other lagging services. Also, modular features work well with a “release early, release often” quickly iterating dev cycle. Improvements to EC2 don’t depend on S3 and won’t effect S3’s availability (and vice versa).

Another intelligent design choice: the cost structure. S3 is flat, linearly scaling cost structure, which couldn’t be more vanilla. Alexa Site Thumbnail is a simple cost/image served. By simple I am referring to the intuitive notion of simply paying for what you use, no more, no less. I wish I could get a cellphone service like this (that wasn’t a pre-loaded phone, gimme a break).

Smart choice number three: they eat their own dog food. Amazon built these services because they were already using them and/or needed them. They are selling their own rack space, their own storage. If you (as a producer of a service) will not use your own service, why do you expect that anyone else will?

The distributed/grid computing future that Jeff is painting looks very realistic to me. I have talked to a number of entreprenuers who are using these services successfully, but don’t take my word for it. Check out the list of success stories. These are all real web services, most of them bootstrapping, which is only made easier by Amazon Web Services.

Jeff’s move is a long tail play. He knows that web services are going to get more specialized and focused going forward.  The number of web services that can be relatively successful through capital efficiency, cheap resources, and an abundance of advertising demand will continue to grow. It’s unlikely that they will become the next YouTube, but that doesn’t matter to Amazon because if Amazon can get a large number of small, niche services running on their infrastructure, Amazon will be a win regardless of the size of the individual services.

The decentralization of web services is not a novel concept. It’s a central theme in the Web 2.0 movement. So, why are so many analysts scratching their heads over Jeff’s new Amazon Web Services move? Is it just because Amazon’s current core business is retail? That’s a silly reason to doubt. Every company needs a labs project. Something to keep research moving forward and innovation alive. Jeff’s new strategy should be judged based on its own merits, not the origins from which it developed.

I eagerly anticipate seeing the wide array of new services that will leverage Amazon Web Services.  It will bring foward a new wave of mashups and cheap, lightweight services that can now ignore heavy-lifting problems like storage and CPU cycles.  They will focus on creative, new solutions, unhampered by infrastructure problems and growing pains.