My goal with this blog

I write about relevant changes in the way that people use the web and how startups are built to provide services and products for this ever changing wonderful thing we still know as "the web." As a former entrepreneur turned early-stage investor, my greatest hope is for this to be useful to other folks that are like me in the hopes that they can avoid some of the mistakes I've made.

Open the cloud through open and portable runtimes

Jonathan Zittrain, of "The Future of the Internet and how to stop it" fame, had a great op-ed piece in the Times this Monday, hot on the heels of Amazon pulling a 1984 by remotely wiping purchased copies of Orwell's books from users's Kindles. In it he cautions us the get our heads out of the clouds and stop thinking about cloud computing as necessarily an improvement on what we have today. He finishes the piece with a warning that echoes the main theme of his thoughtful book:

The market is churning through these issues. Amazon is offering a generic cloud-computing infrastructure so anyone can set up new software on a new Web site without gatekeeping by the likes of Facebook. Google’s Android platform is being used in a new generation of mobile phones with fewer restrictions on outside code. But the dynamics here are complicated. When we vest our activities and identities in one place in the cloud, it takes a lot of dissatisfaction for us to move. And many software developers who once would have been writing whatever they wanted for PCs are simply developing less adventurous, less subversive, less game-changing code under the watchful eyes of Facebook and Apple.

While the Platform as Service providers like Facebook, Apple, Google, and soon to be Microsoft, will continue to push their integrated siloed stacks for hosting code and data in their proprietary runtimes, I believe most developers are smart enough to realize that they are giving up a tremendous amount of independence for the benefits of virally leveraging the social graph, tighter integration with other Google services, or the AppStore effect. As many Facebook application developers have already learned, building in these silos is at best a Faustian bargain which will someday come due.

To me the only way cloud computing works in the long term is with full portability of both data and execution context. And ironically, despite the bonehead Kindle move, Amazon's use of generic x86 instances as the unit of computation for hire is a great first step in this direction. In theory anyone with enough Xen skills could compete by the CPU hour with EC2— an quite a few folks already claim that they do!

Assuming we continue along this model in the development for the open (or as Zittrain calls it "generative") cloud, we need to then move to moving this computation around from virtualized servers to any device we want, be it the old PC connected to your cable modem, your dual core laptop, or even some future version of a mobile computing device.

In a sense this was the promise of the JVM— write the code for one virtual machine and run it on anything from a Java ring to an enterprise class server. Unfortunately the JVM ecosystem did not bring with it three decades of rich UNIX tools, millions of man years of expertise in networked OSes, and a thriving open source ecosystem upon which to build. Perhaps as projects like LLVM get tailwinds due to the aggressive move to multi-core, we'll finally get a VM-like experience that allows us to bring the whole turtle's shell of stuff that Linux and OS X make suck brilliant use of with us— a la ubiquitous EC2 instances.

I'm not sure how this is all going to turn out in the end but what continues to amaze me is that I don't think we've ever been at a point in computing where there are just so many execution context choices facing people looking to bring new functionality into the world. This week on the 40th anniversary of the moon missions, Google publicized the Virtual AGC project which contains all of the Apollo lander code and an emulator as open source. This was the state of the art in 1969— running on some proprietary computer that I'm sure cost millions in taxpayer dollars. And yet from my little Arduino board to the EC2 instance I rent from Amazon, there is not a single computational execution context in my life that couldn't run this thousands of time faster than it did when those guys bounced around on the moon thanks to it.

Amazing.