While pursuing ideas related to the Connected Age, I’ve read some stuff on collaborative consumption and the sharing economy (think Zipcar, AirBNB, Freecycle).
The Industrial Age was defined, in large part, by ownership. In Connected Age, we’re seeing a move to access.
This lead me to think about “the cloud”, where our data is stored, and how Amazon’s EC2 exemplifies this shift to access. You don’t need to have racks and racks of your own computers — just rent access to their service.
But then I thought about their major outage a few weeks back, and how the sharing economy could actually do EC2 one better. The problem with EC2 is that there’s still a singular point of engagement, and if it goes down, you’re hosed.
So, here’s the billion dollar idea I have no idea how to make happen. Provide Amazon EC2-iike services, but with a SETI@home-like distributed computing model. There has to be lots of spare cycles and storage, accessible through high-bandwidth connections. And with a distributed model, you could avoid the single point of failure that seemed to bring down EC2.
I’m sure this would be an enormous technical challenge, but the upside seems enormous.
That’s already kinda happening (see: http://gigaom.com/2007/10/30/grid-networks/). I still am unsure of how the implementation works, but it’s definitely an interesting spin.
What you’ve described sounds like BitTorrent (or possibly Freenet with a monetization layer on top of it.
Peter,
One way to think about achieving this would be to create an abstraction layer over the varying clouds (EC2, Google, IBM, Azure, etc.) One of the problems with today’s cloud platforms is that as each of them are fiercely competing for your rental of service dollars, each of them are making their cloud “bases” non-interoperable. Six to twelve months of programming could overcome most of that non interoperability, and then there are just workflow hooks involved as to when to switch from one cloud provider to another. Another defense against AWS downtime would be to build one’s own app with a multi cloud use design from the beginning (this would be the individualist approach to building an abstraction layer over the various vendors’ cloud services).
Just like the trend toward selling excess power back to the grid, we need a measurable (and secure) pipeline to and from each computing unit.
I would like to get paid for my idle computing power. It would need to pay more than the electricity cost of leaving the computer running all the time.