Achieving Connectivity From the Edge
This is the talk I gave at Ecomm2008 on March 13th. The conference was run on a tight schedule so I compressed a complex topic into 15 minutes.
Most of the sessions at the conference presented new products and services. From my perspective we had two conferences intertwined. In one we saw the people taking advantage of the possibilities created by the Internet -- in order to create a solution you need only concentrate on the users at the end points of a relationship. The other conference was about telecommunications and all the problems associated with working around the impediments between these two end points or, from the carriers' perspective, maximizing the money they could make by owning the path.
These approaches are in stark opposition yet we treat them as a single entity. We do this because we are so inured to the hardship that we don't realize it is unnecessary. We associate pain with gain and confuse the effort spent working around impediments with progress. As I explain in my talk we don't need a telecom industry or a centralized Internet because we can achieve connectivity from the edge. If we own our local facilities --the copper, glass and radios -- we can avail ourselves of the inherent abundance we associate with the Internet.
The Post Office understands routing -- it delivers to each address using fixed routes. You take responsibility for finding your friend's address. This is far simpler than using the IP address which confuses the names and addresses thus creating a monumental housekeeping task. As Skype demonstrates we can decouple the end point names (or identifiers) from the path (or address).
We can compare networking to driving. If I want to get from Boston to Seattle by driving I just hop on I-90. In practice I may take shortcuts when driving. But a packet has no problem scooting across the country taking advantage of high speed, high capacity fiber.
Today routing is complicated by worrying about whose "highway" we are using because we have to pay for transport and thus can't take advantage of the inherent low cost of fiber. Today one percent of one of the "broadband" pipes that pass my house is enough for all my TV and Internet services and I have three available. That means I'm paying 300x the actual cost! If this were shared infrastructure -- what I call the bit commons -- we'd be paying less than 1% of today's cost. If we use the shared infrastructure for many other purposes purpose we have a savings rather than a cost.
This means that today's telecom industry isn't a sustainable business and exists only because of a regulatory system that is a reflection of our collective suspension of disbelief.
In the early 1980's ATT saw that their business model had to change (though they had to be prodded) and divested itself of their local Bells. Today we have an opportunity to revisit divestiture. But this time the Internet as a proven model. We must learn the lesson of the first divesture and not create new local operating companies that are in the style of telecom providers. If the city defines the services then we still have a carrier -- a muni-bell.
Once we can assume abundant infrastructure and ubiquitous coverage (with and without wires) we are ready to take advantage of the Internet and focus on the relationships between two end points and not get lost in the muddle in the middle.
You can also download the slides to see the animations and effects.