Peer to peer is starting to become accepted as a way for content owners to keep their distribution costs down, rather than just a route for distributing content without the agreement of rights owners. Start-ups like Joost and established U.K. broadcasters such as Channel 4 and the BBC all use peer to peer under the hood of their on-demand video services. It means that they don't have to serve each viewer their own copy of the content. That means they save money on servers and bandwidth.

The problem: peer to peer does not save much money now, and the situation may not improve much in the future.

Few figures have been published, but I understand that today's mass market peer to peer systems still serve about 90% of traffic from their own servers. The much vaunted P2P technology only delivers a 10% cost saving. Video platforms will hope new peers on the network will take the strain, as their systems become more popular. I'm not convinced.

The 'A' in ADSL stands for Asymmetric. Your connection is much slower when uploading content to someone else than when downloading content for yourself. The technical standards dictate that uploading is at most half the speed of downloading. BT's standard DSL now provides a download speed that is up to 8 times faster than the upload speed. There's a further asymmetry in customer's attitudes to their connection. If I'm downloading a TV show to watch I don't mind if the software maxes out my connection. I'm less flexible when serving content to strangers; browsing becomes unpleasant if the upload speed is much more than half the capacity of my connection.

A simple way to look at all this: if I use software to download TV for an hour, I have to be willing to keep the software running for another 16 hours before I have given back as much to the network as I have taken. If I shut the application before the 16th hour the broadcaster pays. Today this is expensive for broadcasters. If telcos have their way, the costs will keep going up.

A possible solution: wireless mesh networks

In addition to the normal home setup of broadband connection, wireless network etc, people who like on demand TV could install an extra wifi adaptor for media distribution. Today it might be a simple USB card plugged into the back of the PC, costing around $20. Tomorrow it might be an invisible part of the set-top box. The wifi adaptor joins a special wireless network with other homes in the neighbourhood. This new network would be used mainly for peer to peer traffic, and would be several times faster than DSL.

The great news (with thanks here to Gareth for a great chat yesterday) is that the technology to make this possible is already under development. It's called mesh, and is being considered for providing networks in countries with poor telecoms infrastructure. Cringely was recently talking about Google using the technology in the U.S. to start a new kind of ISP. Some people had a crack at doing just that in London. Maybe mesh has a small role in developed countries, not as the primary network, but as an optional second network for those who want faster or higher quality media? And maybe it does not take Google to kick-start it?

So, what would it take to get something like this started? Or is the 'chicken and egg' situation impossible to overcome?

UPDATE: Gareth has interesting thoughts on this. He points out that routers would be better than USB network cards