P2P doesn't actually solve a problem for developing worlds either. If an end user has a small pipe they are just going to be spinning their wheels trying to participate in the cloud vs. actually downloading the bits.

Where P2P actually starts helping is if (1) you have a large pipe at the end point and (2) there is either a choke point between you and your download or at the download it self. If you don't have this *specific* situation P2P is at best no faster or slower than a normal download. At worst, it's dramatically slower.

Ok you are so far wrong in your assumptions, assertions and comments it's a bit saddening. Let me set the record straight on where your at right now.

Ok, your partially right here, but in reality P2P in all it's forms is *not* to avoid censorship at all, they are there to provide a mirroring infrastructure where non can or does exist. And BitTorrent is in fact a rather bad mirroring infrastructure at that. Don't try to apply political motives or any other traits to P2P, it's a technical solution to a technical problem (mind you most of the OSS world, and in most everything doesn't actually have this problem and there are a number of papers, reports and findings that are clearly showing the BitTorrent and their ilk do not even remotely compare to the capacity of an even small mirroring system)

Redlined? Cracking? Debian? I'm sorry to burst your bubble but by the last numbers I have, when Debian releases the internet does not strain under the load. When Fedora releases (currently the largest / most popular Linux distribution at release time) the internet does not load, and it's not uncommon for the Fedora mirroring system to move 100's if not thousands Terabytes in a week. The internet doesn't redline or strain under this load, in reality most people don't even know it's going on. This isn't 1998 anymore, and there are a lot of mirrors out there with 100mbps, 1000mbps and even a few now with 10gbps uplinks to the internet.

This is quite the worthy goal, and I applaud you for it.

Here's where your argument falls down though. If you use a traditional mirroring system, yes, your relying on the kindness and sponsorship of others to deal with the distribution of your project.

How is this any different than P2P? Your now relying on the sheer *kindness* (which in a P2P landscape is not to the pullers advantage) to continue to participate in the P2P cloud. While they are in they are sponsoring you and you are dependent on them. The only way to be truly independent is to run your own mirror on your own system and buy your own bandwidth.

P2P still doesn't really solve that problem. The problem your relating is one of end user's bandwidth vs. bandwidth of a mirror or P2P cloud. If an end user has a small pipe participating in P2P is actually rather detrimental to them as a lot of their traffic will be eaten up trying to participate vs. actually downloading. Slow pipes are also bad for the cloud in general.

Mirroring here is still a better way to go. Get a local mirror on site there (and yes there are mirrors everywhere) and let them get the content and then download from your local mirror.

What's interesting here is that the original 'GitTorrent' idea is dead, and after the core developers met for the GitTogether'08 it was made clear that the idea, really, is dead on the table. Where GitTorrent is currently headed is more towards a stateful mirroring and distribution system.