In response to the question about the swarmed downloads, basically they will no always go into effect. They will only go into effect for files that are exactly the same size, so this almost always means that a given group will contain several swarm "buckets" of exact file matches. Then within this group, you will only swarm from hosts that successfully are connected to. So, in practice this means that often a group of 10 hosts will not generate a swarmed download, although sometimes it will.

We may also implement some changes that should improve this considerably (implementing persistent connections using HTTP 1.1, for example), but this is the way it is for the currrent release.

Let me know if you want any more details!