Now you can see who is leeching in 3D!

The most obvious one is that it seemed to lack the endgame mode, in which bittorrent tries to get the last pieces from more than one person, so you don't have to wait for the slowest one. A look in the release notes shows that he fixed that in mid february. The way it connected to other peers also lost against the original bittorrent in the 2 tests I made. Moreover claiming "Bits on Wheels is the first 100% Native BitTorrent client" shows that the implementor has no idea how complicated and hard good networking and p2p is. If you listen to the talk mentioned above, then you'll see how much caveats and problems there are. And bram is doing this like forever. There are many subtleties in his algorithm, that make them work the way it does. A good time investment would be to improve the bittorrent in the areas already identified. BitsOnWheels may get level to level with azureus some time. But even azureus does not achieve the throughput of the original python client. Which btw just got updated to 4.0.1 with again serious improvements. I'd love to see people that put this litte python core into a nice really good GUI client, instead of reinventing the wheel and sucking at it.

Terrific! Really great looking client that gives you much more info about your download than the standard OS X BitTorrent client. Love the 3D swarm graph!

Bits on Wheels works just fine. What exactly makes it " the worst bittorrent implemenation ever?"

This client may look good, but it's the worst bittorrent implemenation ever. Don't use it. Bram has put a lot of effort into making his original python implementation the best performing p2p software. Just reimplementing from scratch based on the protocol description won't do. Watch http://www.boingboing.net/2005/03/06/bram_cohens_stanford.html to get some insight about it.