> 16.5% of total Internet traffic on an average day comes from BitTorrent. Since BitTorrent traffic goes both ways (upload and download), 8.75% of this is downstream traffic.
Uh, what? No. In a closed network, 100% of all traffic is downstream somewhere. It's like saying that 50% of all phone calls in the world are phone calls to someone, and the other 50% are phone calls from someone.
That depends on how they count traffic. They may double count all traffic (i.e. count traffic on local links, so the upload and download are counted separately).
Or they could count only transit traffic, in which case it would only be counted once - but they would have to make sure not to miss traffic from close neighbors.
Personally I would count local traffic (i.e. to each single computer) and record upload and download separately - including servers of course.
If I send you a file then yes, obviously that is one file being sent, not half a file. But the way bandwidth is measured, it will count my bandwidth uploading it and yours downloading it.
Moreover, bittorrent connections are almost never as simple as Alice sending a file to Bob. The network topology is far more intricate, with Bob and Alice sending and receiving chunks simultaneously from dozens of seeders/leechers.
For some reason I was thinking the quantity downloaded could exceed the quantity uploaded in a torrent swarm, but that's just wrong. The inputs equal the outputs.
I now see that { A->B, A->C, A->D } is three separate one-to-one mappings, instead of a single one-to-many mapping, in the context of A sharing a chunk with B, C, and D.
Uh, what? No. In a closed network, 100% of all traffic is downstream somewhere. It's like saying that 50% of all phone calls in the world are phone calls to someone, and the other 50% are phone calls from someone.