Slashdot Mirror


The Centralization of BitTorrent Networks

Writes writes writes writes "A group of graduate students from the University of Washington have posted a a new independent report about the extent of centralization in regards to BitTorrent communities. The report indicates that irrespective of the recent damage dealt to global torrent sites, the communities are still very active, even despite their large degree of centralization (and perhaps exposure/liability). Furthermore, the report attempts to determine if the torrent communities follow the 80/20 rule, by measuring the Long Tail of torrent distribution."

11 of 162 comments (clear)

  1. BT links on FreeNET by nurb432 · · Score: 4, Interesting

    This would help reduce the liability of centralized 'link sites'.

    Sure, you can still be tracked once you hit a tracker, but at least the source of the links is safe.

    And before you shout ' freenet is slow', getting a simple BT link from FN would work well. Thats what FN is designed for.. Small bits of data..

    --
    ---- Booth was a patriot ----
  2. Re:Umm, no, it won't ever die. by araemo · · Score: 2, Interesting

    Yeah, I wanted the latest knoppix iso, so I of course used the torrent. Within 5 seconds I had hit 250KB/sec, not exactly my physical maximum, but pretty good for my connection(From very rare sites, I can get 400-500KB/sec, but 90% of internet sites cap out at 250 for me.)

    When it's something legal(And hence there is no reason to fear keeping your seed running) Bittorrent is still pretty fast.

    Especially since a lot of traditional mirror sites for *nix distros have added torrent seeds to their ftp mirrors. This seems to me like a good way of sharing the bandwidth load while also keeping download speeds fast.

  3. Re:A matter of access and exposure by p0rnking · · Score: 2, Interesting

    Suprnova is now registration only .... and it looks like they are using exeem now too

    "Download Torrents, Movies, Shows, Music, Full Albums and more!
    Exeem Suprnova - register now and get all of these benefits and more"

  4. Support It by Zone-MR · · Score: 2, Interesting

    If you are like me and believe that sites which simply trade hashes of illegal files should not themselves be illegal, you might want to consider heading over to www.lokitorrent.com and making a small donation to their legal defense fund.

    Who knows - if one site acheives the budget to stand up for themselves in a legal battle, it might set a very welcome precedent.

  5. double-tailed by TheSHAD0W · · Score: 4, Interesting

    The reason why the "extra long tail" is so amazingly long is because the authors are merging two different types of BitTorrent usage. BitTorrent was designed for legitimate content, and for content distributors to run their own trackers. For example, my tracker is used just to distribute my own projects. Distribution is off the main website, with only one torrent shown. This is an example of BT's legitimate use, and even the largest legitimate BitTorrent sites pale in comparison to the piracy sites. There, you'll see much higher numbers of torrents, and few servers that only distribute small numbers of torrents.

  6. Thank you all. by FreeLinux · · Score: 2, Interesting

    But, I am familiar with how to use BitTorrent and where problems can occur. I sit on a multi-megabit high-speed pipe (bidirectional). The BitTorrent client was set to limit upload speed at 110KBps, the client's default. I did try lower upload limits as well as limiting the number of connections but, it didn't change anything with the download performance. After the download completed (12 hours) I opened up the number of connections and removed the upload limit. Leachers enjoyed upload speeds of 600KBps until the up/down ratio hit 3 and I then removed the seed.

    I remain unconvinced about BitTorrents abillity to meet its design goals. Whether that is due to BitTorrent itself or selfish users, I don't know, but performance isn't what it should be and it seems to be getting worse with time.

  7. Re:Umm, no, it won't ever die. by Morosoph · · Score: 2, Interesting
    There might be a texhnical fix to this: modifying the protocol to do better than tit-for-tat, perhaps as with this feature request of mine (see also follow-up).

    If clients bias towards good propagators themselves, then they will be themselves rewarded by those who do likewise. Leechers that do not upload anywhere near proportionally are, by definition bad propogators.

  8. Re:Umm, no, it won't ever die. by Morosoph · · Score: 2, Interesting
    Which works fine for a single file. All of your stats are reset as soon as you switch to a new torrent/tracker. The only way around this is keeping all torrents on the same tracker (see: empornium), but that has its own failings.
    This is true, but if your are sniffed out within a few minutes by your fellow clients, it'll be a slow business getting any file. This makes leeching pretty pointless.

    Ratio will do it, of course, but if a techinical fix could be found, that would help to keep things open.

  9. Re:Umm, no, it won't ever die. by kitty+tape · · Score: 2, Interesting

    I wish I could be a grad student and publish a bunch of bullshit with graphs and get my degree!

    This is not an academic paper. It is just an informal paper that happens to be done by graduate students. You really should make sure you have something to be an anti-intellectual snob about before you start spouting nonsense.

    --
    ----- "Type theory is like pretzels on crack." -- random friend
  10. Re:Umm, no, it won't ever die. by Morosoph · · Score: 2, Interesting

    You don't look at what people are telling you; you infer rates from the distribution of files in the wild, and how they change. It's a problem involving stats and matrix manipulation.

  11. Re:Umm, no, it won't ever die. by Anonymous Coward · · Score: 1, Interesting

    There's just this one problem with the ratio-based things. See, as bittorrent is fundamentally a peer-to-peer application, there is no way (at least not now) to report upload/download statistics reliably. So they are reported by the client, mostly for the benefit of a statistics-gathering tracker or some such thing. (A future revision of bt could have the peers report on upload/download counts of other peers, but that has some severe privacy implications.)

    Ergo, you can pretty much fake any ratio you damn well please with a properly hacked client. Being that the reference bittorrent client implementation is written in Python, it wouldn't be all that hard to simply change the reported upload count to something like upload*2 + download*3, and no one would probably be any wiser. Sure, someone could come up with weird-ass heuristics to try and counter this, but those will eventually end up kicking honest people out too as there is inevitably some "this and this chunk had bad checksum, downloading it again" and "oops my computer crapped out before I reported having uploaded 300 megs more" type inconsistency between upload and download totals.

    No, it's entirely unreasonable to expect other people to seed your stuff. BitTorrent is more like a bandwidth-easy replacement for putting large files up for grabs on a HTTP server, a way to cope with say the slashdot effect gracefully. If you want to run a free-for-all warez hub, your best bet is to try and create a reasonably tightly knit community where people actually feel like they have a responsibility to give back since the hub and its users have already given them so much.