Re: returning results for partial downloads
Michael Rogers <[email protected]>
| Newsgroups | gmane.network.gnutella.devel |
|---|---|
| Message-ID | <[email protected]> |
Serguei Osokine wrote: > But is the best fighter algorithm necessarily the best one for > the network as a whole? Let's say you have two algorithms: 'generous > giver' and 'shameless leecher'. You load them on the network, and > shameless leacher beats the generous giver hands down in you metrics > of bytes downloaded per bytes uploaded. But does it mean that the > shameless leecher algorithm has any useful qualities? Not necessarily. > > Am I missing something in what you're going to do? You're right, it's possible that selfish algorithms won't benefit the network, but in the context of the prisoner's dilemma all of the successful algorithms have been both selfish and cooperative. All-out defection (ie leeching) is only successful in short games (which could be another reason for making the decision about who to share with as fine-grained as possible - more "rounds" in which to determine who's reciprocating and who's leeching). Leechers beat generous givers hands down, but they don't beat utility maximisers, because utility maximisers prefer to upload to other utility maximisers in order to receive downloads in return. They quickly determine that leechers offer nothing in return, so they don't upload to them - as in BitTorrent, that means leechers get little or no service. They aren't excluded from the network, just passed over. Cheers, Michael