> IC> The entire Freenet mechanism depends upon the "path
> IC> compression" effect which you get when data is requested,
> IC> however (as was pointed out the last time this was suggested)
> IC> with this mechanism this path compression would not occur.
>
> I learn new stuff about Freenet all the time, so I have to ask: I
> don't understand why it -depends- on that. I figured that if I had a
> network like this:
>
> A - B - C - D - E
>
> and A requests data from B that lives on E, the data will be dragged
> back through D, C, and B and cached before it gets to A. It seems that
> if A re-requests the same key, it could get it either from B or from
> E, and it'd be both places. Although I realize it accelerates things
> greatly, I don't see why requests -have- to go to E the next time
> around.
The point is that nodes are supposed to "learn" where data is elsewhere in
the network, by noting where data which has been requested through them
originates (this isn't entirely accurate, but is approximately so). If
you don't allow nodes to learn about new nodes, then they won't learn
about the location of data elsewhere in the network, won't be able to
route requests effectively, and you basically end up with an even crapper
version of Gnutella.
> IC> I think that it will always be possible to "fish" for IP
> IC> addresses (ie. build up a list of Freenet nodes but not in any
> IC> directed manner), but the likelihood that for any given IP
> IC> address it will be possible to shut it down is slim.
>
> I think you're purposely putting blinders on, Ian. Do I really need to
> prove to you that ISPs will cancel an account at the simplest
> complaint? There are just tons of examples, ranging from DeCSS to
> Hotline to Napster.
How does my ISP, given my IP address, cancel my account (I am on cable)?
What difference does cancelling an account make in a country like the UK
where there are hundreds of free ISPs which require nothing more than an
email address for sign-up? (And where the DCMA doesn't apply anyway)?
What would prevent abuse of this mechanism such as giving the IP address
of someone you don't like to Media Enforcer (and - assuming that M.E even
bother to check that there is a Freenet node running, install a virus to
pretend to be Freenet)?
How does an ISP avoid getting sued for cancelling someone's account given
the complexities of the entrapment issue? Remember, courts are quite fond
of compensating the little guy versus the big corporation.
I don't deny that in theory some people *could* get their accounts
cancelled - but can you imagine any better positive-publicity for Freenet
than that? Every script-kiddie outside the US (and many inside) will be
creaming themselves over the prospect of installing something that
violates the evil DMCA, just look at the publicity boost which the RIAA
gave to Napster (if they had kept quiet most people would never have heard
of it), or the number of people who dutifully downloaded decss even though
they have no idea what to do with it!
Ian
PGP signature
