On Thu, May 11, 2006 at 12:57:57AM +0300, Jusa Saari wrote: > On Wed, 10 May 2006 19:31:56 +0100, Matthew Toseland wrote: > > > On Wed, May 10, 2006 at 09:09:10PM +0300, Jusa Saari wrote: > >> On Tue, 09 May 2006 00:12:47 +0100, Matthew Toseland wrote: > >> > >> > When did you try it last? :) > >> > >> Well... It has been a while. Maybe I'll give it a try again, when and if > >> 0.7 non-darknet version becomes available. > > > > Interesting (and common) attitude... connecting to a bunch of strangers > > through opennet is better than connecting to a bunch of strangers through > > darknet how? > > Three reasons: > > 1) I don't have to hunt for more node references after the initial > connect. Since no node will stay online forever, a darknet nodes > connectivity will deteriorate with time unless its given more noderefs > regularly. This is an extra hassle, likely resulting in most people using > some kind of automated means (possibly a Frost board with noderefs, or > something like that), making the "dark" part of darknet completely untrue.
The hope is that now that the /. crowd have disappeared or assimilated most nodes will be semi-permanent. And yes there are many Frost boards for noderefs. :( > > 2) Related to the above, if I take my node offline for any reason (as I do > every night - this computer can't run 24/7, it's in my bedroom and makes > a noise like a jet engine), it won't result in significant reduction in > the connectiveness of other nodes. A darknet node will likely have very > few connections, since getting them is such a hassle, while a non-darknet > will have dozens, if not hundreds, of nodes to connect to, and is > constantly getting more. An opennet node will not likely have hundreds of simultaneous connections. We will probably set the limit fairly low, say 30. > > 3) If it turns out that one of the people I'm connected to is a pedophile > and gets busted, and I was connected via darknet to him, it will cast > shadows of doubt on me. After all, I chose to connect to this particular > person; even if it can't be proven that I'm a pedophile myself, I'm still > associated with one. On the other hand, if our nodes simply established > connection automatically without any interaction on my part, as they do > in a lightnet, I'm not associated with a pedophile - it is simple to show > that the whole thing was completely automatic, and didn't require me to > have ever even heard of this person, much less interacted with him in any > way. > > This last point is especially important: a darknet can *increase* your > risks. In a lightnet, you are simply connecting to a *network*, and while > your node can be sending IP packets to a pedophile or a terrorist or a > general scumbag, it has nothing to do with *you* - just a conincidence, > you aren't associated with these people anymore than if your IP packets > had happened to be routed through a router operated by Mafia. On the other > hand, in a darknet, you are connecting to *specific* *people* **you** > *selected*. You are supposed to connect to people you know and trust; you > trusted a pedophile/terrorist/scumbag - so maybe you are one too ? You > routed your IP packets through a machine operated by Mafia *on purpose*; > maybe you didn't know that it was operated by the Mafia, or maybe you did. > > It is a lot harder to claim that you knew nothing about the illegal > activities of someone you deliberately associated with, than some random > stranger you happened to pass on the proverbial street. And while > associating with scum may not make you guilty in a court of law, the court > of public opinion is quite a different matter. I suppose there is a certain level of truth in this. > > >> > Frost is primarily about boards, and it can't be easily gatewayed to > >> > regular email because it doesn't have the same features. Something > >> > that > >> > >> Actually, it does. Assuming email-over-Freenet is going to use the > >> insert/request model (and not some kind of direct messaging), then it's > >> going to require message senders inserting messages with guessable keys > >> and message receivers polling for them. Basically, any > >> email-over-Freenet application is going to be, in essence, a message > >> board. > > > > Fundamentally yes. That doesn't mean it uses the same data formats and has > > the same features. It doesn't for example mean it uses one queue per > > board. > > What essential feature neccessiates these different data formats ? What > does e-mail do that Frost doesn't do ? What justifies the effort of > writing and debugging this new freemail application, and the inconvenience > of having two incompatible messaging systems ? > > And don't say "1-on-1 messaging", since that is simply not possible on > Freenet. It isn't? Why not? > > >> > could be would be useful; we could gateway the lists, for instance, > >> > and save people the considerable trouble of setting up 2-way mixmaster > >> > accounts. It's also been specifically asked for by rguerra, who has > >> > considerable experience and contacts amongst people working in dark > >> > places. > >> >> > >> >> Or, if you want to use an email program for communication, add POP > >> >> protocol to Frost. It is open source, after all. > >> > > >> > Frost doesn't do the same thing as email. > >> > >> Yes, it does. It lets me send a message to another human being over > >> Freenet. That is what email does. > > > > Frost is many:many and doesn't support email headers. Because of the > > latter it can't easily be gatewayed to real email. Because of the former > > it is inefficient for 1:1 use. > > All communication over Freenet is 1:many or many:many. There is no keytype > that would let many people to insert messages for you but let none but you > to read them. So what? They're encrypted. They're stored temporarily on the network, so that we don't have to have a permanently online email server to connect to. > > As for e-mail headers, they are simply lines of text inserted at the begin > of the message. Apart from address (To: field), there's no special support > required from the transmission layer. Nonsense. Technically maybe you could insert the headers at the beginning of the message. In practice, Frost messages do not contain anything resembling the References: field and therefore cannot be cleanly gatewayed to email. They also typically include the entire thread up to the current message, above the reply, in order to compensate for this weakness. They are not compatible. One or other needs to change. > > >> Simply setup a board to act as your inbox, tell it to people along with > >> your public (Frost) key, and you're done. > >> > >> You aren't going to get a direct 1-to-1 mapping with regular Internet > >> e-mail with Freenet, since there's no way to send messages directly from > >> one host to another. > > > > I don't see why we can't have a 1:1 mapping. It may not be very efficient > > and elegant, but it's been done before. In 0.8 we will use various > > mechanisms to make it more efficient, but IMHO these should probably be > > passive requests rather than 1:1 connections. > > BECAUSE, there is no keytype that supports that. So what? > And it's simply pointless > to waste effort to try to make messaging over Freenet look just like > e-mail. People don't want to send e-mails, they want to send text (and > occasional attachments). People want to send e-mails. Ask the average person what they do with the internet; the killer app is email. It always has been email. > > Sure, you *can* make it look to the user that he's using e-mail. You can't without breaking Frost compatibility. > However, > it's a wasted effort for appearances sake. Also, since we are talking > about Freenet, the whole thing is going to be very untrustworthy, with > messages getting lost on the way. > > Hmm... That may justify writing a separate application: reliable (retry > sending until confirmation is received or timeout is encountered) > messaging. Freemail does indeed retry. In fact, ordinary email retries. The difference with freemail is it's virtually immune to traffic analysis. And if well written it can be fairly resistant to intersection attacks too. Which even email over I2P isn't. -- Matthew J Toseland - toad at amphibian.dyndns.org Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20060511/2804441b/attachment.pgp>
