[freenet-chat] Arguments against the Darknet

Roger Hayter roger at hayter.org
Sat Jun 24 16:50:34 UTC 2006


In message <6E346768-CB71-4AAE-94EA-422C2F5AE3E1 at sq7.org>, Colin Davis 
<Colin at sq7.org> writes
>I'd like to use this opportunity to disagree with the current .7 
>strategy of the darknet- I've done it before, but this is the Chat 
>list, so It's not Off-topic to have a discussion about it.
>
>
>I'd like to start of by admitting that I probably know the least  about 
>the subject, compared with anyone else in the room. This isn't  a 
>pissing match, I just want freenet to be the best it can be.
>We all do. That's why we're here.
>
>The darknet model should be supplemented with an opennet SOON- I 
>personally know 4 friends of mine who I have spoken with in person, who 
>have not wanted to use freenet in it's current state. I'll  attempt to 
>discuss some concerns below.
>
>I've tried to intend my thoughts below, for ease of reading.
>
>
>
>The first is the idea of trust in Freenet .7 is complete- You argue 
>that you should only join with your real life friends,
>That your peers are people that you trust, but this breaks down for a 
>few reasons.
>       1) The first is that it doesn't fit the Social networking model 
>that  we're looking to model after.
>       In Milgrim's experiments (which, btw, were seldom as successful 
>as  his first attempt), he found that the best results came from the
>       occasional long distance link.
>                       a) In real life we often have these- For 
>example, I am < 4 steps  from Bill Gates- My Pastor knows a friend of a 
>friend, etc.
>                               The problem is, these long distance 
>links, such as My Pastor,  aren't necessarily people I trust on a deep 
>level.
>                       b) These weak links are often Largely grouped- 
>Again, My Pastor-  There are several hundred people in his congregation.
>                               If each of these people linked to him, 
>we'd get ubernodes, which  you disagree with (See my other e-mail).
>       2) Trust isn't universal
>                       a) The freenet .7 model gives them complete 
>trust- I trust my  flatmate not to download CP, but I don't trust him 
>not to download an
>                               illegial MP3 file. Do I link with him? 
>Do I need to find people  with whom I agree about everything?
>                       b) People are desperate- Think about our chinese 
>dissident- He  wants to learn more about the Western world, and to 
>write and publish
>                               about democracy. So he links with other 
>people who are writing  about democracy.. He wants the information. But 
>he knows that a  number of them
>                               are otherwise untrustworthy people. Even 
>though they all share a  love of democracy, should they link to one 
>another?
>       3) We may never get to the point that freenet is "Big enough"
>                       a) Sanity has argued on a Gmail model, where 
>freenet is the "in- thing", and people are looking for ways to get into 
>it.
>                               I) The number of people who want a 
>truely anonymous network are  far fewer than the people who want a GB 
>of free e-mail.
>                               II) And even if we DID get that level of 
>success, those links  weren't traded to people that were trusted. there 
>were automted gmail  invite traders
>                                       Gmail invites were sold on eBay, 
>etc. It's not the best model  for emulation.
>                       b) People go to the path of least resistance-
>                               I) It's always going to be easier to go 
>to #freenet-refs, than it  is going to be to find friends who use the 
>service. Promiscuous  linking is just easier!
>                               II) More people care about speed than 
>absolute privacy- Look at  how popular BitTorrent is.
>                                               * The best model will 
>allow top speed for people who don't care  as much, but more privacy 
>for people who do
>                                               * The right way to bring 
>this about is be KEEPING THE DARKNET,  but layering an openet on top of 
>it.
>                                               * People who care a lot 
>about privacy use the darknet- People  who are desperate or want speed, 
>use the OpenNet.
>       4) People are afraid of the Darknet
>                       a) People are afraid of manually choosing to link-
>                               I) For the reasons we outlined earlier, 
>people don't know 100%  that their friends can be trusted in every 
>domain
>                               II) Because of that, they are afraid to 
>link with people. they  don't want to be associated with a "Bad Guy"
>                                               * Rightfully or wrongly, 
>they are worried (I've had actual  potential users say this), that they 
>will link with a person who,
>                                                       unknowingly to them when they linked to him, does Bad Things.
>                                               * Freenet has always had 
>a bit of a public scare because of CP  concerns. I'm not going to 
>debate that here.
>                                                       But people don't 
>want to SPECIFICALLY, and KNOWINGLY  ---- CHOOSE--- to link with a 
>person who might do CP.
>                       b) We need money!
>                               I) We need users to get donations.
>                               II) Users are afraid of the darknet
>                               III) It's worth spending developer time 
>on things that will  improve the number of people in the network, and 
>number of people  donating,
>                       c) Implementing Opennet will help get more 
>users, which helps get  more donations.
>
>I'll be happy to discuss this with anyone who's interested. It's a 
>serious issue, and I'm trying to go about things the Right way.

FWIW, I agree with all your points. And I would add that no-one is more 
than 2 steps away from a police spy - I find random connection *adds* 
plausible deniability:  although not (and this is a valid point that has 
been made by the developers) if running Freenet is itself a crime.  But 
if every friend has at least one friend who is a police spy, they are 
going to know you are running Freenet anyway.  The only defence is to 
have so many people running Freenet that they don't bother to prosecute 
unless they already suspect you of something, in which case they will 
always find something to prosecute you for if they want to anyway.

But is not the routing model for Freenet 0.7 dependent on some sort of 
affinity network rather than the old open/random connection model?
-- 
Roger Hayter



More information about the chat mailing list