On 22 Aug 2006, at 01:27, Michael Rogers wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Ian Clarke wrote:
>> Clearly connection churn is the price to be paid by higher update
>> probability with destination sampling.  Are you sure that mroger's  
>> sims
>> will tell us anything about that?
>
> I'm hoping to simulate churn once token-passing is done. I don't know
> anything about destination sampling, but I do feel that opennet could
> make load limiting more difficult for two reasons:
>
> 1) a selfish or malicious node can connect to a large number of  
> opennet
> nodes and use its 'fair share' of capacity at each one

Why shouldn't a node with sufficient bandwidth be permitted to  
connect to as many other nodes as it can legitimately serve?

> 2) a selfish or malicious node that's detected by some kind of
> tit-for-tat/reputation mechanism can just generate a new identity and
> start again (hash cash and similar mechanisms place a significant  
> burden
> on innocent users and don't deter attackers who are willing to spend
> more CPU on the problem than innocent users)

It is true that trusted connections allows for a form of negative  
reputation, which is normally not possible on the Internet, however  
BitTorrent seems to get along just fine using tit-for-tat without the  
need for negative reputation - all reputation is positive and must be  
earned.  Why can't we do the same?

> In my opinion one of the greatest strengths of the darknet approach is
> that it avoids problems of this kind by using the existing trust
> relationships between users.

Clearly it has advantages, but it has the overwhelming disadvantage  
that most users are not willing or able to find people they trust to  
connect to - hence the need for an opennet option for such users.

> Thinking out loud, is there any way to make it easier for users to  
> find
> one another while preserving some of the "social pressure" of the
> darknet approach? For example:
>
> * Each node has a flag, "public" or "private"
> * If two of my peers are public, I automatically introduce them to one
> another
> * The process is not recursive - if Alice introduces me to Bob, I  
> don't
> automatically introduce Bob to my other public peers
> * Bob's load-limiting tokens come from Alice's bucket, so Alice has an
> incentive not to introduce me to too many people (and no incentive to
> invent people)
> * If, at a later stage, I become friends with Bob and make him into a
> fully-fledged peer, he will be introduced to my other public peers and
> will get his own token bucket

The logical tit-for-tat approach is that nodes start with the benefit  
of the doubt, but from there must earn the right to make requests of  
other nodes by answering requests themselves.  This is the approach  
employed by BitTorrent, although BT has the disadvantage that the  
trust relationship must start again from scratch every time the user  
downloads a different file.  We don't have that limitation, our nodes  
can earn and retain trust over days, weeks, even months.

Ian.

Ian Clarke: Co-Founder & Chief Scientist Revver, Inc.
phone: 323.871.2828 | personal blog - http://locut.us/blog

-------------- next part --------------
An HTML attachment was scrubbed...
URL: 
<https://emu.freenetproject.org/pipermail/tech/attachments/20060822/2cee8400/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
URL: 
<https://emu.freenetproject.org/pipermail/tech/attachments/20060822/2cee8400/attachment.pgp>

Reply via email to