fish wrote:<snip>
On Sun, Nov 09, 2003 at 08:16:08AM -0800, Martin Stone Davis wrote:
For the greedy operator to benefit substantially, he has to create new IP addresses faster than it takes to figure out that he's being greedy. So if he can get a new IP address only once a day, but it takes us as little as 20 minutes to cut him off for being greedy, then we can tolerate it, and negative trust can work.
If the user is smart, it's not very hard for him to convince pretty much any
DHCP server to give him a new IP every few minutes, especially with PPPoE
based services (such as 90% of DSL) - just bring down then up your PPPoE
link, bam, new IP.
You guys are getting too complicated with this. Adopt a self-centered perspective - I detect another node 'abusing' my node. Given the choice, should I do nothing, or make an effort to discourage this behavior ? We have already agreed that "blacklisted" node information cannot be safely exchanged. Now if every (benevolent) node takes this approach, it makes life _a_little_ harder for Mr. NastyNode. And we need to be careful in classifying malevolent behaviour. Tracking IPs is basically worthless - I'm much more concerned about organizations with the resources to construct high-powered clusters and make coordinated efforts to thwart the Freenet. Do you remember the fictitious evil AAIR regime ? I forget, which country do they rule ? All this 'negative trust' everyone keeps referring to means that Mr. NastyNode will simply go find a new connection. But what if every node keeps opposing him (to a degree) ? It could be made to cost him some time to get 'integrated' with the network at each connection step. That's why I believe he needs to demonstrate good behaviour before he can attempt to do much damage. And if the remote nodes are measuring his behavior, he won't get very far in a short time period. It's not a perfect solution, but at least it begins to address the issue. Some day we may allow a user to maintain a list of "trusted nodes" that would be permanent members of their RT. The more this is used, the stronger the network becomes. It would seem that for this to work, both sides of a trust relationship would need to specify the other node in their list of "trusted nodes." So what would differentiate a trusted node from a regular node ? Priority in query processing, mainly. Thoughts? (beyond "there is no benefit to negative trust")
The topic was on negative trust. We resolved that negative trust won't work. You are talking about positive trust. That's a different issue. ATM, I'm clueless on that.
This thread is closed, as far as I'm concerned. The answer is "no".
-Martin
_______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
