Dynamic Querying solutions - back on track

I just want to get this discussion back on track. 

Topic: How can we make Gnutella more spam resistant? 

So here are the ideas from the brainstorm again and I want to ask you to give 
each one a glance and then post your thoughts on each. 

I know, I can't really expect everyone to look at them, but I think it is 
vitally important for Gnutella right now, and every comment can be a step on 
the way to get DQ fixed. And this list has been quiet for far too long. 

Possible solutions to Dynamic Querying from the brainstorm (without a promise 
to be useful, because that's the nature of a brainstorm, but open to 
discussion):

- Eliminate most/all of the Out Of Bounds query hits (that are sent
directly from the query match source to searcher) from Dynamic Querying.

- Give UltraPeers a chance to sanity-check the data sent from leaves.
Bogons, LAN ips, hostile ip ranges (according to the UltraPeer's own
blocklist) could be blocked automatically instead of forwarded on.

- If a Leaf or UltraPeer sends an UltraPeer a HUGE list of (seemingly
good) query hits, search it last for future searches. "Chatterboxes"
aren't good for maintaining reasonable network traffic anyway.
-> "http://infolab.stanford.edu/~byang/pubs/guess.pdf" (input from Bill 
Pringlemeir, please check his mail!)

- Have each Leaf and UltraPeer rate all its neighbors based on whether
they have received any good query hits from them and especially
whether a download from/upload to them works.

- Drop Dynamic Querying.

- Implement Credence and ignore bad result for DQ ( http://credence-p2p.org
) (I added that just now)


Please comment on them.

We need to find a suitable solution to spam, because it really blocks DQ
and Gnutella in general.

Wishes,
Arne


[Non-text portions of this message have been removed]