Rare Files Are Now Nearly Impossible To Find
 

From the Wikipedia article on Gnutella:

"...with DQ [Dynamic Querying] the search stops as soon as the program has acquired enough search results, which vastly reduces the amount of traffic caused by popular searches."

It appears that this design change has made rare files nearly impossible to find because of the thousands of malicious nodes on the network that return bogus results for every search in an attempt to lure the user to install malware or to visit malicious web sites. Limewire stops returning results after a few hundred results have been delivered, even if all of those results are bogus. Something needs to be done to prevent malicious users from making gnutella unusable.

Malicious Users
 

Freon

You said: I agree with you but, I must stress, that we are not dealing with one or two malicious users but organisations bought and paid for by the RIAA, MPIAA and their various world wide associates.

And you also have to understand that they have and do support MPs, Congressmen, etc. worldwide, have strong lobbyist in most, if not all, western governments and have also assumed, again in most countries, police-like powers.

So Freon, how do you fight something like that?



UK Bob

Here's one suggestion...
 

I have investigated this problem some more. It is possible for the user to mark the bogus results as junk, and Limewire recognizes the files so marked as junk in search results for other keywords and in future searches. For starters, the search algorithm should be modified so that results that a user has marked as junk N E V E R count toward the number of results that terminates a search. For efficiency, a "junk list" of hash codes identifying results the user marked as junk could be sent with the search code so that nodes that are several levels down don't return results matching the "junk list." Note that the "junk list" must be generated by each user to prevent malicious marking of files as junk.

Me thinks the junk lists would become very long.

I search for a file, get results, mark many of them as junk, search again. Eventually get the file I want; no need to search further. So, all those junk file hash codes remain in my junk list?

Yeah, "eventually" I often get the file after weeks or months of searching. Millions of results get transmitted all over the network before this happens. Very efficient. Lots better than long "junk lists." So I guess the current design can't possibly be improved.