Well, we don't have to have any manual blocking. My way around it then would be to use any gnutella client that let me spam a whole bunch of repeated keywords and then everyone else on gnutellanet would AUTOMATICALLY add that keyword to their banned list.

I am tempted to do that right now in order to get rid of all these "*.mpg", "*.zip", ".zip", ".asf", etcetera, wasteful "denial of service (DOS)" searches that are appearing via the Monitor function. I know that they are DOS because they keep appearing in cyclic fashion from someone's automated robot. I am also seeing a lot of "a mp3", "b mp3", "c mp3", etcetera types of searches in cyclic fashion too. Whoever is doing this is pretty clever. They're starting to use a lot of asterisks in their keywords to avoid the exact name matching of people's filters.

In my feeble attempts to filter out these spammers trying to reduce gnutellanet, I end up setting 1 outgoing, 5 incoming connections in the hopes that more people would use the filtered feed from my machine. *sigh*

No one has comments on these spammers? It is almost like the RIAA or MPAA is practicing Denial of Service tactics. I thought DOS was illegal for them to practice on internet?

Wow, this thread has gotten out of hand!
I never would have imagined that this feature would raise so much controversy, although this isn't the only place it's discussed; it's come up in #gnutella IRC more than once in the past. I simply think of it as a feature, and nothing more -- I don't understand all the talk about it
I think people are too dependant upon this big gnutellaNet (what we know as the gnutellaNet, that is). Let me take you back a few months ago, when I first got my hands on Gnutella 0.45. When I first discovered what it could do, I thought, "Wow! I could setup private networks amongst a group of people and it would be totally anonymous." Not to bore you or anything, but I still keep that view in mind today. It's just my opinion that people get on Gnutella and think of it as this global file-swapping solution like Napster is/was. I really have no news to report about changes with this feature in the upcoming 0.7 release (we just fixed the BBS bug (don't ask), which means we're very close to release!) Anyway, that's just my $0.02.

-Yytrium-

As if enough wasn't said already...

I do agree that one must have the choice of whether to propagate any particular query or not. People that strongly feel towards an issue may find it uncomfortable to encourage sharing of certain information. I feel we must account for their right to express themselves.

Yes I do feel that argument stating that one should only choose whether to participate in the GnutellNet or not, rather than whether to propagate a certain type of queries, is reasonable in a number of ways. One, it makes client programming much easier. However, I feel that just because it is POSSIBLE to block some queries by any client on the network, we must account for future (or present) development of this feature.

In a word, I feel that if it is possible, it must be done or accounted for. Gee, maybe we should even have to classify each client according to the degree of restriction it puts on the queries it propagates.

Below are 2 extremes:

• A fully UNrestricted (or fully open) client will propagate any query, no matter what it is.
• A fully restricted client will propagate NO queries.

There will be a number of intermediate classifications:

• A client which propagates everything, except globally banned queries.
• A client which propagates everything, except globally banned queries, and a few manually banned queries (ding-ding, we've heard this one before from Neo_Geo!)
• A client which propagates everything, except globally banned queries, and a bunch (see below) of manually added queries.
• A client which propagates only a specific set of queries (say those that match a certain substring).

Exactly how many is a "bunch" (as used above)? We might want a formula, in which some queries may be given more weight, while others less. Weights of all queries would be added, and if the sum excxeeds a certain number (say, based on the current GuntellaNet average), client would recieve a certain classification.

If course, we can add to this thoughts on how a group of users can create a GnutellaSubnet for the purpose of sharing a custom query ban list... Gee, my mind is just running wild, can't you tell? :-)

All this sounds awfully complicated. Am I possibly adding features to the next generation of the protocol?.. Hey, some time someone has to come up with these thoughts, for them to even be considered.

But back to the present... I do feel that for now users should be able to add manual queries bans, which would persist from session to session. For each of then the user will specify whether the ban will be distributed to the GnutellaNet (in other words, will the neighbors be told to ban this query also). The bans recieved from the GnutellaNet will be added to the list but will not persist from one session to another.

I know it is easy for me to say all this, since I am not coding... Just my $.02 ... :-) Or more like $5.00 considering the length of this post...

-= onTy =-

Yytrium, I don't know How nor Who I would set up a private network with, so I am dependent on the gnutellaNet. That is why I think about the banning of certain keywords as being so important.