If the files are no longer refered to by their names, but rather by their hashes... Would there still be legal concern from the RIAA?

If the RIAA only sees that they are downloading part of a file that they can only see by it's hash, then would the liability on the person who is sending the file be reduced at all?

This would confuse the hell out of em', especially if network clients started aiding in the distribution of files accross the network.

As for implimentation, a client returning a query response to a text based message wouldn't identify their location to the requesting party.. Just that the file exists and what the hash is. Only after requesting a part of the file would the requesting client know who COULD have sent the origional query response.

__________________
- Nick Catalano

Gnutella is a *general* file search protocol. However, if you are introducing features that specifically hides the illegal use of Gnutella, you are narrowing this general use down.

Doing so, it will be more likely that Gnutella will be the focal point of legal actions. You will not be helping out other people this way, as some may like to believe, but only digging your own 6ft hole.

Someone else suggested a similar thing for Gnutella, which is supposed to help people against legal actions by the RIAA. I wouldn't care what other Gnutella developers decide, but I will NOT include any features that protect piracy or child molesters (yes, those features will help them too). If this someone decides to sue me for that, so be it. I rather take on a lawsuit from this someone than the RIAA or a similar agency, or the Government itself. It's not about having "enough balls" to do it, it's about common sense.

And it is certainly not that I don't want to deny an end user what he believes is his right to do with Gnutella. It's fine by me if he or she decides to use it for piracy. But if piracy is illegal in the country he resides, it is also his responsibility to prevent getting caught and to assume the consequences. I'm not to police Gnutella or the Internet as a whole. I have better things to do.

__________________
Okay, so I've been gone a while - but hey, meanwhile online translators have gotten better:

Ce n'est pas ma faute. Blâmer vous-même.

----

Ne me blâmez pas. Blâment votre individu.

This also increases proformance. Dealing with only partial hashes allows more files to be shared more effeciently.

__________________
- Nick Catalano

Could you explain that a bit more? I think I've missed something.

__________________
Okay, so I've been gone a while - but hey, meanwhile online translators have gotten better:

Ce n'est pas ma faute. Blâmer vous-même.

----

Ne me blâmez pas. Blâment votre individu.

The entire idea here is that the searching client doesn't know WHERE the file is when they search for it, just that it is indeed there and what the hash is.

Then, a search for part of the file is proformed (after the user selects which file he/she wishes to download) and that is done completely on hash. Thus you arn't nessecarily dealing with filenames or any identifiable information, but just hash numbers. Plus, after an open URL syntax is made a lot of people will find their files from HTTP-based sites, so an all hash query method wouldn't be completely offbase.

__________________
- Nick Catalano

add a bit of auto-mirroring and proxies and poof ;-) What did you download? Did you download anything or just mirror something? Maybe you acted as a proxy? Who knows?

But wouldnt it be hard to have all the hashes in mind?
Would you remember that DJ Zyclon - Theme From Myrher
in a hash code?
Well, if there's a sort of translator it would be ok.
hm...

__________________
Never touch a running system.
www.guox.de

its not the file name, its the content. Now if you add a few extra bytes to the file, that will change the hash but ruin any chance of people finding the file. Catch 22.

Well I think I forgot that
But then how do you think are people gonna find those files?

__________________
Never touch a running system.
www.guox.de

There would simply be two queries.. One returns metadata and a hash.. the other is for a hash and returns locations of files..

__________________
- Nick Catalano

Nick, wouldn't that increase the bandwidth consumption a lot? Gnutella developers finally managed to reduce the flow with the ping/pong cache scheme, and now we're talking about increasing it again

__________________
Okay, so I've been gone a while - but hey, meanwhile online translators have gotten better:

Ce n'est pas ma faute. Blâmer vous-même.

----

Ne me blâmez pas. Blâment votre individu.

Nope.. Reduce traffic actuially.. Since you would only do a search for hash when you allready know what file you are downloading. Otherwise the clients need to search using normal search methods, in order to find a hash.. No real increase in traffic..

Also, since a search by hash would be so amazingly small, the TTL could be upped to say 10 or 12 so that a client can find multiple locations of a file.

__________________
- Nick Catalano

Of course there would be an increase. You said "There would simply be two queries.. ". One for the initial search (plain text results), the second for the hash.

But as each client responds to a Query, they already include the hash as defined with HUGE. So you already have all the locations the search has to offer. If you send out a new search with just a hash from those result sets, the same ones that sent them in the first search will be sending a result back to you. That's double work for the same results you'll get.

The only exception is when you search using the hash on a new connection. But that's not much of a use. Once you connect to a client to download, you can get alternate locations based on the hash. So you automatically expand your horizon for that particular file. You could even do that *without* any further searches or even download.

Now I'm wondering though: You are familiar with HUGE, right?

__________________
Okay, so I've been gone a while - but hey, meanwhile online translators have gotten better:

Ce n'est pas ma faute. Blâmer vous-même.

----

Ne me blâmez pas. Blâment votre individu.

With this one, you get a response for your first search with just metadata about the file.. This metadata can come from any source and isn't dependent on Gnutella.. So the serving clients don't nessecarily have to announce they have a file, they just need to announce (to their Ultranodes) that they have xyz hash.

I have a rough understanding of HUGE.. The idea here is that this would be two independent search channels.. One for metadata and another for hashes.. Thus you can only confirm that a specific client has xyz hash and not the actual file xyz. Makes it much harder to find specific users who are sharing the same file because you literaly have to download every file you wish to track and even then with automated file distribution, partial files, and (with the new GSP spec.) streaming+chaining there will be NO way to confirm that a user has a file.

As for extra bandwidth, the first search (that is done now) returns thousands of results.. If you turn arround and search again for a specific hash you are looking at 20 repsonses being sent back. Also, since hashes are smaller than normal metadata the TTL can be much higher and thus you will find more locations of a harder to find file.

I am more wondering if it would really make any difference from a legal standpoint than one from a technical standpoint. No sense researching something that would hurt more than it would help.

__________________
- Nick Catalano

Ok, I see where you're getting now. I still have some reservations though.

But what does the metadata return? It should contain some sort of human readable identification of the file, such as a description or filename, correct? If not, how do you know the file you selected is the one you want (you'll see a lot of unrelated matches in Gnutella itself - you search for music, but get porn in return). If you send out the human readable identification, you have the matching hash, it'll be easier to "backtrack" illegal files.

__________________
Okay, so I've been gone a while - but hey, meanwhile online translators have gotten better:

Ce n'est pas ma faute. Blâmer vous-même.

----

Ne me blâmez pas. Blâment votre individu.