I have an idea that I have expressed many parts of before. Personally the advancement of gnutella is moving much to slowly for me. Not that I have a right to complain as I am not a programer and thus cannot contribute to actual software development. What I try to do is offer the best ideas I can. I have brought up the idea of a blend of gnutella and freenet many times but would like to try to get some developers together to try to create it and see what it could become.

The implementation would consist of the current basic gnutella protocol with a few tweaks. Therefor the easiest thing to do would probably be to take an excellent open source client like gnucleus and modify it.

All clients on the network would be required to share some amount of harddisk space for network usage. This space would be used to store encrypted files (or parts of files). The amount of space would probably be at least 100mb but should maybe be more. Users would also be able to share unencrypted files just as they currently do, though this would be optional. Probably users should not be able to download a file bigger than the space they are willing to share with the network.

All comunication between clients would be encrypted, including searches. If a client initiates a download it should not know if it was downloading an encrypted file or an unencrypted file. Some percentage of transfers would be proxied (maybe multiple times), thus clients would not be able to be certain of the source of the file (let alone if the source even knew it had the file).

Parts of files or whole files could be stored in the encrypted space. In addition the network should allow fot the sharing of partially downloaded files. Therefor the network should base hashing on the TigerTree method. This would also solve the problem of changing hashes due to small file changes as the rest of the unchanged file could be swarmed.

The protocol should also be tweaked to allow for better searching power (+, -, " ", ect), superpeers, metadata, distributed network file rating, ect.

It would also be great to see this as an explorer plugin so that people could host pages and files on the network and link to them on the web. Since it is a plugin people would have to have it running to use the links to the FreeTella network and they would be giving back at the same time.

I think the benefits of such a network should be obvious. Clearly it would be much harder to stop and network resources would be greatly increased. Freeloaders would be eliminated but people would not have to worry about not having any of their own files to share to get on the network (they are sharing resources).

Questions and comments are welcome.

If anyone is interested in helping get the idea of the ground let me know. It would be great.

As a person, who tried to take-n-modify Gnucleus code I must say: it's not exelent at all! Both coding style and design leave much to be desired. I am really sorry, but it is to be said.

As to the rest of the message -- why dont you want to call all this "Gnutella Protocol v0.7"? Or may be 0.8? With regard to Free-net like "encripted local storage" I have serious objective: I dont want to store and distribute anything like child porn for example, even if it is encripted. Any ideas how can I not do that in the mentioned p2p net?

--Max

The question is, why encryption?

Do you think that you can share you file in a secure way?

NO!

The RIAA needs only a client... they try to d/l from you - see which file it is... and they have you!

So please tell me why should the devlopers waste their time for this?

Morgwen

Just as with Freenet, if all of the clients have encrypted files (that they dont even know what they are) they cannot be held responsible for that content. And because of proxies you cannot even say for sure what client certain content came from.

Yes,

but you need only a client to see what this file is...

how do you choose which file to download?

And when the file is downloaded they know exatly the file content and the source!

I didn´t test freenet yet!

Morgwen

The easiest way to do this would require a closed source (or at least partailly closed) client. Basically they client would be able to decript the files but would not permit the decreption of the files stored in the shared/encypted directory.

Because the client would be able to decript the files it could do many things, such as creat an encrypted list of files hosted and the meta data. Searches would then be able so proceed as normal but the client would have to look at the encrypted list of files.

It is all very possible and very doable but the more I look at the great work going on over at www.freenetproject.org the less use I see in this. Freenet just needs to come up with a method for searching. If you havent looked into it you really should. Its not only the future of file sharing but of the NET!

Ok, I will have a look...

Morgwen