Untraceable File-Sharing Algorithm

**qprider** · February 17th, 2005, 09:04 AM

sup ... I'm a grad student in computer science and I thought I'd take a shot at creating a absolutely untraceable file-sharing algorithm, no caveats, no 'almost untraceable' ... just absolutely impossible to trace. I think I've come up with something pretty darn good, but the people on this forum are no doubt much more knowledgeable about this field than I am.

So, I open my algorithm up to the comments of the community :) ...

http://www.strangebunny.com/sharing.php

- Thanks,

Victor Palmer
AIM: goodtimes9863
email: [email protected]

**Afn** · February 17th, 2005, 09:48 AM

I described a system that would eliminate sharing files and replace it with a system that would search and share numerical patterns of files, but not the files themselves.

1. create all known variations of 1kb into a look up list
2. hash the file to the list
3. hash the file name
4. share the numerical pointers to the hashed list, but not the files themselves.
5. download and upload lists of pointers
6. autogenerate the file by hashing the pointer list in reverse from a database of all known combinations of 1kb or larger on the destination machine.

**qprider** · February 17th, 2005, 10:51 AM

The only problem with the algorithm you proposed is that even by sharing 'numerical patterns' of files (as you have described them) you are still legally responsible for sharing copyrighted material.

The problem is that a given a 'numerical pattern' I can always easily backtransform it to the original file ... so even though I'm not sharing the original file, I theoretically could know that I am sharing copyrighted material (by backtransforming='autogenerating', whatever), and that makes me liable.

For example, it's still illegal to share ZIPed versions of copyrighted files ... even though those ZIP files are not the files themselves. Because I can readily uncompress a ZIP file and check its contents, I could know that I'm sharing copyrighted material and I would thus be liable. In fact, if you think about it, you're algorithm is precisely another compression scheme. And looked at in such a light, your algorithm would go something like:

2. Compress the file.
3. Compress the filename.
4. Advertise the compressed version of the copyrighted file
5. Upload the compressed version of the copyrighted file (still illegal)
6. Decompress the file.

In my algorithm, I've tried to mix encryption into the file-sharing algorithm precisely to counter this effect. If a person shares an encrypted version of a copyrighted file - AND doesn't have the key to that encryption - there is NO possible way to for him to know that he is sharing a copyrighted work, and thus he cannot be liable for copyright violation.

- Victor

**notbob** · February 17th, 2005, 10:53 AM

Originally Posted by Afn

I described a system that would eliminate sharing files and replace it with a system that would search and share numerical patterns of files, but not the files themselves.

1. create all known variations of 1kb into a look up list
2. hash the file to the list
3. hash the file name
4. share the numerical pointers to the hashed list, but not the files themselves.
5. download and upload lists of pointers
6. autogenerate the file by hashing the pointer list in reverse from a database of all known combinations of 1kb or larger on the destination machine.

wow all you need is trillions of combinations and a database the size of texas to share files (and a computer with more processing power and data storage than a supercluster at los alamos)?

what are the negatives?

**Afn** · February 17th, 2005, 11:16 AM

Originally Posted by qprider

The only problem with the algorithm you proposed is that even by sharing 'numerical patterns' of files (as you have described them) you are still legally responsible for sharing copyrighted material.

The problem is that a given a 'numerical pattern' I can always easily backtransform it to the original file ... so even though I'm not sharing the original file, I theoretically could know that I am sharing copyrighted material (by backtransforming='autogenerating', whatever), and that makes me liable.

For example, it's still illegal to share ZIPed versions of copyrighted files ... even though those ZIP files are not the files themselves. Because I can readily uncompress a ZIP file and check its contents, I could know that I'm sharing copyrighted material and I would thus be liable. In fact, if you think about it, you're algorithm is precisely another compression scheme. And looked at in such a light, your algorithm would go something like:

2. Compress the file.
3. Compress the filename.
4. Advertise the compressed version of the copyrighted file
5. Upload the compressed version of the copyrighted file (still illegal)
6. Decompress the file.

In my algorithm, I've tried to mix encryption into the file-sharing algorithm precisely to counter this effect. If a person shares an encrypted version of a copyrighted file - AND doesn't have the key to that encryption - there is NO possible way to for him to know that he is sharing a copyrighted work, and thus he cannot be liable for copyright violation.

- Victor

You could add additional layers, still with any proposed system, once the file is de-encrypted at some point, you have a possible liablity issue. Even if you had an optimized hash list system, large enough database (mysql references up to a billion records) they could not claim you copied the file because you did not.

Ok, so that will not work. Now what if you had a robot or computer that viewed, watched, read or listened to a work, and then was able to recreate the work in the robot's or computer's brain (1)?

let's play with the idea... (2) a computer views the work and makes slight modifications that does not change the material, just inserts or compresses it. A computer sees an episode of ST:NG or all eps of a series, for example, and then creates a program called Star Nut and creates a synthetic version of the program retaining all elements but in a compressed or different form than the original.

After reading this, you may think I am on acid, but think of this, if copyright exists, google would not be possible, you would have to get "permission" to use any text for any reason, and fair use, a powerful and important educational device would be replaced with a system of information haves and information have nots.

And that is exactly what is happening. When you get critical mass of have nots, you get a french revolution.

**cpugeniusmv** · February 17th, 2005, 11:59 AM

It doesn't matter how many times you encrypt, compress, or rename copyrighted material--you're still sharing it, and it has to be reassembled at some point.

And as long as you're on the (current) internet, you can't hide the IP addresses of computers connected to one another (despite efforts). Sure, you can proxy all you want to...but eventually the file has to get to the person that wants it.

Complete anonymity is not possible on the current internet.

**MushroomheadXIII** · February 17th, 2005, 12:08 PM

I'm no programmer, but why not create a program to convert files into .txt and share them with random names and codes, and then have the same program to join em together while downloading?

**cpugeniusmv** · February 17th, 2005, 12:14 PM

Originally Posted by MushroomheadXIII

I'm no programmer, but why not create a program to convert files into .txt and share them with random names and codes, and then have the same program to join em together while downloading?

All files are made up of text, it's just not human-readable. The file extension will make no difference.

Add on splitting and rejoining, and you've got every multi-source p2p application already available.

**Afn** · February 17th, 2005, 01:08 PM

Originally Posted by cpugeniusmv

It doesn't matter how many times you encrypt, compress, or rename copyrighted material--you're still sharing it, and it has to be reassembled at some point.

If 60 million people think copyright is archaic, and use the system against itself, there is nothing the corporations can do to stop 50 or 100 million people. Cars at the turn of the century in the 1900's were fined for spooking horses and other nonsense laws.

Copyright will go the same way as the millions of people wake up and take control of the political process. It is just a matter of time.

**crackerjacker** · February 17th, 2005, 01:09 PM

Originally Posted by cpugeniusmv

It doesn't matter how many times you encrypt, compress, or rename copyrighted material--you're still sharing it, and it has to be reassembled at some point.

And as long as you're on the (current) internet, you can't hide the IP addresses of computers connected to one another (despite efforts). Sure, you can proxy all you want to...but eventually the file has to get to the person that wants it.

Complete anonymity is not possible on the current internet.

ditto that.
:)

**Afn** · February 17th, 2005, 01:20 PM

Originally Posted by qprider

The only problem with the algorithm you proposed is that even by sharing 'numerical patterns' of files (as you have described them) you are still legally responsible for sharing copyrighted material.

Numbers can not be copyrighted.

A split key system with multiple layers of encryption would be the best approach.

A system needs 4 things 1. wide area easy que/subscription 2. wide area search 3. multiple levels of encryption to obsure the source and the destination. 4. multiple senders that store chunks of files encrypted, but do not know what the encrypted block is used for.

An underground waste network like napshare that relays newsgroups might be a realistic target for development.

**Roamerick** · February 17th, 2005, 01:33 PM

You may want top check out "The Big Hack"

It runs somewhere along the same lines :)

http://thebighack.org/modules.php?op...rder=0&thold=0

**eivioolla** · February 17th, 2005, 03:31 PM

Originally Posted by qprider

sup ... I'm a grad student in computer science and I thought I'd take a shot at creating a absolutely untraceable file-sharing algorithm, no caveats, no 'almost untraceable' ... just absolutely impossible to trace. I think I've come up with something pretty darn good, but the people on this forum are no doubt much more knowledgeable about this field than I am.

So, I open my algorithm up to the comments of the community :) ...

http://www.strangebunny.com/sharing.php

- Thanks,

Victor Palmer
AIM: goodtimes9863
email: [email protected]

I hope you don't think this idea is new. Google for Freenet, MUTE, ANTs, I2P, TOR, Nodezilla, Entropy etc for starters. All of them use an idea more or less like this one, none of them works in practise. In other words, there has been more attempts to implement this than you can count, and not a single one of them is usable. I can't pin point the exact reason for failure, but they are all insanely slow and don't seem to scale.

**Dave_Man** · February 17th, 2005, 03:46 PM

i use tor all the time..
never had any problems..
the speed has slowed down.. but thats the price for privacy.

**eivioolla** · February 17th, 2005, 03:49 PM

Originally Posted by Dave_Man

i use tor all the time..
never had any problems..
the speed has slowed down.. but thats the price for privacy.

Oh? What do you use it for? What kind of speeds are we talking about? Maybe I should try it. I haven't tried that one, I just presumed that if it worked I would have heard about it. :)