Posted by: Anonymous Coward
on September 04, 2003 07:39 AM
So glad to see this thread here!
Yes you are on the right track. I have been working on a project that is pretty much this very idea since the start of 2003. DINT (Distributed Information Network Technology) is the rather stopgap handle I gave to the concept at the start. I have teamed up with another computer scientist to work on an open source implementation. We are still recruiting hackers to help with this. At the moment the project is at the detailed RFC stage with a good analysis of the algorithmic framework.
There are many issues that I will outline beifly.
1) Legal: By splitting data into thousands of distributed chunks users keeping such data are immune under current copyright laws. The enforcement of any case against the power of collective resposibilty on a massive scale renders the current law unenforcable to all intents. Another interesting effect is that a browser can be made such that the full copy of a copyrighted work never even exists on the users machine...it is effectively streamed down and erased from FIFO buffer on the fly. This is a true 'Ghost Network'.
2) Practical: A distributed archive system, like a hologram, the human brain and the internet itself as originally concived by ARPA, is effectively bomb-proof. Data encoded in DINT would be 'unerasable' once distributed. Having no localised data means no single point of weakness. As a concept it has side effects far beyond filesharing, including ultra-safe archives for proprietry data, or even ways for journalists working in dangerous conditions to disseminate material without personal risk or fear of information being "shut down".
Here are some of the issues we have studied:
i) Architecture
One of the most promising ideas is a algorithm based partly on Conways 'Life' and Kohonens self organising maps. Packets have a kind of inbuilt intelligence that allows them to know about their neighbors on other machines and how long to stay alive, when to replicate and how to reassemble themselves back into the original data given a catalytic 'key', a term we refer to as recrystalisation.
ii) Poisoning
In such a open system there are a myriad of issues to deal with concerning security and data integrity. One of the problems is incorporating hashes and other checks that allow data to avoid being poisoned or impersonated. No doubt the RIAA etc would not think twice about employing such abuse if holes existed.
iii) Namespace
How to prevent unintentional crosstalk between similar files distributed in the same space has plauged our studies.
iv) Transactional value
Given that each user would have to allocate an area of public space what incentive can the system give. Also consider that the methods are non deterministic, no guaranteed delivery times, or even certainty that all the data can ever be recovered at any time. This is somewhat mitigated by the redundant distribution of the data making it 'Fuzzy' , in other words you might be able to recover a partial version of a time based file like audio which although 'complete' in time is of lower quality (like a dirty CD).
My job means I cant dedicate nearly as much time as I want to this project. I hope to publish a detailed RFC towards the end of the year. In the meantime I am very interested in hearing other ideas or from groups and individuals also working on this concept. To see other people on the same road is very encouraging, I truly beleive this is the future of filesharing communities in an opressive 'intelectual property environment'
email me and exchange PGP keys if you want to talk about this in more detail.
Learn about Linux
Download Linux
Get Linux help
Re:Cutting record companies $ is GOOD
Posted by: Anonymous Coward on September 04, 2003 07:39 AMYes you are on the right track. I have been working on a project that is pretty much this very idea since the start of 2003. DINT (Distributed Information Network Technology) is the rather stopgap handle I gave to the concept at the start. I have teamed up with another computer scientist to work on an open source implementation. We are still recruiting hackers to help with this. At the moment the project is at the detailed RFC stage with a good analysis of the algorithmic framework.
There are many issues that I will outline beifly.
1) Legal: By splitting data into thousands of distributed chunks users keeping such data are immune under current copyright laws. The enforcement of any case against the power of collective resposibilty on a massive scale renders the current law unenforcable to all intents. Another interesting effect is that a browser can be made such that the full copy of a copyrighted work never even exists on the users machine...it is effectively streamed down and erased from FIFO buffer on the fly. This is a true 'Ghost Network'.
2) Practical: A distributed archive system, like a hologram, the human brain and the internet itself as originally concived by ARPA, is effectively bomb-proof. Data encoded in DINT would be 'unerasable' once distributed. Having no localised data means no single point of weakness. As a concept it has side effects far beyond filesharing, including ultra-safe archives for proprietry data, or even ways for journalists working in dangerous conditions to disseminate material without personal risk or fear of information being "shut down".
Here are some of the issues we have studied:
i) Architecture
One of the most promising ideas is a algorithm based partly on Conways 'Life' and Kohonens self organising maps. Packets have a kind of inbuilt intelligence that allows them to know about their neighbors on other machines and how long to stay alive, when to replicate and how to reassemble themselves back into the original data given a catalytic 'key', a term we refer to as recrystalisation.
ii) Poisoning
In such a open system there are a myriad of issues to deal with concerning security and data integrity. One of the problems is incorporating hashes and other checks that allow data to avoid being poisoned or impersonated. No doubt the RIAA etc would not think twice about employing such abuse if holes existed.
iii) Namespace
How to prevent unintentional crosstalk between similar files distributed in the same space has plauged our studies.
iv) Transactional value
Given that each user would have to allocate an area of public space what incentive can the system give. Also consider that the methods are non deterministic, no guaranteed delivery times, or even certainty that all the data can ever be recovered at any time. This is somewhat mitigated by the redundant distribution of the data making it 'Fuzzy' , in other words you might be able to recover a partial version of a time based file like audio which although 'complete' in time is of lower quality (like a dirty CD).
My job means I cant dedicate nearly as much time as I want to this project. I hope to publish a detailed RFC towards the end of the year. In the meantime I am very interested in hearing other ideas or from groups and individuals also working on this concept. To see other people on the same road is very encouraging, I truly beleive this is the future of filesharing communities in an opressive 'intelectual property environment'
email me and exchange PGP keys if you want to talk about this in more detail.
NOSPAMprisonernumbernineNOSPAM@yahoo.NOSPAMcom
#