Decentralizing wiki technology: Difference between revisions
wikademia>Remi added see also and external links |
wikademia>Remi m →External links: fix link |
||
| Line 31: | Line 31: | ||
==External links== | ==External links== | ||
* [http://www.gossamer-threads.com/lists/wiki/foundation/121420 'Possibility of a git-based fully distributed Wikipedia' | * [http://www.gossamer-threads.com/lists/wiki/foundation/121420 'Possibility of a git-based fully distributed Wikipedia'] | ||
Revision as of 17:45, 18 February 2008
Decentralizing Wikipedia and its sister projects is a project to create a framework of how decentralizing the technical details Wikimedia projects may work. This is just an intellectual exercise. It may or may not work in practice. It may or may not be practical, and it may or may not be a feasible method of reducing the load on central servers and allowing for greater capacity and empowerment of content and users, respectively. This could maybe be a fork of BOINC, GnuTella, or FreeNet or some synthesis of the aforementioned.
It might work something like the following. Anyone can setup tracker and content servers (that runs as part of Wikiversity Screen Saver or just as a service in the background). These servers and nodes track changes and where you can find articles on the network.
Trackers could be large or small. They could be dedicated machines, or run in the background of peoples computers.
For example, since Wikipedia is quite large, Wikipedia would probably have the most tracker servers.
Wikiversity would probably have considerably fewer, but could be dynamically balanced and tweaked according to need.
When someone views an article, they basically ping the tracker server, saying in computer talk, "I want to see an article on Cows".
Their machine could attempt to tell this to multiple tracker servers, and could hopefully get a rather speedy response.
The tracker server would say, "In order the amount of trust you can put into the integrity of the article, you can find the most up to date revision of the article on Cows on the servers x, y, z, and c".
The person who requested the article on cows would then send a message off to machines x, y, z, and c". Hopefully, x would respond first, in a speedy manner and the article would pop up speedily.
Presuming this goes right, the individual may decide he or she would like to edit the article.
They make their changes and send them off to a few of the servers that contain the article on cows. This would then be propagated and perhaps reviewed by editors and then integrated into the best version.
Maybe it could simply be integrated. Suppose someone came along and saw that it was vandalism. They could revert it as normally is done, and then that revert could be propigated.
Maybe OpenID could be used to verify identities. That is an interesting challenge. How could identities be trusted on a decentralized system. OpenId?
There could be separate nodes for images.