PeerToPeerWiki

MeatballWiki | RecentChanges | Random Page | Indices | Categories

Let's think of a PeerToPeer wiki in a new way. What if instead of each peer having a server where they "own" a few WikiPages, we really think of it as an implementation of AllInOneWiki where there is only one wiki for the planet?

The simple text of each WikiPage is stored and a custom WikiBrowser does the work of converting to a suitable display. (The custom browser is probably the biggest leap of convention here but read the rest before you kill me).

Each peer has a Public and Private set of WikiPages. Private is of course for your own private notes and ideas. Public contains a copy of any pages you've recently viewed or changed. The P2P tool allows you to easily move a page between the folders.

The scary thing is that there could be hundreds of people editing and saving distinct versions of the same WikiPage at the same time. This sounds like horrible chaos (and it is) but we gain a few cool things by allowing and encouraging this:

There is always a backup if a page was accidentally or purposely deleted or screwed up.
Different thought lines will develop that otherwise might not have.
Definitely no worry about somebody's changes being accidentally overwritten.

How do we handle all those versions? First, whenever we open a WikiPage, we pull it from whichever peer we pulled it from last time. Then, from there you always have the choice to look at other versions of the same page. Only a few distinct choices will be offered:

Most Popular: more people have chosen to look at this version more than any other.
Smallest Change: this encourages spelling and grammar improvements as well as adding a simple link to related information.
Most Respected: more people have consistently chosen changes from this peer.
I'm sure there are more valid options...

Most of the time if you're just browsing around, you can leave the default choice. If it is a topic you participate in, you will have to put in the extra effort to see what choices are available. You are also going to need some way to compare files so you can see exactly what was changed to help you better decide which version to use.

I see a few interesting affects of all this:

Specialized interests will have greater consensus and less worry about versions simply because there are less people making changes.
Popular pages will always have multiple versions out there. There will be a continual process of branching and either manual reconnection or splitting into distinct sub-pages.
Best ideas win -- a silly or prank change will naturally fall out.
Since this an AllInOneWiki that doesn't attempt to have any kind of categorization or folders, it would probably lead to very long wiki names.

-- Robert

CategoryInterWiki?.

A very exciting idea indeed! I think it works a little with the idea I am developing over at TheyCallUsTroubleMakers? - a self organizing wiki structure. - MarkDilley

I like your ideas about blogs, personal wikis, and public wikis all being interchangable. I think that would work very nicely with this. Your private wiki pages will link to public pages. If you link back the other way, the page will be automatically copied to your public page folder.

(warning non techie/newbie thought) Hey Folks, I am thinking of the wiki in not so technical terms, because I don't have those skills yet. I have been thinking of PeerToPeerWiki, what are peoples thoughts on that/resources. Best, MarkDilley

What exactly do you mean by PeerToPeerWiki?

My thoughts are running along the lines of each individual having a wiki on their machine, that they edit and decide when it goes into the shared file of a peer to peer system. you control what is on your wiki, if someone changes it, you can change it back, you have access to, for lack of a better term, censor content, words and IP addresses. You do searches for anything you want on the PeerToPeerWiki and get your results. . .

p2p wiki - gps - radius wikis, sharing info at a local level, voting systems of interest - would this create a hatsfields vs mccoys?

Wiki:PeerToPeer [1] [2]

: This sounds interesting: '...each individual having a wiki on their machine, that they edit and decide when it goes into the shared file of a peer to peer system....', it might be similar to what i am thinking of while writing my comments on Wiki:AllInOneWiki. Everybody having a wiki and the connection of all these (by p2p protocol or perhaps another one) leads to the vision of OneBigWiki IMO, it's just a question of view (interface). - Hi Mark, your comments always inspire me! - Thanks. :) I think i should refactor all my comments on the related pages to this topic the next days. - cu, FlorianKonnertz

[Mesh] system of Peer to Peer technology strikes me as being technically similar to this idea. (strikes me as a non techy) --MarkDilley

: This is very interesting to me, could a wiki server be a plugin also? THE JABBER DIFFERENCE Jabber is an XML-based protocol designed to enable real-time IP messaging services. Jabber enables anyone - yes even you - to run their own online chat server. And that's just the beginning! Even now work is underway to expand the Jabber protocol to include additional features like whiteboards, chat rooms, and more - all running on the existing Jabber infrastructure. As many of these new features are introduced, you will be able to use them in conjunction with your DreamHost?-provided Jabber server.

: The JABBER DIFFERENCE is very interesting to me, could a wiki server be a plugin also?

: Wiki plugin for Slashdot -http://www.onlamp.com/pub/a/onlamp/2002/01/17/slash_plugin.html?page=1

re: [CsWiki] idea peer to peer protocal, a call to task; I don't think a p2p discovery protocol is needed for Wikis; the users do that pretty well. Even if there were so many wikis that we couldn't keep up manually, why would the software need to know about them? -- BayleShanks

: So, I am sitting at my computer and I have access to all my friends and (other important people) wikis (through an rss reader?), because they have told me they are online with a wiki. But what if I want to search all the wikis out there who are writing on topic XYZ? Or, what if I want to write to a common XYZ page? Because there are many redundant wiki pages. How to use wikis, is a common redundant theme. So the idea, for me, is that I search a p2p wiki system for pages that I want to read/write too. Yes, users do it well, but that can't scale, imho. -- MarkDilley

Hmm.. I was going to reply to your message on my page that while InterWiki tools like MetaWiki are valuable, that the list of available wikis used by these tools would have no need for dynamic discovery, that users would be able to add new entries to things like InterMap as fast as needed. However, I thought of a counterexample to my own argument; I'm interested in WebOfTrustModeration, but the InterWiki analog is almost exactly what you are talking about; messages would be dynamically cross-posted to certain "nearby" wikis. So I take it back; a peer to peer discovery protocol would be useful. In addition, I guess that, as you say, there might be times when you would want the equivalent of Google for wikis; a MetaWiki that searched not only your neighborhood but rather, searched all wikis. -- BayleShanks

I am really surprised to find this page. I implemented the [Bouillon project] which may be understood as P2P wiki with reputations. You may find a user manual at http://oc-co.org:8000/page/Bouillon_Manual (guest login 'foaf', password 'foaf', Firefox 1.5 is required). Bouillon employs XMPP (Jabber) for message exchange. Also, existing Jabber contacts are used as a social network for message propagation. 'A piece' is the basic editing unit in the Bouillon network. A page is a tree of pieces. Piece is an XML fragment with metadata (version id, parent piece, author, reputation distance to the author, etc). The current reputation scheme is pretty trivial: percentage of previous successful recommendations defines reputation of a contact. `Reputation distance' is multiplied reputations of every link a message passed. The farther is the nearest recommender of a piece the smaller and greyer font of the piece. The Bouillon addressing scheme is flat in the wiki way. Content doesn't depend on the server; it depends on your social vicinity only.

Bouillon software itself is

an external component for a Jabber server to interface with other Bouillon servers;
also a web server to interface with its users.

And so on and so forth, see the site. -- VictorGrishchenko

PeerToPeerWiki

Discussion