The web became more collaborative with wiki's introduction of the Internal Link. Link names were drawn from a shared space backed up at the time by shared storage. Federated wiki's internal links preserves the shared space but not the shared storage making it the ideal link for the distributed collaborative web.
The creative commons' attribution agreement combined with html5's cross-origin resource sharing provide the missing mechanisms that make the distributed collaborative link work.
In short, we know where a page came from and we know how to find more.
When we mashup a page that has been well circulated we can still expect its links to work even if the cited pages have not been similarly copied. A click on a collaborative link works backwards through current and previous authors until the desired page is found.
When communities meet and share a page or two they have opened a portal between each web that future readers will follow within what appears a unified site.
We now describe collaborative links in the abstract and as implemented in federated wiki.
As pages are copied (forked) they accumulate attributions sufficient to locate originals in their original linking context.
Federated wiki uses individually owned domain names as attribution expecting author identification and related pages to be served according to well established conventions.
Link click handling consults the accumulated attributions in a most-recent first priority search for the desired page.
Community awareness is provided by acknowledging older and newer versions of pages as their availability becomes known.
Federated wiki fetches automatically maintained sitemaps for each attribution it encounters. This informs available versions as well as 404, search and recent changes.
The reader controls the collection of sites dynamically merged into one based on their own needs and interests of the moment.
Federated wiki maintains a distinct neighborhood for each browser tab. Features like drag and drop between tabs and collected rosters of likely participants provide quick and controlled neighborhood expansion while browsing.
Federated wiki creates new sites on first reference. We've found this to be liberating in that each site provides a unique vocabulary for the writing that takes place there.
We fork pages that have utility in multiple contexts. This frees each site to be as independent and connected as convenient. We've adopted an abrupt, to the point, easily refactored writing style that regains context by surrounding pages.
Emergent site structures tend to be topical, chronological or reflective. Pages are written quickly and then updated occasionally. A page copied to a new environment will be adjusted to fit.
We designed for desktop authoring while viewing three to five wiki pages at once, tablet reading while viewing two or three page and smartphone reference retrieval showing only one or two. We have not yet shown this to be the prefered use of touch devices.
Our design favors non-conversational collaborations suitable for more thoughtful writing. Authors are not guaranteed to know what has become of their words.
Random identifiers are created for each paragraph. These survive through editing and copying. Pages sharing a paragraph of common origin remain implicitly and bi-directionally linking by id. Search engines then become the external link-store that makes bi-directional links possible.
We have been crawling the visible federation since October of last year. We found 776 sites then which has grown to 1067 now. We see 300,000 uniquely identified paragraphs, a number which is growing by 3000 a month. We settle for slow but steady and lasting growth by encouraging participants to launch their own servers.
TEDx Portland 2012. Describes improved link.
Writing with Strangers 2015. Emotional reflection.
Natural Genetic Infrastructure 1996. Call to action.
Decentralized Web Summit 2016. Inspired this page.