I was reading "Emergence: The Connected Lives of Ants, Brains, Cities, and Software" by Steven Johnson on the plane yesterday flying home from Dallas. In the book is some discussion regarding an adaptive web. A web with emergent properties. A self-organizing web. Interesting ideas that inspired several fragmented thoughts of my own which I recorded in my Treo. I don't have the time right now to dig into these ideas or even to form a cohesive article around them but I wanted to record them and to maybe incite some discussion around them.
- pattern recognition: from tables of data to a semantic representation of that data (I think Dabble DB already does this); inferring site type (blog, corporate site, news, etc) from page structure
- webpages that when users view them activate machine level links - i.e. you go to a webpage and it 'pings' all the other pages it links to; the pages that are pinged, if pinged by multiple pages for the same user, may somehow make the user more aware of them
- a way of recommending new bookmarks: analyze all the inbound links for a group of bookmarked pages, if any page links to more than a couple of a user's bookmarks then other links on that page may be of interest
- ant trails... but are most people coming to a webpage looking for the same thing? (depends on the type of page) even so, is a link the most common exit point or is the user hitting the back button and selecting a new link on the previous page more common?>
- tracking link clicks - should be able to accomplish with a javascript function for the onclick() handler of <a href> tags; the script could make an ajax call to the server before the browser follows the link to the new page (i recall seeing sites that show the relative popularity of outbound links - they probably accomplished that like this)
- tagged bookmarks - people with highest page count for a given tag may be an authority for that tag
- the most popular link isn't always the most appropriate recommendation - people visit pages for different reasons; would be better to take into consideration their referrer, users that arrive the same way may leave the same way so recommend exit links giving extra weight to those that users who got there the same way followed
- visualizing site traffic - exit links are as important as entry links, especially for hub style sites where people click a link and then hit the back button to click another (or simply open several links in new windows/tabs
- is a network of neurons at all interesting if you can't do an MRI on them?
- page structure is also important in traffic analysis - if users predominantly click links near the top of a page but never near the bottom then surely that suggests something
- 'link blogs' have become popular recently -- I started (and since neglected) one here -- where entries are too small to be considered articles; they may not even contain actual links. What if the entries were networked? Like an online mindmap? linking individual mindmaps into a global mindmap... streams of consciousness metaphor where streams join into rivers feed into and ded by lakes and oceans etc...
Lots to think about. Lots I haven't even touched on like two-way links, semantic content, etc.
