Switching to del.icio.us
A basic collection of del.icio.us related resources including its REST API, XML-RPC blog posting mechanism, Mozilla-to-del.icio.us and del.icio.us-to-Mozilla integrations, XHTML conversion and more:
So, why no longer hand-rolled QuickLinks? Why don't do all the publishing on your own server? Let's have a look at the drawbacks first:
- No more hand compiled collections by topic with multiple links as del.icio.us allows just one link per item.
- No attribution as there is no "via"-field. No attribution = no back link = no good.
- No means to arrange similar tags into groups.
And now at its merits:
- Power through simplicity: super-simple, super-robust, easier than keeping bookmarks / atomized content in a blogging system could ever be
- Dead simple posting via a RESTful API and dead simple data retrieval via the API or RSS feeds (even though the API works on your own bookmarks only)
- Keyword system ("drop the taxonomies")
- Collaborative effort: the group-mind always knows more than the I-mind
If it is so powerful, why limit it to posting QuickLinks? No need to do so. Instead, let's think of using it for all four groups of bookmarks I have:
- QuickLinks, carefully crafted links collections for your enjoyment
- URLs inbox of unread, but potentially interesting items. First attach "toread" or "totryout" tags, then after checking drop the item or use it further
- URLs collected for a specific blog entry, wiki page, research theme or project
- URLs collected for reference use
So let's see how we can handle the drawbacks:
- QuickLinks: attach a "quicklinks" tag, post into single blog entries at intervals, voila
- Attribution: use special "via:"-tags as a "via"-field substitute, leave away the http:// part of the domain name, have the via-links converted into backlinks when retrieving links via the API and publishing them as XHTML (on the blog). Example: If I bookmark something on Papa Scott's weblog, "via:papascott.de" would be the via-tag.
- Cloud forming: no instant solution, room for development
How will the weblog integration look like?
Both retrieving XML via the API and transforming it with XSLT or the Magpie RSS Parser and the XML-RPC route look promising. As an ad hoc solution, I could use Feedburner to merge the blog and bookmark RSS feeds into a single feed.
Extending the API use beyond bookmarking, building visualization tools, building keyword handling, editing and grouping tools, building more nifty interfaces might be further steps. So are gluing together thematically related blog entries, wiki pages and bookmark collections (finally a solution to that!) or applying the tags architecture to entirely different applications like a photo collection (hint hint) or this weblog (there is already a hack that replaces Movable Type's category system with del.icio.us tags). Generally, tagging discipline will be an important subject as it is a cornerstone for building powerful queries.