I’m probably not the first to note it but it seems new tools for organizing knowledge are moving from directories-based (i.e. trees) into flat tag-based. We know that ubercool applications such as del.icio.us, citeULike, flickr use tags, but i realized today that even gmail does not allow you to create folders (and (sub)*forders) for your email messages but only to tag them. In this way emails remain in a single big pool but you can have different views over them based on the different tags you used.
I’m wondering if (and how) filesystems can move in a similar direction, or at least the “explorer” visual interface of a filesystem. VennFS (snapshot) seems an interesting direction even if I’m not sure the representation metaphor is easy to grasp. It is selfevident that the fact many people are used to directories-based structures does not mean this is the best way to organize knowledge.
A quick search resulted in 2 keywords related to tag-based system you may want to analyze more: folksonomy and ethnoclassification.
Pros of tags-based tools: an object can fall under more than one categories, you don’t have to think once forever your categorization structure and then be stucked with it but tags support evolution.
Cons of tags-based tools: unless the tool incentives re-use of tags, you can easily end up creating too many tags and forgetting about them, resulting in inability to find the information you previously categorized.
Ramblings by Paolo on Web2.0, Wikipedia, Social Networking, Trust, Reputation, …
Have a look at
http://www.namesys.com/whitepaper.html
for some interesting ideas on namespaces and filesystems.
BTW, I have contributed to the developement and deployment of some knowledge management / CMS applications, and I have found that many non-technical users find unnatural and difficult to work with hierarchical structures as a way to organize content.
Francesco
I actually think that Gmail’s feature of organizing mails with labels rather than (sub)directories is one of themost interesting things of this service.
My personal point is as following: hierarchies and tags/labels may be combined and used as relaxed topologies (I use kind of fuzzy topologies called Indra’s nets). I.e. multiroot, shortcuts, multiple inclusion, etc. I’m currently working on that. If anybody is interested, my e-mail is given here.
My personal point is as following: hierarchies and tags/labels may be combined and used as relaxed topologies (I use kind of fuzzy topologies called Indra’s nets). I.e. multiroot, shortcuts, multiple inclusion, etc. I’m currently working on that. If anybody is interested, my e-mail is given here.
Hi!
My personal point is that hierarchies/taxonomies and tags/labels might be combined in relaxed fuzzy topologies (I call it Indra’s nets). I.e. multiple roots, shortcuts, multiple “parents”, etc. I’m currently working on that.
Jeff Raskin in “The Human Interface” argues that we should get rid of file names and directories, use file content as the the file name, and searching and colocation of files and their descriptions (tags) instead of directories.