Taxonomy in a Digital World - Part 4

Concluding my notes on Everything is Miscellaneous...A good read on the whole.  Does a good job of summarizing the information management trends on the web today.  If you are an information architect or consultant working in the field of knowledge management then I would say you should have a copy on your bookshelf.

Chapter 8 - What Nothing Says

The limitations of software force us to often make explicit binary choices where in reality complex implicit social interactions exist. (I am reminded of Scoble's recent dissection of Facebook friending as an example of this). Weinberger talks about CRM systems - good salespeople know more about their clients implicitly than they can say explicitly. He also talks about the fallibilities of internet dating and social networking services and the limitations of the boxes you have to tick and fill-in.

So the problem here is this: Human social interactions don't fit neatly into tick boxes.

  • How do we capture and catalog the implicit stuff?
  • How do we categorize information when the boundaries are blurry?

Weinberger points to and tagging as an example of how giving users free reign over categorisation enables (at least some of the) implicit relationships to emerge. He also discusses the implicit data trail people leave behind as they search and browse and interestingly points out that our idea of privacy is in a transitional mode right now as we figure out what we can and can't do with this data (as Google found out recently).

One problem with tagging though is that as a young technology we don't yet know if people will tag for themselves to refind information, or to help others find information.

Folksonomy: a taxonomy that emerges from how people tag items.

Discusses (possible) limitations of folksonomies including their inability to handle equivalence, synonyms, hierarchies etc. Some argue that this will cause them to fail at significant scale. Weinberger argues that improvements in technology will allow computers to be able to get better by inferring relationships, synonyms from how people tag, and by learning more about us and our preferences. Flickr homepage has no editors - implicit metadata from tags (how many, who from and "other stuff") is used to select photos for home page. Implication? Tags may become more useful, meaningful and relevant the more there are.

Databases and rigid taxonomies strip out the open-ended context from information. Tagging, links, playlists and the rich interconectedness of the web (connecting the leaves in human ways) is adding this context back. We can gain insight and meaning from this connectedness.

Chapter 9 - Messiness is a Virtue

Most if not all categories do not have clear cut boundaries. We learn to categorize things based on prototypes: A robin is a good example of a bird, an ostrich or penguin is a bad example. The hard lines of categories come later when we are forced to draw them. A business that forces its products or employees into a predefined set of categories is performing an unnatural act.

The Semantic Web - adding structure and classification to the mess of the web. Largely so machines can work better with the information therein.

Weinberger discusses the Semantic web from a sceptical point of view. He likes some simple ideas - smart links - links that contain metadata. He dislikes the vision of rationalising, driving out the ambiguity, minimize the miscellaneous.

"This implies a massive agreement about exactly how the metadata should be expressed. Everything works perfectly if everyone just agrees on the terms and the rules to follow...Turn language into a machine and our machines will work wonders...A seamless whole that drives out ambiguity would also drive out the richness of implicit meanings."

Resource Description Framework (RDF) - a description language for defining relationships between two items. This can be used to build an ontology to describe a particular subject area. Problems with this are similar to Dewey system. People want to define huge all encompassing classification systems upfront. Knowledge is ever-changing.

Something can be 73% in a category. The edges are fuzzy. The folksonomies that are emerging bottom-up are characterized by ambiguity, multiple classification, and sort-of, kind-of relationships.

It's a sort-of, kind-of world. Things don't fit neatly into categories. Accept it.

Chapter 10 - The Work of Knowledge

This final chapter concludes with a summary of the changes happening on the web today, a look to the future and a call to share information freely.

Information has more value when it is set free. It provides an opportunity for new information based businesses to emerge. New novel uses for that data. Meta businesses - aggregating and cutting data in different ways. Travel web sites displaced the middlemen that added no value. Then web sites emerged that compared prices across travel web sites. Now the latest breed display charts of how fares change over time. And add information such as average travel delays at specific airport terminals.

As one meta business succeeds others come along to take it up a level.

  • It's not who you report to or who reports to you, it is how messily you are connected.
  • It's not what you know, or who you know. It's how much knowledge you give away.
  • Hoarding knowledge diminishes your power because it diminishes your presence.

Technorati tags: Everything Is Miscellaneous, Davd Weinberger, Taxonomy, Folksonomy, Semantic Web