IA Summit: David Weinberger Plenary

by March 25, 2006

David Weinberger’s opening plenary at IA Summit 2006 walked us through his thoughts on today’s shifting uses and storage of knowledge.

For twenty five hundred years, we’ve been organizing ideas and knowledge using the same principles use to organize physical objects. As a result, most of our information organization methods and principles are artifacts of the limitations of publishing on paper. Now that we have digitized everything, can we use new principles to organize information? If so, what are these principles?

Something is going on with knowledge

There are important changes in our understanding of knowledge afoot. Most of us accept the transition of Date to Information to Knowledge to Wisdom. But this implies a false continuity. In fact, this sequence probably is backwards.

  • In our efforts to fit information in our databases (computers), we’ve stripped out all of the context.
  • Information has become a refinement of data but without the context surrounding that information –how can we get to knowledge?
  • Going back in time, knowledge was thought to develop differently: Impressions to Resemblance/Contiguity/Cause to Reason. In our current definition of knowledge what has happened to Experience, and Wisdom?
  • We have come to think everything is information. For example, DNA is always depicted as information but it is really a clump of molecules.

Seven properties of knowledge

Traditionally knowledge has been defined with seven properties.

  • Knowledge has to have a knower in order to be known
  • One & the Same: There is only one knowledge -it is same for everyone
  • Simple: Knowledge is simpler than the world. World is very confusing in appearance, knowledge helps us make sense of it.
  • Filtered: Most things aren't knowledge; need experts to find that refined view
  • Impersonal: it doesn't matter how says it. People are abstracted from knowledge.
  • Outlast Us: Knowledge is bigger than we are
  • Orderly: Knowledge has structure, it provides context, leaves things explicit

For years, we preserved knowledge in objects with these attributes so knowledge (and way of representing it) has been guided by medium: most notably paper.

Other ways of understanding knowledge

We don’t operate from a rational set of definitions. For example, a robin is a much better example of a bird than a penguin. As a result, a robin is a prototype for bird. To define something we point to a prototype of it and cluster other objects around it based on similarities or difference to the prototype. We don’t rely on a rational definition (feathered biped). For example, Flickr & del.icio.us are prototypes of tagging –they can be used to define it.

  • First Order = organizing actual things (stacks of books)
  • Second Order = separate metadata and organize them (card catalog). This often results in a huge reduction of knowledge.
  • Third Order = what can you do easily digitally that in the real world is hard. Here messiness is a virtue (the messier the better). When everything is digital, the difference between metadata & data blurs.
  • Through faceted browsing, people can construct their own hierarchy (tree) to suit their own interest.


Knowledge has primarily been built up by excluding what we don’t need to know. As we move from a tree (a traditional hierarchy) to a pile of leaves (tags, hierarchies, blogs, front pages, and more), excluding information has negative value. We can afford to include everything and people can filter information by any system they found useful.


Why do we care about authority?

  • It has utility –helps us understand a world that is way too big.
  • Also provides social standing, institutional power, control conversations, personal virtue, fulfillment of species destiny, money
  • Britannica has credibility and as a result, authority
  • Why do we believe what we find on Wikipedia : if it is a major topic, we are more likely to know something about it; lots of edits; lots of discussion; notes that indicate it needs to get cleaned up. It tells us where it is week.
  • Major sources of information opt for authority vs. truth.
  • Publicly negotiated knowledge: the greatest expert that cannot negotiate in public discussion will render his information irrelevant if he/she can’t publicly negotiate.
  • We know have knowledge without a knower –the convergence of agreement on Wikipedia
  • In the future, does Wikipedia represent a commoditized set of knowledge we all agree on? Does this lead to: separate knowledge? Base-lined knowledge from which controversy emerges? reclassification of knowledge alliances? fragmentation or re-alliance?
  • Tagging is fulfillment of postmodern philosophies. Authors are not the best judges of what their works are about


  • Books are unlinked knowledge sources
  • Digital info is highly linked, which is essentially to usefulness.
  • To understand a hammer you need context (nails, wood, carpentry, construction, etc.); you need to have meaning (messy set of potential relationships between things)
  • Tags will never rebuild that context: we simply cannot make everything explicit
  • There is value is in the implicit not just the explicit.
  • If the measure of tagging is to rebuild the context of the world, then it will never succeed. The bigger the set of data, the better the clusters we can form but it is still very hard to rebuild hierarchical systems.
  • Good enough is all we need. We won't get to hierarchy but we are adding valuable data.

We will continue building hierarchies, playlists, tags, blogs, links and more that pull together leaves at a pace we can't imagine. We’ve spent twenty-five hundred years building knowledge. We are now focusing on building meaning.