NFAIS 2013 opened with a glorious bit of philosophizing about knowledge in a keynote from David Weinberger, author of Everything is Miscellaneous and Too Big To Know.
Some people might not care about “what knowledge is,” but surely anyone in or around the field of education gets the importance. Teensy example – last year I encountered various theories of economic development which my Penn professor at the Graduate School of Education (the very smart Dr. Ghaffar Kucher) related to epistemology, to different paradigms for knowledge. So I was poised to love Weinberger’s keynote talk, and I did.
Here’s my selective interpretation of Weinberger’s remarks.
The first startling thing is that, Marshall McLuhan-like, what we think knowledge is gets mixed up with the medium by which knowledge is delivered. When knowledge came in physical containers (books, journals, etc.) you couldn’t physically access it all. So it was important to condense it, to get value out of it by making it smaller.
We thought of knowledge as being part of a pyramid. By condensing the raw data into information, people could sift through that and create knowledge. You don’t have wisdom by just plowing through data, you distill at each layer. We managed to get to wisdom by “reducing the amount of stuff we have to deal with.” We filtered stuff out.
Another thing the pyramid shows us about our up-til-now view of knowledge — knowledge has traditionally had the property of being that which is settled. If we’re still arguing about something, we say we don’t know. We tend to think of knowledge as something we agree on; in a true state of knowledge there isn’t any more reasonable disagreement.
Theorists who first used the knowledge pyramid explained each layer as built on the other –information is structured data, and knowledge is information-you-can-use. The model is somewhat limiting in that knowledge, in Western culture going way back, has always been thought of as something more than just a way to get things done. Knowledge was being truly human, understanding our place in the universe etc.
Other things from the world of knowledge-presented-in-tangible-media. In the physical world you can only have one organizational scheme – pick one. So, the taxonomy (where we place an organism in the tree of life, for example) becomes the truth almost. We forget that’s it’s a representation, a model, and instead we take it for reality.
And that leads us to seeing knowledge as a series of stopping points – we get the answer, boom, move on.
Knowledge in the West shares the properties of its medium (print).
Enter the Internet. Our new medium for disseminating knowledge is a NETWORK which gives knowledge the properties of networks. Networks don’t “end.” They are infinitely extensible (I think).
Knowledge starts to look more like a network. (I can see our culture in transition here–many disagreements seem to me based on what counts as knowledge.)
Now compare properties of knowledge from the print world to knowledge in the networked world. Something sparks a response throughout the network – it’s all relevant and it’s all linked. Guess what – no one agrees! (cf the print world, where getting published meant a lot of settled agreement about what you have to say before people would undertake to print your work). So, if I understand Weinberger right, we begin to have a deeper knowledge because when everyone is saying something (slightly or even dramatically) different, we benefit from a whole lot of available multiple perspectives trained on any particular issue or problem. And we’re not saying everyone is equal — we’re not saying crazy or unfounded opinions help. We’re saying these multiple perspectives come from thousands of people with expertise and reasons to weigh in. We would actually learn less or understand less if all these smart people were saying the same thing. The value comes from multiple perspectives ad infinitum rather than from filtering for less.
It’s almost paradoxical sounding to hear Weinberger say “Disagreement is how you scale knowledge – it’s how knowledge can get really big.”
He then showed us a diagram with a picture of an ordinary robin in the center, surrounded by labels in all directions to show that when the world at large talks about the robin, it can have many “meanings,” including a robin, a songbird, a symbol of Spring, a disease vector, a work of art, etc etc. This helps me understand that knowledge is not a finite amount of statements about a robin or one placement of the robin in some ontological scheme. The knowledge is the infinite number of semantic relationships a robin can be part of. If the knowledge I’m seeking is the kinds of birds I can see in Medford, Mass. in mid-March, then somewhere there may be a data set with geo-tagged, dated photos of robins. And that’s a pretty mundane example — a robin could be included in climate change knowledge, a study of birds as symbols in literature, or even the “livability” of a city.
If we can cope with the messiness of a network as compared to a pyramid, knowledge can get really, really big. Which I take to mean, I could conceivably “know” a lot more because with the right tools I can produce knowledge precisely because I have access to sooooo much information.
Then Weinberger remarked on something that has troubled me ever since I started to work in libraries. Curation. In the world of tangible media (where space is an issue) but also in the world of licensed electronic information, you can’t have everything so what do you acquire, manage, keep? Anybody’s “principles” (e.g., buy what people want) are in conflict with someone else’s (e.g., buy what they’ll need, but don’t know it yet). It’s so obvious, there is no right answer. (In my public library days, it just depended on who was boss.) What a relief to have someone talk about how problematic curation is. You simply cannot predict what will be of interest or need, anymore than you can limit the ways a robin might have meaning.
I do think some libraries got over this, realizing that what their constituents could “access” was just as important as what the library might own. Still, we spend a lot of time selecting stuff for individual libraries, and as Weinberger pointed out in his earlier book, it’s cheaper to just have everything and filter on the way out. Selecting, or curating, is a very expensive use of staff time. Since libraries can’t buy everything (even electronic stuff costs money) they are forced into selecting. But the illogic is increasingly clear, as libraries struggle with what they should stop doing in order to provide impactful services for users with new needs, and perhaps this will be one of the drivers to radical new business models.
What Weinberger calls filtering on the way out I think I would call discovery — it’s sort of the same thing when you can go to a place where “everything” is, and then drill down to what you want. You don’t filter by putting only a few things into the bucket; you filter by going into the Bucket of Everything and using discovery tools to bring what you want to the surface.
So, in this new world, we’re no longer going to get value by filtering on the way in, by weeding through and letting only some information become knowledge (by selecting, by only publishing some info, by only providing access to a subset, etc).
Instead, thinking about knowledge by thinking about the properties of networks, Weinberger talked about 3 ways to squeeze more value from information.
Iterate forward into knowledge (rather than seeing it as something settled that you obtain.) His example is the stackoverflow site. The likelihood that you are the first person to have a problem is minimal to vanishing. People can help you iterate forward. Iteration at webscale is an efficient way to produce new knowledge when information is superabundant.
Weinberger alluded to some unfolding cultural effects. We’re still not used to what happens when millions of people can be brought to work together. According to traditional ideas of knowledge production, Wikipedia shouldn’t work, but it does.
Platforms continually increase value. I can’t remember what example Weinberger used but the one I have heard is facebook. Facebook didn’t succeed by putting content out on the web and attracting people to it; it succeeded by being a platform on which people could share content, and by designing the platform so that people’s ordinary interactions could make the content more and more valuable. If my circle of friends are all recommending a particular book or a data visualization tool, no centralized source is bringing that to my attention, it is the result of many individual behaviors.
Weinberger sees the library as a platform. The top layer is where the value is created (people using stuff and connecting with other people). As a knowledge network where people discover, build, share. Everything they are doing in this layer should be feeding back in to your system to keep adding value. This would be a terrific discussion topic back at the ranch! (It’s the sort of thing that makes me want to talk about why “tagging” in academic libraries never seemed to catch on.)
Again, the top layer is users doing whatever they feel like and adding value by their behaviors. The middle layer is tools, services that allow users to do “whatever.” The bottom layer is the data, metadata.
This led to another interesting philosophical point — the only difference between metadata and data in the digital world is functional. Anything that is data can be metadata for something else. The example is “Melville, Moby Dick, Call me Ishmael.” Any one of those terms could be the metadata for any of the others, depending on what a user is trying to do.
Similarly to how curation is problematic, Weinberger commented that plans don’t scale. People who try to anticipate what will happen are going to be wrong. The pain of being wrong becomes greater as we invest more in our plans. Seeing the library as a platform allows the development of value without recourse to planning which will be mostly wrong.
#3. Linked data
What is happening to knowledge (it’s getting massive and interlinked ) is also happening to data.
Weinberger applauded some experiments with “filtering on the way out” such as Data.gov - which is full of big, messy stuff. Just chuck it all in there, and let people develop tools and services around it.
Because data can be meaningfully linked to and used, we don’t need knowledge to be a settled, agreed on thing. Knowledge is now useful arguments between people…at scale. The usefulness comes out of linked data. We can have evidence at our fingertips. At this point it almost feels better to say that we’re always going to be in the process of knowing rather than to talk of knowledge as if it’s some disembodied thing. (Although complicated ways of talking like that usually make me impatient….) But it’s really a point to ponder. What can possibly be seen as a “right answer” when complexity theory demonstrates how mind-blowingly complex problems are, with googobs of interrelated and interacting sub-parts, and we need to think about optimizing in a constantly changing situation rather than getting to one right answer.
So, if we need to have computational ways of making all that data useful to us, we will need semantic search.
Weinberger closed with several points.
These things don’t scale:
The web is a good metaphor. We built this without having control (a project plan, named managers, etc). There’s no agreement on the web, there’s no real order, no one is in control.
And this Syllogism:
- Interests are unpredictable
- Value arises from interest
- Linked openness enables value to scale
By the way, Weinberger says he works in a basement (at the Law Library at Harvard). Now that seems too bad. Let’s hope it’s a nice office down there.