These are my notes from a talk that Larry Prusak gave at my invitation at the Kellogg School back when I was on the faculty there. (For posterity’s sake, here is a link to the original blog post, although that version is suffering from bitrot. The Magic Number 300: Knowledge and Community)
His reference to the “magic number 300” is what I now recognize as Dunbar’s Number, which most people seem to estimate at closer to 150. The point remains largely the same.
I persuaded Larry Prusak to swing through Evanston last week (February 2002) and talk about knowledge management to students here at Kellogg. Over the course of 90 minutes he shared his perspective on knowledge as part of organizational life.
On a side note, Larry has abandoned Powerpoint. He’s always favored storytelling anyway. It does make the audience have to work harder, but that’s a good thing.
Unit of Analysis
Larry started with an old research design question – what the right unit of analysis. His answer? The group. Not the individual, not the enterprise. While he argued that there are important ideas tied up in the terms “network” and “community of practice,” he’s also concerned that the terms are well down the devolutionary path toward buzzword. “Group” is a nice simple word and you then have to listen to the actual points he makes instead of tuning out.
His research reinforces others in setting 300 people as a “magic number” — about the limit on the number of people you can know by name. He added one tidbit I hadn’t thought of before; that the typical military working unit has also averaged about 300 through history (worth tracking down the evidence for that later).
Given In Good Company, it wasn’t surprising that Larry tied the issue of group size and knowledge to social capital. He sees knowledge largely in terms of trust and of the conditions where reciprocity works. His evidence says that incentives are useless in knowledge management settings. You have to look to the level where people share without explicit incentives.
As Larry put it, “people want to find each other and want to talk.” You can force this into an economic framework, but that 300 number hints that we’re dealing with something pre-economic. “Gift culture” is the technical term for what Larry has tagged as one of the driving forces of knowledge creation and uses in organizations. What’s interesting about that is that it creates a point of tangency with open source software development as a social phenomenon. Eric Raymond’s writings on The Hacker Milieu as Gift Culture is a good entry point. I’ve got an old essay on linking KM and Open Source ideas that I’ll dust off and post as a story.
Here’s Larry’s densely packed definition of knowledge: “embodied, tacit, pattern recognition.” Whew!
Here we tie in directly to the community of practice literature and to the learning processes that take someone from novice to expert. The late Herb Simon estimated that it takes 10-12 years to become an expert in a field. Experts reveal themselves in their ability to see the patterns in a situation that are invisible to a novice. I had a partner who was an expert on call centers. Dave could walk into a call center and tell you in a matter of minutes where the problems were. If you grilled him for a day or so, you might be able to figure out about half of what went into his judgments. This was why expert systems never succeeded.
If you believe that capturing knowledge is a vain quest, then you need to direct resources toward community. That’s certainly where Larry comes down. While I believe that community is the more important and more malleable element, I’m not quite so quick to dismiss investments in capturing and organizing knowledge.
The keys to encouraging community relate more to providing time an space than providing money. That’s actually trickier, because it’s easier to throw money at problems. Chapter 4 of In Good Company is the place to go for more details.
Knowledge and Learning
Community is a means to the end of learning how to put knowledge into practice. To IBM’s great disappointment, Larry isn’t keen on document management as a path to better knowledge and he’s pretty skeptical about the value of technology tools to nurture communities. I think he finds much of the money poured into formal knowledge management systems wasted. As he put it “asking about the ROI on knowledge management is like asking about the ROI on babies.”
While Larry is no Luddite, some of his technological skepticism is generational. He’s squarely in the middle of Negroponte’s digital homeless. His skepticism is a useful counterbalance to the usual utopian promises of technology advocates. But I take his input more as design constraints than a meaningful argument to avoid technology.
Larry closed with a really great question to use when poking around in an organization trying to get a sense of its attitudes toward knowledge and learning — “can you make a mistake around here?”