Doug Cornelius .com

Dave Snowden, of Cognitive Edge, laid out the most thought-provoking session I have heard on knowledge management. (My head is still sore from trying to assimilate his presentation. ) He is posting his podcast of the session and the slides from his presentation.

Dave espoused his theory on naturalizing sense-making. We should focus on how we make sense of the world so we can act in it. Knowledge management should be about decision-making and innovation.

He takes the position that knowledge management lost the battle and is becoming a subset of IT. He is seeing a revival of knowledge management because of the Web 2.0/Enterprise 2.0 technologies. He thinks there will be much more success working on fragmented information rather than structured information. People like a mess. Put too much structure around information and people lose their place.

I have to agree with him. Whenever I hear someone talking about a KM system, I shudder. Knowledge management needs to be built into and latch onto a person’s regular workflow. You can ask them to make some changes to what they do so that information is better captured. But you better offer them lots of concrete, easy to show benefits for changing their workflow.

Dave moved onto systems theory, talking about ordered systems, chaotic systems and complex systems. With an ordered system, the system constrains agent behavior so with a particular input, you can expect the output. With a chaotic system there is no agent constraints resulting in turbulent and unstable process. But, with the use of statistics and probabilities, it can give you a predicted outcome. In a complex system, the agent constrains the system and the system constrains the agent. These two-way constraints make it harder to handle than the chaotic system. It is also highly sensitive to starting conditions and cannot be broken into simpler subsystems.

He used planning a kid’s birthday party as a metaphor for various management theories and how they relate to these three different kinds of systems. First being uncontrolled management, which is just giving the kids a few bottles and let them run free until the house burns down. Second, he moved onto a structured management approach, starting with a PowerPoint presentation to the party-goers with a set of goals for each attendee to achieve and various incentives for them achieving the proscribed milestones of happiness. (Listen to the podcast; he’s much funnier than I am.) Lastly, we moved onto the complex system. You use a few strict ground rules to limit behavior, throw in a few activities and adjust activities to the behavior. This is easier to manage and how people actual act and react to their environment.

Dave moved onto a session about pattern recognition. The slides showed a sets of dots, the lines you can make with the dots and various patterns you can make from these various lines. As the number of dots increase the number of possible patterns increases by many magnitudes. The lesson was that hindsight can be 20/20, but is highly unreliable to predict future behavior. Seeing all the data points, with the outcome in front of you , it easy to see how the data showed the future behavior. But those data points could lead to a multitude of possible outcomes.

He also did an experiment with a group of people passing basketballs. Our assignment was to count how many times the people in the white shirts passed a basketball. The video had three people in white shirts and three people in black shirts moving around quickly passing several basketballs. Dave then asked the audience how many passes we saw from the white team. Then, to the surprise of most members of the audience, he asked who saw the gorilla. Replaying the video, someone in a black gorilla suit walks right through the group passing basketballs. In hindsight the gorilla was obvious, but the audience was focused on other data.

Humans are built for pattern recognition and pattern matching influence, not information processes. People do not remember the same thing twice, because we are never presented with exactly the same set of circumstances. People have fragmented memories, blending multiple patterns to reach decisions. People scan a small percentage of the information presented to them and match it to remembered patterns.

Failure leaves a stronger impression than success. People are more afraid of failure than they enjoy rewards of success. We need to be sure to capture the lessons from our failures as much as we capture the lessons from our successes.

Dave finds narratives to have more impact than databases or lengthy best-practices manuals. He finds that when people hear patterns this creates pattern recognition. Fragmenting information into narratives is better than the big bang approach of a full size manual.

Knowledge Management should embrace social computing. He finds the messiness of it along with the narrative and flow of knowledge as a more effective way of conveying knowledge to one big overarching database.

knowledge management

Kathy Curley, Boston University, Panel Moderator, Eric Lesser of IBM Global Business Services, Dave Snowden , Mike Zack , Joe Horvath and Kate Ehrlich wrapped up the meeting by looking ahead.

They only thing they seemed to agree upon was that they disagreed about the future.

There was some commentary about the opening of systems, being more chaotic than controlled company networks. Dave thought that the concept of enterprise software would erode, while other panelists though that the enterprise systems would open (a little).

Another theme was what to do with the changing workforce dynamics. There is a demand to capture the knowledge of retiring baby-boomers. At the same time the younger workers are coming into the workforce expecting transparent information and the visibility of knowledge. Dave pointed out that baby-boomers will not fill out surveys and databases. They will tell stories and will continue to tell stories after they retire. Enterprises need to harness the power of the narrative to collect the retiring knowledge of baby-boomers.

They panel had some agreement on the increasingly common ability to form a network and form a community electronically. Mike was particularly forthright that he wrote his thesis on the importance of face-face contact for effective collaboration, but is not retreating from this position.

Eric put forth the idea of the enterprise creating a platform for its workers to succeed. It needs to give them the ability to collaborate, to provide flexibility to work when and where they want, to allow them to create a network of connections, and to improve their employ-ability. People no longer think that they are going to work at the same company forever.

Mike Zack, an associate professor at Northeastern University, College of Business Administration, threw out the concept of knowledge strategy. We should not just focus on what we know, but also on what we should know.

Knowledge is a barrier to entry into new markets. An enterprise needs to know what it needs to know to be successful in a particular marketplace.

There are internal strategic gaps are the difference between what the enterprise knows and what is should know. The external strategic gaps are the difference between what the enterprise knows and what its competition knows. Knowledge management should align its knowledge and learning initiatives to close these gaps.

He is putting together an interesting study on how knowledge can be a competitive advantage. The challenge was finding an industry that could be analyze to determine the impact of knowledge. He stumbled upon the wheat industry. Everyone has the same tractors, the same land and the same tools, so knowledge of how to use them must be a major differentiating factor.

He pointed out that when selling knowledge, knowledge management becomes a core technology and strategy.

Joe Horvath, of Millennium Pharmaceuticals, started off by pointing out how learning and knowledge management have been historically separate disciplines. With them converging, there are challenges and opportunities.

Knowledge management is designed for providing a future need for an indeterminate task and connecting the user directly with the content source.

Learning is designed to make a change in the learner to improve performance for a specified task, adding value to knowledge, beyond the source, to optimize instruction.

Knowledge management is moving to a task specific orientation and training is moving to direct learners to the source materials.

Joe pointed out some of the factors that are driving learning to KM, including the time crunch on employees, the long lead time in creating training materials and the low value learners associate with training. On the other side, compliance -related knowledge is driving KM to learning. You need to verify that employees have been to certain types of training (sexual harassment, SOX, etc.).

One theme that emerged from the talk is the move to more granular training in smaller bites. The enterprise should be looking to leverage more user-generated content in the learning process. The strategy should be to turn the training material into a hub for post-training reference. People will often better remember the event than the specific content information. People do not want to have to go to different systems to find information. The training repositories should be tied into the knowledge repositories and tied into the working environment of the user.

One challenge is keeping training content synchronized with related content. (This is always a problem. That binder of content given out at a training session quickly becomes out of date.) You need to reconcile quality and compliance standards with the need for speed and flexibility for effective support.

Kate Ehrlich, IBM Research, presented on finding people.

Kate proposed that there are three directions for expertise search:

1. What you know
   -Skills
2. Who you know
   -social networks
3. What you do
   -Enterprise 2.0 tools

What you know
Pro
Need reliable valid information
In technical settings, expertise matters
In staffing, you want the right skills

Con
Hard to define expertise (varies by the person looking for it)
No single expert for everyone
Responsiveness – An expert cannot be reached has little value

Who you know
Pro
You typically get a better response from people you know
You are more likely to reach out to people who are trusted and whose knowledge is validated
It is easier and faster

Con
You might not get the right expert
Sometimes you just want the answer and not a person
You can spend extra time being redirected to another person

What you know
(This is the core of most location systems. They rely on some database of information to transform into expertise.)
Pro
People are judged by what they do rather than what they know
Participation in public venues builds a reputation

Con
Coverage -Not everyone uses the application
Reliability – You are not what you bookmark.

She went on to show how she developed these concepts into the Small Blue application at IBM. (Small Blue because it makes Bib Blue IBM feel smaller.) She took the approach of mining a system, instead of self designation, to create the analysis of expertise. In this case, she chose sent emails. Users need to opt-in to the system.

Personally, I am skeptical of a system that relies only on mining to generate expertise, especially for the transactional side of a law firm. I have found that they work well for the oddball requests, but miss core skills. In part this is because of the lack of rich language. For example, who is the expert on the UCC. I have hundreds of documents and emails with UCC in the text but I am not an expert in the UCC.

Small Blue Find delivers your search results of experts and shows how close the person is in your social network, up to three degrees (like LinkedIn). Small Blue Reach shows you how to reach through your network to that person. It also shows their recent blog posts, bookmarks and their community. Small Blue Net shows a visual representation of the network, color-coded by business unit with pictures of each person.

She had some interesting data on what caused someone to select someone from the list of experts presented to them. She found that page ranking had an impact, but the closer the person was in your community was the biggest factor. Participation in blogs was a big factor. The person blogging was advertising that they are willing to be contacted about the content of their blog.

She found a notion of expertise sufficiency. The expertise searcher gets to a point where expertise is sufficient and then other factors kick in.

She found that big users of the system are people interviewing for other positions within IBM. They use the system to find information about the person who is interviewing them and who they know that knows the person to find out background information.

Small Blue has been deployed for a year, with 2500 of the 300,000 employees opted in. Since each person in the system brings all of their contacts into the system they get great coverage.

knowledge management

Today’s Bentley Event for the KM Forum was a great event.

I had planned to do some conference blogging, but ran into some technical problems. There was a shortage of plugs so I had to run on batteries. I sat near a wall, but still no plugs. Also, the room was so bright that I could not see the screen with the lower level lighting on battery mode. Lastly, I could not pickup the wifi connection.

I will post up my notes shortly. Here was the program:

Finding experts: Who you know matters more than what they know – Kate Ehrlich, IBM Research

Convergence of Learning and Knowledge – Joe Horvath, Millennium Pharmaceuticals

Raising the Strategic Profile of KM – Mike Zack, Northeastern University, College of Business Administration

The Case for Tacit Knowledge – Larry Chait, Chait & Associates, Moderator and Speaker

Sense-making and Knowledge Management – Dave Snowden, Cognitive Edge

Where KM will be in 5-10 years – Kathy Curley, Boston University, Panel Moderator, Eric Lesser of IBM Global Business Services and all speakers

knowledge management
KM forum