As you can see I have explored my HTML guide, consulted our web master and I can now load images: expect a colourful new year in consequence! This post is designed for those who found (or may find) my post of yesterday on identity a bit too metaphysical for a first reading; i.e. those who are primarily interested in how natural numbers apply to communities and networks. It is designed to give some practical guide lines to community/network development and use of tools and updates my 2002 article on knowledge management which was one of the earliest outings for the Cynefin framework in an application context.
Now in the early days of knowledge management if one was to be cynical, there were two rules in operation:
Firstly, install an intranet/portal, create a taxonomy and wonder (with increasing despair) how to motivate people to codify their knowledge.
Secondly, design a standard approach to building a community of practice, with roll out plans, templates, process etc etc. Then, after the novely has worn off, wonder (with increasing despair) how to motivate people to engage.
Both of these approaches represent an ideal not a naturalising approach. I have previously discussed this but I will summarise the essential difference. In an ideal approach you define how things should be and attempt to achieve it; in a naturalising approach you introduce technologies and practices on a safe-fail basis and see what works. Amplification of good patterns, damping of bad patterns allows something to emerge that is more resilient and risk free, not to mention a lower cost, solution. What I want to do here, linking to natural numbers, is to outline a naturalistic approach to simulating network/community formation.
Please note – this blog entry was updated with corrections on 4th January 2006 my apologies for the lack of proof reading in the earlier version
—– Informal communities —–
When I wrote the 2002 article I had just gathered some data on number of on line communities within IBM Global Services, both formal and informal. Like most other service/consultancy organisations IBM had rolled out the approach to communities of practice. One look at the one on knowledge management and I never went back, all information management based on formal process and an over structured taxonomy. Some worked, mostly those in engineering communities which seems to be a wider pattern around the world. It also makes sense, the early work on CoP all studied engineering communities who understand codification and sharing within limits. The urge to share knowledge however is not confined to Engineers. Others also will do so, but the form will be particular to the group.
Let me continue my story. in parallel to the formal CoP program, IBM had stripped down a Lotus product (team room) to a basic, easy to set up utility called a Work Room. These were ad hoc, set up as needed without any real process. They were often unstructured and critically participation was a voluntary act. When they fell over (well it was stripped down system) people just fixed it and moved on. In the formal system they blamed the company for system failure.Checking the figures we discovered that the ration between formal and informal communities was around 1:1000 and the number of the informal virtual communities was around half the number of staff. Subsequent work in other companies validated those rations as reasonable heuristics. One scary happening however. I presented this at a conference where on the senior IBM knowledge management staff was present. Her response to the news was to start a process to bring the informal communities under control …….
Now this was four years ago, pre-blog and pre any wide-scale adoption of social computing. What we witnessed then has been validated in other environments and by the growth of blogs, wikis etc. Give people simple tools with no formal process or designed taxonomy and they will naturally find ways to share and create knowledge. This type of activity is complex, it is not structured and cannot and should not be controlled.
At the same time as we were making those observations in IBM, list serves also started to gain mass use. In knowlege management actKM was one of the most successful, although it seems to have dried up recently (hopefully that is just the Christmas break). List serves allowed people simply to say what they thought, comment and link to other people. Again they were voluntary in nature and allow for threads to be taken up, dropped, diverted without central control. I think there is now some evidence that the growth of blogs is starting to hit the list serves, but its early days yet. Add in the growth of public wikis, and the development of wiki’s that look like word processes rather than a motley collection of HTML, and you have a significant increase in infrastructure and tool capability. People are responding to, and using that capability. The critical point is that informal ecologies of knowledge exchange work, they are also critically and wonderfully messy, human systems not a nightmare of structure and process, as such they are not stable, but they are resilient.
—– A naturalising approach —–
Now of course, it was this phenomena that the early researchers in knowledge management had found. One thinks of Etienne Wenger’s pioneering work in observing naturally occurring use of virtual environments by engineers. The problem was when people went from a researchers description of what had grown naturally in the past, to a prescriptive recipe things went wrong. People never accurately report all the factors which led to the success or failure of a project, retrospective coherence clicks in. A researcher, now matter how gifted tends to pay attention to data that appears causal and which fits their emerging hypothesis. Also the fact that something worked once in a specific context does not mean that it will work again even in the same conditions, or that you can accurately replicate the starting conditions.
So lets look at an alternative naturalising approach to building communities for knowledge flow, organised in seven steps (so the median of the readers short term memory):
- Install software for blogs and a user friendly wiki (don’t skimp here on price, you need something which is natural to use and which looks like a Word document). Don’t lay down rules and regulations for their use just set it up. Run multiple on line and f2f training programmes around the organisation; ideally you train the trainer here with support staff in the business units. Find the opinion leaders and get them up and running, then get them to train other people. Do everything you can to avoid this being a corporate programme. Learn from what other people have done using blogs but under no circumstances try and copy what someone else has done no matter how successful. Your context is different. You might want to throw in the odd list serve around specific issues but you will need to make sure that a small cadre of people commit to early participation.
- Now be patient, you can stimulate this, get senior people to blog internally or externally. Richard Sambrook at the BBC is a good example of this. But people have to choose to use the technology. Don’t spend money on roll out programmes, use it on training super-users amoung the opinion leaders. You can use social network analysis software on the blogs to see who is connecting with who and also to produce objective measures of connectivity where you need to have some sort of performance target. It may happen quickly or may take a year, don’t get heavy handed, fertilise the soil do a bit of covert or even overt weeding from time to time but wait for a critical mass of activity to emerge.
- Next you talk to the experts and knowledge workers, Find out what is working and what is not. Start to put in some utilities to allow sharing material across the boundaries of the informal networks. A good document repository with a search engine is one basic; teach people how to link to it. By now people should be used to HTML references. If they are anything like me (and others I know), once you get used to HTML you want the same facility in everything you write. Don’t make use compulsory, but make it easy and help people. Start to look at progressively removing attachments from email and get people to reference centralised systems. Take small groups of experts and get them to act like review editors in a journal. Taking material from blogs and wikis and formalising material that requires version control, placing it into your more structured systems. Don’t do this unless it’s necessary. Now you can get your power users to start referencing this material.
- For things like health and safety manuals and other formal material, start to get people to write in limited authorship wikis or blogs in narrative form with hot links to the documents. Don’t get into the old way of working, use an informal, conversational front end to a more formal backend system. Make sure that anyone who uses a document can either change it (if they have authority) or at least comment or suggest changes. Talk to a Cognitive Edge practitioner about using Sensemaker™ to hold a lot of the material in narrative form and link it to the more formal documents. Now what you are doing in this and the previous stage is to shift knowledge/information/activity from the complex domain in the cynefin framework to the complicated one. From a purely informal system to an expert network.
- Some of the material will stabilise to the point where it can move from the complicated to the simple, from expert to formal with broader populations. Material which goes there will not recycle, it is now static and stable and you have to determine its life span before destruction. Try and avoid the temptation to put material in the simple domain into formal review mode It’s better to regenerate in the shift from complex to complicated to simple; its cheaper and the material is more up to date.
- So we now have a cycle from the informal to the formal, in effect the tacit-explicit aspect of conventional knowledge management. But we also need to make sure that our expert communities in the complicated domain should not be allowed to consolidate and structure their knowledge without a regular disruption. The history of experts in science is not good when it comes to innovation. There are various ways in which this can be done. One of the most effective is to create a short term programme in which a group of experts are required to work with people on difficult or intractable problems with people from radically different backgrounds. Starved oftraditional resources, under pressure of time and with the perspective shift of people with very different backgrounds the pre-conditions for innovation are present. This shifts from complicated, through a shallow dive into chaos and back to the complex. During the innovation process the participants use blogs to record their experience, link and connect with different knowledge bases. This is a process of renewal, a way of preventing the ossification of knowledge. I will blog on this process early next year.
- Now there is one final opportunity. The best chance any organisation has to do things differently is during a crisis. During such a time people’s willingness to do things differently, to be open to working with people who they normally avoid is higher than in normal circumstances. Appointing someone to work in parallel with a crisis team to seek opportunities for innovation is one of the great knowledge creating opportunities.
—– The dynamic learning cycle —–
Now the shifts between domains are illustrated above. The dynamic learning cycle starts with the informal, looks to shift some knowledge from there to fthe domain of experts on a just in time basis. That knowledge is either stabilised into the formal (although links are maintained) or cyclically disrupted to prevent entrainment and ossification.
Now we can look at natural numbers. Again this is illustrated in the picture. Informal communities link back to natural levels of trust, they need to be less than 15. For expert communities some degree of knowledge of the other participants is necessary, but deep trust is not. SItuation trust, contextual and professional trust all come into play. The 150 limit therefore cuts in here. For a formal community it does not matter, you have enough structure, and the material is at a low enough level of abstraction that anyone can use the material. For a crisis you need very small focused teams which is where the 5 limit fits in.
Now a critical qualification here. The numbers can relate to individuals,and they are good guidelines as such. However given the way social computing works we don’t have to restrict participation in this way if instead we think about identities or coalescences. Provided the central actors are limited by the numbers, the number of actual people can be large. Look at the number of lurkers in any virtual community, while the active participants tend to be about 15 overall, 5 in any particular thread. By observing natural clusters in the use of social computing tools, those clusters can be given roles or functions in wider groups. In effect this is a nodal network. The nodes stabilise their linkages and act as a focus for activity. By using simple analysis software you can not only monitor those patterns, but you can also measure and target their connectivity.
One final point here. All communities are networks, in so far as they are linked, but not all networks are communities. A community has a common purpose, it may not be stated, but it is known. It is the way we do things around here, which is not a bad definition of culture.