The Emperor’s Eunuchs

I’ve missed blogging for a couple of days, in the main because I want to complete the Emperor series without interruption and that didn’t allow for any shorts in the meantime. As it is, its going to be three more blogs. Today I want summarise the last three and then look at some of the inherent problems involved in creating any central function for knowledge management (especially under that name). Tomorrow I plan to outline a possible alternative, and then on Sunday (probably in the air to Los Angeles so it may not be up until Monday) look at what I will call necessary compromises to fit within the expectations of power. Many thanks to all those who commented as well, I plan a general response once I have completed the series.

So where are we?

In The Emperor’s Chess Board I argued against integration approaches to data which attempt to join up the dots. This was in the context of the idea that previous failures in the case of 9/11, Katrina etc were failures of knowledge management. and the implication that a centralised KM function in government would have prevented them. I extended this argument in Between Empires to introduce the argument for distributed cognition as opposed to top down directives. The Empire Repeats emphasises the danger of assuming that this time we will get it right. I also argued against the danger of allowing hindsight as a substitute for foresight. Speaking truth to power is also important and preventing political filtering of advise doubly so. However the reality is that there will always be someone who, with the benefits of hindsight, we wish we had paid attention to. After the event we can see who got it right, but it does not mean we will in the future unless we radically change the process by which information is presented to decision makers; centralising KM is not only not radical enough it will make it worse. The overall theme here returns to my quote from Lincoln in the opening post: As our case is new, so we must think anew, and act anew. We must disenthrall ourselves, and then we shall save our country.

So what is wrong with a central function?

Well, where to start? I decided the best thing was a set of bullets, some of which are connected some of which are not. Oh and a note, if you had not already realised it, all of this has been in the context of current proposals in Washington so its US centric, but most of the points are universal. I am also going to be a bit extreme, deliberately, to make a point.

• For a start, if you create a centralised function or co-ordination based on existing knowledge work in different agencies then you inherit their practice and their politics. There is much good and bad practice around, and a much that needs to be defended. Most people will have already staked their reputation on a pet approach, which may or may not have worked. They will defend that and advocate it, horse trading if necessary with other interest groups. People are blind to failure.
• A variation on this involves people accommodating unorthodox success into their pre-existing models. I remember a presentation in the US recently when I talked about the success of blogging during the Iraq war. I repeated a prior comment that the US army had the best method for knowledge capture I had seen anywhere in the world, but the worst method of distribution; far too much effort into taking raw field data and synthesising it into doctrine. The blogs on the other hand were direct raw data transfers, horizontally and immediately. At the end of this a delegate who shall remain nameless came up to me and told me that he agreed with everything I had said, and the US Army Lessons Learnt Centre were now taking the blogs and summarising them into material with higher utility. He just didn’t get it, and I have seen many a similar form of pattern entrainment. Even in the face of failure people persist in justifying their past practice.
• If you centralise things then bureaucracy creeps in regardless of good intent. I remember in IBM when a new structure for emergent Business Opportunities was set up (EBOs and I was in one). The idea was a single VP and a budget that would be dispenses to new ideas without the constraints of normal process. Great idea for innovation, the problem is that the VP appointed worked form a model by which a VP has a few directors, staff and lots of forms. Within six months most of the budge was going on the process, and less and less was available for innovation. Processes, the need for consensus are bound to slow things down.
• When you have a central function people start to assume that its no longer their responsibility, its now down to the central group. That group now starts to create its own language, its members go to conferences and then start to present case studies based on their work (this is where it often goes wrong as they start to believe their own myths). Their separation from the workforce becomes more extreme over time with increasing separation between reality and myth.
• Politics is the essence of any centralised function. In the Byzantine Empire the civil servants were castrated to remove ambition, but in practice this produced one of the most devious and dedicated civil services of all time. Arguing the service only role, while setting up a centralised function with power of recommendation and co-ordination is Byzantine in nature, hence the title of this blog.

There is another major objection, and the most important. This is not a problem of knowledge management alone. It is a wider problem of decision support, cognitive bias, weak signal detection, trust between silos and many other things. It requires an inter-disciplinary form of praxis not a centralised function writing papers, making recommendations, keeping minutes and fighting for its funding. This is a big problem, not a problem of a single function, let alone a function whose name is now largely associated with technology alone. Of that, more tomorrow.