A grain of sand: Innovation diffusion

Today I was privileged to be invited as to present along with four others at an Innovation Diffusion workshop for UCL in connection with the NHS. The format involved each if us presenting for ten minutes each along with some interrogation from panel members in the morning. In the afternoon the wider audience synthesised the the material and came up a list of characteristics that the health system needs to allow innovative ideas and practice to diffuse through the service as a whole.

I promised that I would summarise my ten minutes here and provide some links to supporting material. I’ll do my best to do that so this post is of general interest not just an aide memoir to those who were there. One of the themes, namely the criticality of weak signal detection is well illustrated by one of Hugh’s Gaping Void gems shown. There are some wonderful ideas out there, but how to make them visible?

Introduction

I referenced Cognitive Edge’s background in weak signal detection (that grain of sand) in counter terrorism, the philosophy of cognitive complexity, taking a natural science approach to social systems and our development of SenseMaker® (described by one Home Office anthropologist as the first example of a system for distributed ethnography). I mentioned our work with the US Army, the developments we have just completed to create a battlefield KM demonstrator that has applications outside of a military environment. I also talked about our recent project for Global Giving in Kenya and some of the conclusions that came from that.

I then gave a basic overview of complexity theory, defining it in terms of constraints and agent behaviour using the children’s party story to illustrate the point. Decided to have a go at sick stigma, BPR with bible belt cultism added on for good measure, high priests of the cult wear different coloured belts etc. Never understood why anyone combines it with Lean which has a different philosophy.

Two minutes in, eight to go …..

Observations

That complete I went on to make some observations on the field:

• There is a big difference between innovation in a market and innovation in an apprentice. The former relies on failure, risk investment, high returns etc. etc. Those conditions are rarely present in a professional or government environment. Here apprentice models, social networks are more effective at diffusion and development of ideas.
• There is a paradox in that highly codified, highly abstracted material diffuses best (per Boisot I-Space), but is the least likely to be innovative or novel. Things that are highly codified are generally established practice rather than novel practice. That means that database and similar traditional KM solutions are not likely to work and we need to think differently
• Context is everything and often neglected. Something that works in one context may not work in another even if they are very similar. Ideas and practices need to have enough flexibility to adapt and ideally to combine with existing practice and other ideas. It means that pilot approaches have an inherent problem in that the initial success results in a specific context with a lot of effort. It won’t necessarily scale.
• While incentives have their place, its unlikely that financial incentives will stimulate innovation per se. Major prizes in public competitions do work but that is very different. Financial incentives tend to attract those who are good at winning incentives rather than true innovators. We had a lot of that in IBM; put money on something and the bonus sharks would arrive.
• Remember Goodhart’s Law – any statistical instrument used for policy looses all value, or loosely translated the minute a measure becomes a target it ceases to be a measure.

Four minutes in, six to go; good audience response so I’ll risk a bit more content than I planned

What should we do? What principles should we adopt

Now moving to action and another set of five (I do like symmetry in rhetoric)

• Key lesson of complexity based strategy, the shift from fail safe design to safe fail experimentation. That means creating space where the cost of experimentation is acceptable. You can only understand a complex system by interacting with it, it cannot be studied in abstract. This type of approach allows you to create a resilient organisation (early detection, fast recovery/exploitation) which is more appropriate than a robust one (nothing is allowed to fail, planning takes a long time etc. etc.)
• Need to get the granularity right, both for information and for organisational structures. This links back to the earlier observation on diffusion and codification/abstraction. Finely grained objects diffuse faster, can be combined with other objects both novel and existing in new ways. Human brain evolved for this synthesis of multiple fragmented items; we are adept at conceptual blending, linking and connecting things in unusual ways. Systems and processes need to do this. With tools such as SenseMaker® we can handle very large volumes in real time.
• Use technology to support human sensor networks. Link and connect people, allow then to exchange ideas and observations. Social computing does this well, although more advanced tools (such as the Auditor module of SenseMaker® ) allow whole of workforce engagement in real time decision making, critically with an audit trail. These approaches allow co-evolutionary development and adaption of ideas, moving away from push or pull marketing to push-me-pull you (with apologies to Dr Doolittle).
• Exaption is more important than adaption; traits that evolve for one context accidentally get used in a different one and we get true innovation. Most of human development including language is the result of exaptive processes. You can’t determine in advance what innovation is needed, you have to create an ecology in which novel solutions can emerge. There are some related techniques here, the use of metaphor for example can allow association or linkage of ideas in new ways. Defining problems via narrative, especially counter-factual metaphors stimulates people to come up with material they might otherwise not recall or perceive as irrelevant.
• Cases (single narratives) and models are nothing to do with innovation As Jim March says they perpetuate what is known. The problem is case based recipe approaches dominate management consultancy thinking to little effect. Cases confuse correlation with causation, models confuse simulation with prediction. We also need to shift from process based outcomes to measuring impact (quantitative measurement of beneficial outcomes that could not be defined in advance) and the good news we can

Concluding I emphases the need for messy coherence, relaxing constraints to allow new patterns to emerge. We handle uncertainty with sound theory, tested through multiple safe-fail experiments. Praxis as they say makes perfect.

Oh, and if I am honest 12 minutes …..

Links to related material

Sometime ago I gave a general presentation to senior leaders in NHS London. That covers a lot of the material. The slides for both the keynote and the workshop can be found here. The podcast for the keynote is here, and for the workshop here. Now it was not really a workshop! I was asked to do one and wrote an agenda assuming a couple of hours not 40 minutes! So I chose to cover the material at a high level and provide backup here. So links and materials relating to both poadcast and workshop are as follows:

Link: Detailed discussion of SenseMaker® (the micro-narrative software which is not open source. That presentation includes Cynefin and the S-curve material. There are also case studies here and here. The latter links to two Huffington Post Articles that I referenced in the keynote.

Link: Videos that elaborate on the analysis side of SenseMaker® (these are real cases)

Link: Various videos of the Children’s Party story, an overview of Cynefin and two presentations on innovation.

Link: The US Navy version of the Children’s Party story told as a Socratic dialogue

Link: Social network stimulation – creating an innovation network (not discussed in the event but its linked)

All methods are open source and there is other material to explore on the site

Articles

Link: The Harvard Business Review article on complexity theory is here

Link: A more academic article with older language with Kurtz, one of the most referenced

Link: Multi-ontology sensemaking (a half way house between the above two

Link: Original Cynefin model paper in the context of Knowledge Management

Link: Academic article on abductive or narrative research

Link: Summary of Complex Adaptive systems from one of the leading authors in the field

You can also find third party articles here, and more by me here and other podcasts here.