Fuzzy group exercise

ThomasBayes I promised to write up the fuzzy group exercise I ran at Measurement Camp yesterday. To explain to those who weren't there how it was done and to report some of the learnings and how they might be applied.

I called it the Bayesian fuzzy group exercise because I had to call it something.After the invention of the Rev Thomas Bayes. No doubt an expert in Bayesian logic will now tell me it has nothing to do with calculating the probability of outcomes with only a sample. I chose the name because too often group activity whether in research or brainstorms aims towards consensus.  And this simple and rapid technique is a very efficient way to build a map of definitions, to assign them a place in a hierarchy – is an idea the opinion of a minority or does everyone agree? So potentially very useful in determining what kind of messages to put into a social media campaign – whether we are aiming for consensus or whether we are identifying ideas which are powerful and emergent but still marginal. I developed  the fuzzy group exercise to use in workshops and I have also used it in research groups where time is precious. It is a very powerful tool for getting a range of points of view out into the open in a very short time. So do have a go at using it.

Here's how it works: 1. You ask a question: yesterday's questions was How does herd thinking contribute to the success of social media campaigns?

2. Ask every one to write down their answer to that question whether they have one thing to write down or several to list. Allow 3 minutes for this.

3. Then ask them to find someone they don't know and to compare what they have written. To see if there are any areas of agreement. If so this is logged as a pair agreement. I allowed 5 minutes for this because it really gets thoughts flowing as people interact.

4. Then put the pairs into foursomes and ask them to log whether there is any agreement at the pair level which can be promoted to a foursome agreement. That takes another 3 minutes.

5. Then put the foursomes into a single group of 8. To see if there is any consensus at the 8 level. This takes around 3 minutes.

So barring the introduction we are now 14 minutes in. Yesterday we had 20 people do the exercise and we have got a lot of thoughts about the contribution of herd thinking onto paper.

6. Now pull everyone back together in plenary session. We start at the top with what the two and a bit groups of 8 agreed on. Nothing.  We went down to the 8 level. And got one level of agreement. Then down to the 4s and the ideas started flowing. Each one of these I captured on a flip chart logging at what level it had emerged.  No idea is repeated – the more consensus an idea has the higher it appears in the hierarchy. Never twice.

At the bottom level I asked if there any individual ideas which hadn't secured agreement from anybody. We got quite a lot of those. And by this time people started offering ideas which had emerged during the course of the exercise. They weren't top of mind. This process took about 10 minutes.

What we have here is a map of what this group of people think is the contribution of herd thinking.  I can't share the map with you because I had to dash off to a meeting so the flipchart is probably still up in Digitas boardroom! But here are some learnings. Which you might find useful in thinking about how ideas spread through groups.

 Bayes chain
Firstly the lower down the hierarchy you go usually the more interesting the ideas get. Consensus squashes good ideas because as they become more 'popular' they are legitimised. They may not be very interesting but we still pay assent to them as something everyone else things. 

Lower down the hierarchy ideas don't get much further because if you can't secure the agreement of one other person they just don't get promoted. But that is where all emergent ideas begin. And those who are able to identify them quicker than anybody else have an advantage.

Ideas which do get agreement at the pair and foursome level have some momentum but have not yet secured consensus.

Someone commented that the role of influencers may be key in promoting ideas in the pairs and foursomes so they have more likelihood of being promoted. But even here the influencers were having to take ideas which they assume were shared by more than one person. It didn't mean that they could force their own opinions on the crowd.

There are issues about creative development. Classic integrated comms tries to wrap everything round a single idea. If this hierarchy of ideas is representative then following the principle of lighting many fires it might be that the creative agency needs to ensure that the fires are different from each other and not pale imitations of the same fire.  To have a chance of finding an idea that will really travel somewhere.

And lastly it raises issues over measurement. The fact that an idea gets strong consensus may mean that its past it. Big Brother is a spent force. But it would secure strong consensus because it has been a mass phenomenon. So perhaps the focus of measurement needs to be on how far and how fast an idea is capable of travelling. Its not about getting to the top. Its about creating momentum.

Let me know your thoughts. And try it as a brainstorming technique for mapping emergent ideas and see how you get on.

And if you want to read more about Bayes fuzzy logic.. I haven't assigned probabilities to these but a mathematician might want to go and have a play. You could always turn the map into a list – put in on an online survey and get respondents to prioritise the definitions..



Twitter Updates
Photo Albums
TV stations JeddahWillie Williams exhibitionGrindleford Grinch April 2007Planners drinks night No1 BucharestPutna Sihastra Monastery 2011

Designed by Matthew Pattman