All models are wrong

Somewhere along the line I got the phrase 'All models are wrong. Some models are useful.' This has come to top of mind during the week when models of all types have entered my consciousness. This week I've been walking round a full size cardboard mock-up of new office space and furniture that the intended occupants are walking around and through, making comments on its viability and suggesting improvements. It's great fun seeing the ease with which the cardboard can be picked up and re-sited with no difficulty. Cardboard boxes piled one on top of the other represent standing work stations, and flip chart paper the computer monitor. Intended occupants are assessing light levels, asking questions about noise, and so on.

This exercise was followed by a trip to an office furniture showroom where the same people now experienced the type of furniture that would go in the spaces. So where we had the cardboard mock-up of six people sitting at what is called benching (essentially akin to a long rectangular dining table that in the café chain, Le Pain Quotidien, is called 'our communal table' and has a little spiel associated with it ) in the office showrooms we went to they were sitting at the real thing and thoroughly enjoying it. But again the showrooms are just a model. We don't know what the real thing will actually be like, and that's where the phrase sprang to mind, because people are fearful that the model is useful in theory but could be wrong in practice.

Sidebar 1: There is a tremendous amount being written about these new ways of working. A Harvard Business Review article (July 2011) 'Who moved my cube' is one that explores some of the pros and cons in a fairly succinct and neutral way.

The thought that "All models are wrong. Some models are useful" has come up differently as some of my colleagues and I try to agree what organizational systems model we should develop, adopt, or adapt to explain the need to align organizational elements. Personally I am not wedded to any one systems model and there are many of them – Burke Litwin, Nadler and Tushman, Galbraith's Five Star, McKinsey 7-S, are just a few well known ones, and I have developed several different ones with clients that match their own organization's language and style. The point of a model is that it should be useful, and in my view they are all 'wrong' i.e. there is no 'right' model for all situations.

Sidebar 2: In my book The Guide to Organization Design I compare a number of the common systems models and have put up the comparison as the May 2012 tool of the month on my website.

A systems model is useful in helping to explain that organizationally things can go wrong, or less well than they might, if you focus on one organizational element neglecting others. So, for example, if we go down a route of introducing benching, then in one scenario the technology has to be available to support a constant turnover of users – fixed desktop computers and monitors would defeat the object of a 'communal table' – if we were going down the flexible working option that implies a turnover of 'guests' at the table. In the benching scenario 2 where we do have fixed monitors because each worker is to be assigned a seat at the table, then we would have to be cognizant of the personality issues that might surface if people were assigned space next to someone they didn't get on with. (As anyone who has been to a dinner and has sat marooned next to someone they couldn't get along with will know. Read the Dylan Jones piece if you've never been in that situation.

Systems models help you work out some of the key things that need to be considered – but the models can never factor in everything you might come up against.

A third occasion this week which brought to mind 'all models are wrong, some models are useful' was today when I was having a conversation today with a business man working in Sudan. We were talking about models of succession planning and talent management and wondering whether the traditional western model (US/UK predominantly) of these would be workable in Sudan? We agreed that what he termed the 'copy and paste' approach would be inappropriate. Any models have to be developed and tested with the purpose and context in mind. But then I'm not sure how easy it is to develop a model that isn't overwhelmed by one's own experience and knowledge of specific models. I'm reminded of a quote and have no idea where I got it that reads "Education and experience both conspire against us when it comes to predicting the future potential of a new idea. Each acts as a funnel narrowing our field of vision so tightly that eventually we only see what's already behind us." (I see I quoted this in one of my blog pieces in 2009 – oh, the power of Google).

One of my all time favorite books is a sci-fi by an astronomer called Fred Hoyle – the book is "The Black Cloud" which I read when I was a teenager and found a recentish review of. The reviewer doesn't mention what I took as the abiding learning from the book which was that the knowledge transfer that took place between the omniscient alien being and the UK scientist should have taken place between the omniscient being and the mentally challenged gardener (who essentially had a 'blank slate' mind). The mental models of the omniscient being caused such conflict in the brain of the super intelligent scientist that he dropped dead (sorry about the spoiler). This serves as a guide point that models are only useful up to a point, and it should be constantly borne in mind that they may be wrong and/or lead you to the inability of being able to accept any others.