On Vector Measurement: the ‘what’ and ‘why’ of

I’ll start this post with an excerpt from a podcast in which Stephen Fry was the guest. He recalled the following passage from a Sherlock Holmes book:

“You know Watson, the statistician has shown that we can predict, to an extraordinary order of accuracy, the behaviour of the ‘average man’

…but no one has yet, and probably never will be able to predict how an individual will behave.” (Sherlock Holmes, ‘The Sign of Four’ as verbally recalled by Stephen Fry)

Fry went on to add:

“We can be talked about as ‘a mass’, and advertisers and politicians…and all kinds of other people are very good at knowing how we behave as a group, but as individuals we are unknowable without face-to-face conversation, and [knowing a person’s] history and so on.” (Stephen Fry)

This made me smile. He is discussing a point that is (for me) quite profound and I’d like to use it to link a few things together1.

A reminder about Complex vs. Complicated

I wrote a post ages ago that explained the difference between a complex and ordered system (ref. ‘It’s complicated…or is it?’)

If you’d really like to delve into this, then I’d recommend looking at the Cynefin sense making framework2.

In short3:

  • An ordered system (whether simple or complicated) is predictable. There are known or knowable cause-and-effect relationships.

There are right answers (which may be self-evident or may require expert diagnosis)

  • A complex system is unpredictable. The elements (for example people) influence and evolve with one another. The past makes sense in retrospect (i.e. it is explainable) BUT this doesn’t lead to foresight because the system, and its environment, are constantly changing.

 It’s not about having answers, it’s about what emerges from changing circumstances and how to respond.

The difference between the two is hugely important.

On working in a system with a purpose of helping people

There are many social systems that are put in place with the intent of helping people with their lives. As an example: most (so called) developed countries have social welfare systems that provide a level of income support. They also help people with their housing needs and gaining employment.

Each of these welfare systems has a choice as to how it sees the people that need their help, and therefore how they choose to design their response.

The ‘average person’ response

The welfare system can look at their population of ‘clients’ and create a host of data about ‘the average’.

They can even break this population down into cohorts and look at more detailed ‘averages’. They might even design ‘personas’ around the ‘average’ per cohort.

But, if they design responses to individuals with reference to these averages, then they are falling into Sherlock Holmes’ stated error – that they believe they can know about an individual from an average.

They would be presuming an ordered (complicated) system and designing answers in response.

Example4:, We can all understand that, on average, it is beneficial for a person to gain employment and thus be able to become independent from the welfare system…so the ‘ordered’ answer must surely be ‘get everyone into work’…so let’s direct all our focus (and targets) to achieving this!

The ‘individual’ response

Each person reliant on a welfare system has a complexity to their life (whether we see this or not).

Expanding upon this, the complexities for many ‘clients’ can be huge – such as:

  • dependents (children, and others that they care for)
  • lack of permanent address and other financial barriers
  • limited education and work experience
  • mental health (incl. depression, anxiety), habits and addictions
  • physical disabilities
  • family violence
  • criminal records
  • history of institutional care
  • …etc.

If those working in the welfare system want to achieve meaningful help, then the starting point is for them to know the client as an individual, and then iteratively work alongside them according to what emerges.

Measurement of success

We would hope that each welfare system has thought about what its purpose is, from a client’s point of view, and that this is the anchor point from which everything else is tethered.

Such a purpose will likely be about helping people towards ‘good’ outcomes (such as independence, safety, …and other dimensions of wellbeing)

A welfare system should want to measure its performance against that purpose.

If I’ve assumed an ordered system (via the ‘average person’ design response) then I will take my ordered answers (in this example ‘getting people into work’) and likely measure things like ‘Number of clients into employment this period’. I might even set targets to (ahem) motivate my staff….

…and I will predictably (though unintentionally) promote dysfunctional behaviour5:

  • A mindset of whose ‘on my books’? (and therefore, who can I get off them)
  • Who can I get into work quickly? (with the reverse effect of leaving the ‘difficult ones’ languishing to one side)
  • What work can I get them into? (as opposed to what will help them succeed)
  • Who needs coaxing/ persuading and what can I use as levers to do this?
  • Who got themselves into work by themselves (i.e. without any help from us)…so that I can ‘count them in my numbers’
  • ….

Attempting to push someone into work might be an incredibly dumb thing to do for that individual (and for those that depend upon them).

The above is not blaming anyone working in such a system. It is to say that these are, sadly, “healthy responses to absurd work” (Herzberg).

To Vector Measurement

If I correctly see that the welfare system is a complex one, then I realise that I need to work at the level of the individual.

The type of measure that fits for a complex system is a vector measure. A reminder from schoolboy maths that a vector has both speed and direction. It is especially used to determine the position of one thing, in relation to another.

I will know the performance of my system if I regularly measure, for each individual, their speed and direction of travel – from where they are now towards, or away from, a better place (as defined by them).

Such a measure causes a total focus on the individual:

  • whether things are working for them or not
  • whether to continue with the current stimuli and perhaps amplify them; or
  • whether to dampen them and work to stimulate new ideas

It’s very likely that there will be a collection of unique needs per individual, and therefore a set of vectors to be monitored (one per need). These may cluster around the various dimensions of wellbeing (see footnote for a wellbeing model6).

Example: My financial wellbeing might be slightly (and temporarily) better off because you pushed me into a job, but my mental wellbeing might be sinking like a stone because it really wasn’t appropriate.

Which leads to…

Measuring over time

Vector measurement shouldn’t be one-off thing. The individual is on a (lifelong) journey. The reference point is continually changing, but the question of ‘better or worse off?’ (or perhaps progressing vs regressing) is a constant.

As such, each person steps from one place to another, sometimes forwards (i.e. towards where they thought they wanted to go), sometime diagonally (to new possibilities that have emerged) and sometimes backwards (requiring reflection and perhaps some helpful interventions).

Playing with a way to visualise such a journey7, it might look something like this:

If we were measuring the performance of a system aimed at helping individuals, we would want:

  • the aggregate of our clients to be going forwards (and most certainly not be stuck and dependent on us); and
  • to spot the individuals who are stuck or, worse, going backwards, so that we can (quickly) help;

Returning to the ‘average person’

Every welfare system says that it wants good outcomes for those that they are tasked with helping. Whether this happens will depend upon how we think this can be achieved.

I contrast:

  • Fighting people into (what we have determined for them as) good outcomes;

with

  • Dancing8 with people towards what they arrive at as their good outcomes

I hope you can see that if we truly help ‘the individual’ then, on average, people are quite likely to move towards employment (and better mental health and…). But this is a case of ‘cause’ (helping individuals) and ‘effect’ (improving the average).

Conversely, I can ‘badger the non-existent average’ till I’m blue in the face…but it would be the wrong place to work from!

In summary:

I’ll end with one of my favourite quotes:

“Simple, clear purpose and principles give rise to complex intelligent behaviour.

[Complicated] rules and regulations give rise to simple stupid behaviour(Dee Hock)

The simple, clear measure of ‘are you better or worse off (as defined by you)’? will give rise to all sorts of varied, individualised and highly relevant actions and interaction.

The complicated rules around, say, how many people a team is supposed to get into employment this period, what counts as ‘getting employment’, what rules determine who gets the credit, how long does this ‘employment’ need to be sustained to keep the credit,…and so on, will give rise to simple, stupid behaviour.

 

Footnotes:

1. Sources for this post: As is so often the case when I am energised to ‘write it down’ in a post, the ideas within are because of a coming together (at least in my mind) between a few separate things.

The main two for this post are:

Whilst the former is about ‘the individual’ and the latter is (I think) written with respect to, say, an organisation (or bigger system), I see them as complimentary.

Richard Davis’ Chapter 4 ‘Using data’ nicely shows an example of an individual with a set of needs (as defined by them), and how they are doing against each over time.

Dave Snowden’s piece adds to, and broadens, Richard Davis’s work by naming, and clearly articulating, vector measurement.

2. Cynefin: There’s also a useful book called ‘Cynefin: weaving sense-making into the fabric of our world’ by Dave Snowden and friends

3. Sensemaking framework: I have chosen to provide a very brief reminder here. I have omitted the ideas of chaos, disorder, and liminality.

4. Employment as an example: I could have used various examples to make this point. I picked employment after reading an article about a new UK Govt. initiative called ‘The way to work’, with a target to get 500,000 into work.

The article is titled ‘Way to Work Scheme: forcing people into jobs they aren’t suited for has damaging effects’.

The article also links on to a systematic review by University of Glasgow on the research in this area. The abstract notes

“…we found that labour market studies…consistently reported positive impacts for employment [i.e. yes, we can ‘force people into work’] but negative impacts for job quality and stability in the longer term, along with increased transitions to non-employment or economic inactivity. …increased material hardship and health problems. There was also some evidence that sanctions were associated with increased child maltreatment and poorer child well-being.”

5. Dysfunctional behaviour: A reminder that this isn’t a case of ‘bad people’; this is normal people attempting to survive within their system.

6. Wellbeing: Googling the (often referenced) notion of wellbeing shows that there are lots of different (though similar) n-dimensional models of wellbeing ‘out there’.

Using the Te Whare Tapa Whā model, as derived by Sir Mason Durie, four core dimensions of wellbeing are:

  • Taha tinana: Our physical health
  • Taha hinengaro: Our mental and emotional health
  • Taha whānau: Our social wellness (e.g. a sense of connection, belonging, contributing, and being supported)
  • Taha wairua: Our spiritual wellness (g. a sense of purpose and meaning in life/ the degree of peace and harmony in our lives )

Other dimensions that feed into (i.e. will likely affect) these four core wellbeing dimensions include:

  • Our financial/ economic situation (now) & outlook (expected)
    • e.g. access to resources (food, clothing, shelter,…)
  • Our environmental situation – where we live
    • e.g. safe, clean, pleasant, cohesive, with access to facilities
  • Our intellectual situation – what we do (education, work, and leisure)
    • e.g. stimulating/ creative, productive/ useful, learning/ developing/growth, autonomous (self-determining)

7. Visualising the journey: I’m sure that there are lots of people far more skilled than me that could really ‘get into’ how to visualise a person’s journey (and probably have).

The point is to be able to see it, as a vector, moving over time, comparing ‘where I was’ to ‘where I am now’ on some useful dimensions. These could be:

  • a set of needs as identified through actively listening to the person (as per Richard Davis); and/or
  • a set of wellbeing dimensions (as derived from a useful wellbeing model)

 To clarify: This would be the opposite of scoring where the person is at on a goal set by the welfare system that is ‘managing’ them.

A reminder that the former recognises the person’s complexity, whilst the latter assumes an ordered reality…which is not the case.

8. Re. Dancing with people: This phrase might sound flippant – I have no wish to be simplistic about people whose lives are really tough. It comes from the rather nice concept as explained by Miller and Rollnick in their book on Motivational Interviewing.

It’s complicated!…or is it?

mandelbrot-setI’ll  start with a question: What’s the difference between the two words ‘Complex’ and ‘Complicated’?

Have a think about that for a minute…and see what you arrive at.

I did a bit of fumbling around and can report back that:

  • If you look these two words up in the Oxford English Dictionary (OED), then you’d think they mean the same thing; however
  • If you search Google for ‘complex vs. complicated’, then you’ll find oodles of articles explaining that they differ, and (in each author’s opinion) why; and yet
  • …. if you were to read a cluster of those articles you’d find totally contradictory explanations!

Mmmm, that’s complicated…or is that complex?

This post aims to clarify, and in so doing, make some incredibly important points! It’s probably one of my most ‘technical’ efforts…but if you grapple with it then (I believe that) there is gold within.

Starting with definitions:

Here are the OED definitions:

Complex: consisting of many different and connected parts.

  • Not easy to analyse or understand; complicated or intricate”

Complicated: consisting of many interconnecting parts or elements; intricate.

  • Involving many different and confusing aspects
  • In Medicine: Involving complications.”

So, virtually the same – in fact one refers to the other! – but I think we can agree that neither are simple 🙂 . They are both about parts and their interconnections.

Turning to the ‘science’ of systems:

scienceWhilst the OED uses the ‘complex’ and ‘complicated’ words interchangeably, Systems Thinkers have chosen to adopt distinctly different meanings. They do this to usefully categorise different system types.

Reading around systemsy literature2, I repeatedly see the following categorisation usage:

Simple systems: Contain only a few parts interacting, where these are obvious to those that look; Extremely predictable and repeatable

Example: your seat on an aeroplane

Complicated systems: Many parts, they operate in patterned (predictable) ways but ‘how it works’ is not easily seen…except perhaps by an expert

Example: flying a commercial aeroplane…where, of note, its predictability makes it very safe

Complex systems: unpredictable because the interactions between the parts are continually changing and the outcomes emerge – and yet look ‘obvious’ with the benefit of hindsight.

Example: Air Traffic Control, constantly changing in reaction to weather, aircraft downtime…etc.

“…and the relevance of this is?”

There is a right way, and many a wrong way, to intervene in systems, depending on their type! Therefore, correct categorisation is key.

A (the?) major mistake that ‘leaders’ of organisations make is they presume that they are dealing with a complicated system…when in fact it is complex4. If you initially find this slightly confusing (it is!) then just re-read, and ponder, the definitions above.

Management presume that they are operating within a complicated (or even simple) system whenever they suppose that they can:

– administer a simple course of ‘best practise’ or external expert advice…and all will be well;

– plan in detail what something will turn out like, how long it will take and at what cost…when they’ve never done it before!;

– implement stuff as if it can simply be ‘rolled back’ to an earlier state if it doesn’t work out…not understanding that, once acted upon, the people affected have been irrevocably changed (and regularly suffer from what I refer to as ‘change fatigue’5);

– isolate and alter parts of the system to deliver a predicted (and overly simplistic) outcome…by which I am referring to the slapstick ‘benefits case’ and it’s dastardly offspring the ‘benefits realisation plan’…

…Management can, of course, invoke the Narrative Fallacy to convince themselves that all that was promised has been achieved (and will be sustained)…whilst ignoring any inconvenient ‘side effects’;

– strip out (and throw away) fundamental parts of a system whilst invoking their constant simplification battle cry…because they can’t (currently) see, let alone understand, why these parts are necessary;

…and I’m sure you can carry on the list.

The distinction between ‘complicated’ and ‘complex’ fits quite nicely with Russell Ackoff’s distinction between deterministic (mechanistic) and organic systems; and  John Seddon’s distinction between manufacturing and service organisations (and the complexity of variety in customer demand).

Put simply 🙂 , in complex systems, it’s the relationships between the parts (e.g. people) that dominate.

So what?

jack-deeWell, if Management understand that they are dealing with a complex organisation then they will (hopefully) see the importance of designing their system to take advantage of (rather than butcher) this fact.

Such a design might include:

  • aligning individual and organisational purpose, by sharing success (and removing management instruments that cause component-optimising behaviours)
  • putting capability measures into the hands of front line/ value creating workers, where such measures:
  • allowing and supporting the front line/ value creating workers to:
    • absorb the customer variety that presents itself to them; and
    • imagine, and experiment with, ways of improving the service for their customers
  • …and much much more systemsy thinking

This would mean creating a system that is designed to continuously adjust as its components change in relation to one another. That would be the opposite of ‘command and control’.

Huge clarification: Many a command-and-control manager may respond that, yes, they already continually adjust their system…I know you do!!!  It’s not you that should be doing the adjusting…and so to self-organisation:

From simple to complex…and back again6

answersThe giant systems thinker Donella Meadows wrote that highly functional systems (i.e. the ones that work really well) likely contain three characteristics – resilience7, self-organisation and hierarchy8.

I’ll limit myself here to writing about self-organisation:

“The most marvellous characteristic of some complex systems is their ability to learn, diversify, complexify, evolve…This capacity of a system to make its own structure more complex is called self-organisation(Meadows)

Wow, ‘complexify’ – a new word?!…and it can be a very good thing…and goes 1800 against the corporate simplification mantra:

“We would do better at encouraging, rather than destroying, the self-organising capacities of the systems of which we are a part….which are often sacrificed for purposes of short-term productivity and stabiliy7(Meadows)

It turns out that complexity isn’t of itself a bad thing…in fact quite the opposite – a system can achieve amazing things as it becomes more complex. Just consider that, through the process of evolution, ‘we’ have ‘complexified’ (can you see what I did there) from amoeba to human beings!

…but what’s REALLY interesting is that this complexity is enabled by simplicity!

“System theorists used to think that self-organisation was such a complex property of systems that it could never be understood…new discoveries, however, suggest that just a few simple organising principles can lead to wildly diverse self-organising structures.” (Meadows)

Meadows went on to note that:

“All of life, from viruses to redwood trees, from amoebas to elephants, is based on the basic organising rules encapsulated in the chemistry of DNA, RNA, and protein molecules.”

In short: Simple rules can allow complex systems to blossom, self-learn and grow.

I believe that a wonderful, and complex, organisation can be created and sustained from living a simple philosophy.

“Simple, clear purpose and principles give rise to complex and intelligent behaviour.

Complex rules and procedures give rise to simple and stupid behaviour.” (Dee Hock)

….so what might such a simple philosophy be? Well, Deming was ‘all over’ this with his ‘Theory of Profound knowledge’* and ’14 points for Management’.

* Deming explained simply that we should:

  • Observe, and handle, the world around us as systems (which, by definition, require a purpose that is obvious to all);
  • Expose, and understand, variation;
  • Gain knowledge through studying and experimenting;
  • Understand psychology and truly respect each and every human being ; and
  • Lead through our actions and abilities.

…and a final warning against that oversimplification thing:

“Collapse is simply the last remaining method of simplification.”  (Clay Shirky)

The short ‘simple’ version at the end of the long ‘complicated’ one 🙂

‘Complexity’ is not the inherently bad ogre as persistently painted by contemporary management. Rather, it can be a defining property of our organisational system that we would do well to understand and embrace.

Let’s feast at the ’complex but right’ bookcase of knowledge, design appropriate and evolving responses based on simple scientific wisdom and climb the mountain…rather than automatically follow the crowd over the ‘simple but wrong’ cliff!

Footnotes

1. Opening Image: This image is a part of the Mandelbrot Set – an amazingly complicated (or is that complex?) image that is derived from the application of a simple mathematical formula. It sits within the fractal school of Mathematics (repeating patterns) alongside others such as the Koch snowflake.

Image Source : CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=322029

2. Systems Thinking Literature: This systems thinking is taken, in part, from a 2011 HBR article Learning to Live with Complexity.

3. Dave Snowden’s Cynefin framework is built (partly) around the difference between complicated and complex….and the importance of correctly identifying your system type before intervening. Snowden’s framework also adds the idea of chaotic systems, where there is some emergency that requires urgent action (without the time to experiment)…where the action chosen may determine how the chaos is halted…which may or may not be in your favour!

4. The inclusion of people in a system likely makes it complex.

5. Change Fatigue: This is my phrase for those people who have worked for an organisation for many years and had the annual ‘silver bullet’ change programme rolled out on them…and got bored of the same lecture and the same outcomes. It is very hard to energise (i.e. excite) someone with ‘change fatigue’.

6. Cartoon: I LOVE this cartoon! It is sooo apt. The vast majority are on the simple road, following the crowd over an (unseen) cliff…or at least not seen until it is too late. A few turn right at the ‘bookcase of knowledge’ – they take a book or two and then travel a circuitous and uphill road to an interesting destination.

7. Resilience vs. stability clarification: “Resilience is not the same thing as being…constant over time. Resilient systems can be very dynamic….conversely, systems that are constant over time can be unresilient.” (Donella Meadows)

8. Hierarchy: I’m aware that some organisations have experimented without a formal hierarchy (e.g. Holacracy). However, even they create a set of rules to assist them co-ordinate their component parts.

It’s worth noting that “Hierarchies evolve from the lowest level up…the original purpose of a hierarchy is always to help its originating subsystems do their jobs better…[however] many systems are not meeting our goals because of malfunctioning hierarchies…

 To be a highly functional system, hierarchy must balance the welfare, freedoms and responsibilities of the subsystems and total system – there must be enough central control to achieve co-ordination towards the large-system goal, and enough autonomy to keep all subsystems flourishing, functioning, and self-organising.” (Meadows)