Values and systems again
When BAA were building Terminal 5, they used the contracting process to signal that they were interested in the system, not its components. They appointed main contractors for the engineering aspects of the work that you would expect to see. They got quotes from all the contractors for how much the work would cost. And then they said that the total cost of those quotes was 150% of what they were going to pay to construct the terminal.
This is not a cost cutting measure.[1] What they asked the main contractors to do was to work out how to save cost by working more effectively together and by innovating the construction process. This is a quality consideration because taking work out of the process is almost certain to lead to a better product. The contractors were paid some retainer fees to take as long as it took to find a way forwards.
The contractors loathed this arrangement. They squealed. They used every trick in the book to get paid for doing some engineering work ahead of the contractual condition. The teams at this stage were co-located in offices at Heathrow and the staff had to live daily their challenge to have meaningful discussions in order to progress. It is legendary how little different types of contractor understand each other’s work practices and needs. It was hilarious.
There was nothing wrong with each contractors’ proposals and quotes. These were leading firms and they were not stupid or on the make. But parts do not add up to a whole in that way. It is precisely in the relationship between the parts that the system exists: its quality and its cost live there, not in the spreadsheet or the project plan. All credit to BAA for nailing this and sweating it out.
Asking people to reduce their price is not at all the same thing as insisting that there is a better way to do things that will therefore cost less: in fact, they are some sort of opposite in their effect.
Urban design
Let’s look at the same issue with Christopher Alexander’s New Theory of Urban Design. In this design exercise an urban area is re-planned, rebuilt in theory, from scratch. The process is an iterative architectural competition with a set of simple explicit values to judge entries against.
It starts by placing some landmark public building, like a theatre. The next building has to work in its own right, to enhance the previous building, and to create good public space. The subsequent steps iterate this process to infill the area with buildings that keep enhancing what is already there.
There is no such thing in this process as a building that is successful, in its own right, in isolation. There can be no “great building” without consideration of context. That context is extremely multidimensional, has a huge number of factors that may be relevant. That is why a series of competitions to bring new ideas and different imagination of what the area could be like makes sense.
The stricture to enhance what others have already done is like the contract that the Terminal 5 contractors hated. People want to shine on their own terms. They want the value of good engineering to be an absolute, to be external to the context. These processes insist that only the situated specific can demonstrate excellence, the rest is hot air.
Just think of all the marketing people who want you to believe that they have just what you need. It cannot be true; it is an impossibility that the context for what you need might be part of their design world. They can get you to want their product, but they cannot meet your need.
Journeys and destinations
These two examples have in common that the destination is not specified, except in terms of abstract values. They embark on a journey that will take them to a much better place than they are in now, but they don’t know concretely what that place is or what it looks like. Alexander’s paradigmatic case is an Italian village that has evolved over the centuries to its current feeling of being just right.
Notice that this is not contract-friendly. BAA had to come up with a cost threshold because they could not specify what they actually wanted the contractors to do. They required a journey with a cut-off point at which the journey was deemed good enough. Alexander’s architecture competition is similarly devoid of a specified endpoint, but it requires extensive iteration where the quality of a what exists in all those dimensions is improved in a progressive fashion.
When we have an issue to address, we would do well not to get pushed into premature specification of what it is we “need” because we probably don’t know. But I want to give two counter-examples:
I knew some consultants part of whose business was to help arbitrate contractual disputes involving software. As I remember an old story, there was a medical practice that were having some administrative software built. They sued their suppliers saying that they had not provided what was agreed. It turned out there was no written specification at all.
I have seen several groups of developers whose definition of how a system is supposed to work is the current system itself. And the famous second-system effect, in which the new system is also supposed to remedy all the unstated defects and aspirations of the original. All too often, the behaviour of artificial intelligence systems is taken to be intelligent!
So, if we are going on a journey with others, we need to have some reason to believe, or some evidence, that the journey will produce the value we seek. Those marketing people will blandish, and they are persuasive by definition. Remember that none of the T5 contractors came to BAA saying they would work with other contractors to improve the process.
Levels of system
So here comes the difficult thought move. In the Italian village, no-one was in charge. No-one got select what might make a place great to live in: it happened by a process of repeated discovery of what works and therefore persists. In natural ecosystems we don’t even have the notion that we can tell a good one from a poor one.
Systems nest within each other, and there are important reasons for thinking that often it is the enclosing system that makes a nested system a system in the first place. Many of these issues are down to understanding what we are doing, and how we affect things, when we observe. No such thing as a neutral bystander or an objective scientist. Not ever.
Sometimes we need to step up and take responsibility for the effect our system has on its enclosing system (and on the systems it is comprised of). Sometimes, as with T5, we need to very clear about the sort of thing we want even if we can’t see the detail yet. And sometimes we need to be in humble awe and recognise that the ecosystems we work within have properties that are lifegiving and yet which we do not understand.
And we need to do the other difficult thing, which is to distinguish the purpose and values of the different level of system. I usually quote the value of black teams in a software development effort. A black team has a purpose and set of values around showing that the software system does not work; it employs people who are good at trashing systems. (One of my own dubious talents is to be able to fail to understand any user interface going.) But the overall software development can value the improvement in quality to be had by efficient wrecking of this sort. There is no need for the black team to buy into their bigger role, in fact it would probably work against the larger system.[2]
The most barren long conversations I have had around systems are with people who cannot recognise this systems-within-systems view and how important diversity of purpose, values and outlook can be. The people who reject this view have, in my experience, an overarching view that if they were in charge, they could order things better. Which of course is only true until we meet the condition that they do not understand and for which it is not true.
Scaling
Is a system that is ten times the scale of the current system the same system, only bigger? Theoretically this is highly unlikely: system properties depend on connections between their components so if there are more components there will be more, and structurally different, connections. Very often if you are asked whether something will scale it is a covert way of asking whether your system has enough aspects of command and control to believe that scale can be commanded. But command and control is not systemic, ever. No component of a system can control the system, especially not the boss.
Many systems of interest to us are human social systems: they work because people communicate with each other at a deep level and because they empathise. Social systems of this nature are mind-bogglingly complex because of the incredibly variety and dimensionality of those communication links. Remember that if you meet an angel you will not know that it is an angel because it is so superior to your intelligence: it takes a special gift to recognise an angel. The behaviour of social systems is similarly beyond our imagination, which is why marketing people need to debase it and isolate people from each other before they can get any control. Not for nothing is neoliberalism individualistic.
To return to Nora Bateson’s symmathesy and warm data labs, warm data is precisely the impossibly complex trace of real social processes. The attempt to even describe the behaviour of real social systems without this complexity ends of being a mirror of the prejudices and limitations of the describer. Large-scale and scalable systems are mostly the fantasies of large-scale egos. Prime Minister Johnson would be a classic example: he will never be able to comprehend the effects of his own fantasies.
Large problems
All of which leads to a fascinating systems problem. If we take a large-scale problem like climate change, like global political corruption, like our current pandemic, how are we to think about them?
We can easily list some thoughts that are not appropriate approaches, that are just the reflection of someone’s inflated ego. Geoengineering is not a solution to climate change: it is embedded in the systems that caused the problem. Herd immunity is not a solution to the pandemic. More interesting is to consider whether representative “democracy” is capable of re-establishing governments that work for the people who elected them.
We need to work between Scylla and Charybdis as ever. If we say that only global initiatives can tackle global problems, we will fall down the hole of thinking about scalability. If we think that it is the change of behaviour of individuals that will make a difference, we will get sidelined by neoliberal values and by marketing. The current wave of vegan propaganda is a beautiful example of how these two extremes work together to keep people enslaved.
The myopia in looking at social systems is failing to recognise what they are part of which is ultimately more fundamental than they are. We need to be in touch with humility and awe at the way the biosystem supports us despite all the damage we do. There is a touchstone of seeing something work robustly and recover given half a chance that we all need to observe and wonder at. I highly recommend Andreas Weber’s Biopoetics as a route into this healing view.
The systems that keep us alive are more subtle and complex than anyone can really get to grips with: Johnson stands no chance. That is not just abuse of Johnson: it says that all political decisions will be fatally compromised until that gap is bridged.
[1] Yes, it’d save cost, but contrast it with the ‘haircut’ type of cost-cutting where you remove the same percentage of budget from everything, without considering which elements are critical or seeking coordination efficiencies.
[2] Note that this is intelligent probing and wrecking, in contrast to the famous Netflix chaos monkey that indiscriminately crashes servers and severs connections. As a subscriber I can confirm that the overall system frequently loses the context of what we’ve been watching, for all that the streaming capability has been made robust.
