Keeping models in their place

The increasing use of algorithmic decision making raises some challenging questions, from bias due to societal biases being baked in to training data, to the loss of the space for compromise (due to the need to codify a loss or reward function in a machine learning system) that is so important in addressing conflicting aims and values in democratic societies. The broader role of models as a means of both understanding and shaping societys is one of the themes of my most recent book, Cogs and Monsters, in the domain of economics. In particular, I wanted to expose in economic modelling the reflexivity involved in being a member of a society analysing the society in order to try to change the society – when its other members may well alter (in often-unanticipated ways) the behaviour that was analysed – because they are subjects, not objects.

Well, all of this is the subject of Erica Thompson’s excellent book Escape from Model Land: how matehmatical models lead us astray and what we can do about it. It focuses on the use of algorithmic models, and has a wide range of examples, from health and epidemiological modelling to climate projections to financial markets. It covers as well as the reflexivity some familiar challenges such as performativity, non-linear dynamics, complex systems, structural breaks and risk vs ‘radical’ uncertainty. The ultimate conclusion is the need to be duly humble about what models can achieve. Alas, people – ‘experts’ – seem to all too often get caught up in the excitement about technical possibilities without the thoughtfulness needed to make decisions that will affect people’s lives in important ways.

So this is a much-needed and welcome book, and I’m looking forward to taking part in an event with the author at the LSE in January.



The joy of messiness

It’s hard to imagine that Tim Harford ever writes a duff word. His new book, Messy, is a pleasure to read. (It’s published in a couple of weeks but The Economist jumped the gun with a review so I’m going to go ahead with mine now.) Messy is closer in spirit to Tim’s Adapt rather than his Undercover Economist or Undercover Economist Strikes Back. It gathers research and stories from several different fields to weave them into the theme of the title, and make the case for appreciating mess and avoiding tidy-mindedness.

For example, early on the book tells the tale of Robert Propst, an engineer (and much else) who created a design for modular office furniture and partitions for the Herman Miller company – working 150 miles away from headquarters. But his innovative design empowering workers to create their own workspaces was subverted by managers who insisted the partitions stick to 90 degree angles to make regimented lines. The cubicle farm was born. “Propst was left to condemn the peversion of his ideas as ‘monolithic insanity’, ‘hellholes’, ‘egg-carton geometry’ and ‘barren, rathole places’. The account is in a chapter about workplaces, where mess is a sign of individual empowerment and creativity, contrasted with tidy centralisation and control. (Tim has an extract on tidy vs messy desks in the FT Magazine.)

There are many other accounts of mavericks and independent-minded innovators. The book gives examples from military campaigns, creative contexts (such as Bowie’s reinvention of his music in Berlin), business and nature. I particularly liked the example of the way 18th century scientific foresters in Germany tried to measure the forests using the metric of a Normalbaum, or standardised tree. The idea was to make sense of the messiness of the forest by assigning standard sizes to trees so it would be possible to count how many trees of what dimensions there were, and thus assess the total volume of wood. What happened? “The mess of old forests began to be tidied up. The confusing patchwork of threes of various ages and species was replaced with stands of particular species – the Norway spruce was popular – and of a particular age. The foresters lined up the rows to make the forests easier to survey, to police, and in due course to harvest. Dead trees were felled, rotting hulks dragged away, underbrush cleared. The Normalbaum, once a statistically convenient idealisation of a tree, took physical form.” In the short term this was profitable. In the longer term, it destroyed the ecology of the forests. Yields declined, and the Germans by 1968 had a term for it: Waldsterben or forest death syndrome.It’s a terrific, malign example of performativity.

This chapter, called Incentives, has a number of very nice examples of the perverse effects of targetting, with echoes of the manic modernism described in James Scott’s Seeing Like A State, and of the phenomenon known to economics as risk compensation. I’m not sure I agree with the prescription make it messy as a sure-fire fix. One example in Messy is the blurring of how urban space is divided between vehicles and pedestrians. The early schemes suggest a messy arrangement is safer: people take greater care because they are uncertain. But we don’t know what will happen over time as everybody gets used to the new arrangements. On the other hand, it’s hard to disagree with the idea that children need to be allowed to play in non-sanitised, non-health-and-safetied spaces if they are to ever learn to cope with risk in life.

The common theme in these chapters, which on the face of it seem to gather together a somewhat disaparate set of examples of non-conformism, is the need to design for unpredictability or risk, but also for the fact that people (and trees, and the rest of the natrual world) respond to actions and constraints. It is so hard for decision-makers in any context to realise that they are not social engineers, somehow looking down on the world and able to manipulate it. I entirely agree with this, so recommend Messy. It’s right, it’s a pleasure to read, and it’s out just in time for Christmas.51nrnvc3xcl-_sx329_bo1204203200_

A philosophical diversion

It was a holiday weekend so I indulged myself in a little philosophy: Patrick Baert’s [amazon_link id=”0745685404″ target=”_blank” ]The Existentialist Moment: The Rise of Sartre as a Public Intellectual[/amazon_link]. Regular perusers of this blog will know I recently outed myself as a teenage existentialist, in reviewing Sarah Bakewell’s excellent new book [amazon_link id=”B017IGPTDQ” target=”_blank” ]At The Existentialist Cafe[/amazon_link]. Bakewell explains (and critiques) the philosophy, and sets it in the context of wider philosophical currents. Baert explores a few years in French history, those of the German occupation during World War 2 and the immediate post-war years, to explain why existentialism and why Sartre in particular struck a chord with the public and become so influential.

[amazon_image id=”0745685404″ link=”true” target=”_blank” size=”medium” ]The Existentialist Moment: The Rise of Sartre as a Public Intellectual[/amazon_image]

It is fascinating in its exploration of why writing came to be seen as so central to French national identity and why Sartre was particularly adept at using writing about politics to appeal to the French public at that time of national defeat and subsequent rebuilding. So, oddly, although Sartre and De Gaulle were poles apart politically, both played an important part in rebuilding the nation’s sense of cohesion and dignity.

Baert’s final chapter has some general reflections on the role of public intellectuals and writing as a performative political act. He argues that the generalist public intellectual of Sartre’s type cannot exist in modern social contexts, but have been replaced instead by public intellectuals with expert domain knowledge. Sartre wrote about social and economic issues with no knowledge of the facts or the social science, and nobody would get away with that now. I’m not sure I buy the argument about the perfomative character of people who pontificate about the economy, at least not in a straightforward way, but having said that, there’s some appeal (to a writer) in the idea that words are sufficiently powerful to shape social reality. Man the keyboards!

Measuring and markets

I’ve been reading Michel Callon’s introduction to the edited volume [amazon_link id=”0631206086″ target=”_blank” ]Laws of Markets.[/amazon_link] It’s about the performativity of economics, a question that interests me (although I do struggle with the academic jargon of sociology; at least my own subject’s jargon is familiar). Callon writes: “The most interesting element is to be found in the relationship between what is to be measured and the tools used to measure it. The latter do not merely record a reality independent of themselves; they contribute powerfully to shaping, simply by measuring it, the reality that they measure.”

[amazon_image id=”0631206086″ link=”true” target=”_blank” size=”medium” ]Laws of Markets (Sociological Review Monographs)[/amazon_image]

Needless to say, the question of how the classification and structures embedded in economic statistics shape the reality of the economy (through affecting understanding, behaviour and policy) is of keen interest to me. For instance, part of the debate about productivity is about what it measures, but also partly about what it defines. What is productivity when products play a minority role in economic activity? The Callon intro doesn’t ultimately enlighten: it seems to me to place too much weight on economics as a subject, for markets existed long before economists did. There has to be some two-way influence between reality and the attempt to make systematic a description of it. In fact, I don’t think economics is as different from some other subjects as the performativity analyses suggest. For instance, classification in biology is not completely dissimilar. I also wish other social scientists would acknowledge that economists *do* think a lot about the specifics of markets as social institutions – see, for one, John McMillan’s brilliant book [amazon_link id=”0393323714″ target=”_blank” ]Reinventing the Bazaar.[/amazon_link]

[amazon_image id=”0393323714″ link=”true” target=”_blank” size=”medium” ]Reinventing the Bazaar: A Natural History of Markets[/amazon_image]


Still, there is something in this territory. It’s particularly important for sustainability that the concepts and measurements economists define and gather place ‘the economy’ in nature and the physical world. To be continued…

Statistics in a disordered world

I’m writing about GDP in particular and economic statistics in general again – can’t keep off the subject (& by the way, [amazon_link id=”0691169853″ target=”_blank” ]GDP[/amazon_link] is out now in paperback!) Today I picked up Adam Tooze’s marvellous 2001 book [amazon_link id=”0521039126″ target=”_blank” ]Statistics and the German State 1900-1945[/amazon_link]: “This book … has sought to portray the construction of a modern system of economic statistics as a complex and contested process of social engineering … A functioning statistical system … implied a particular model of political order and in particular a vision of the relationship between state and civil society.”

The national accounts framework in place today is a modernist project. Like so many of these, it is being unravelled in unpredictable ways by technology, globalisation, and the changing character of the state. My writing task today is responding to the call for evidence on the current Review of Economic Statistics. It’s a tall order to say something succinct about getting from the categorisation of the world coded into current statistics to something closer to (disordered) realities, especially when there is an important element of performativity in statistical categories.

[amazon_image id=”0521039126″ link=”true” target=”_blank” size=”medium” ]Statistics and the German State, 1900-1945: The Making of Modern Economic Knowledge (Cambridge Studies in Modern Economic History)[/amazon_image]   [amazon_image id=”0691169853″ link=”true” target=”_blank” size=”medium” ]GDP: A Brief but Affectionate History[/amazon_image]