34. What would Warren Weaver say now?

Warren Weaver was a remarkable man. He was a distinguished mathematician and statistician. He made important early contributions on the possibility of the machine translation of languages. He was a fine writer who recognised the importance of Shannon’s work on communications and the measurement of information and he worked with Shannon to co-author ‘The mathematical theory of communication’. But perhaps above all, he was a highly significant science administrator. For almost 30 years, from 1932, he worked in senior positions for the Rockefeller Foundation, latterly as Vice-president. I guess he had quite a lot of money to spend. From his earliest days with the Foundation, he evolved a strategy which was potentially a game-changer, or at the very least, seriously prescient: he switched his funding priorities from the physical sciences to the biological. In 1948, he published a famous paper in The American Scientist that justified this – maybe with an element of post hoc rationalisation – on the basis of three types of problem (or three types of system – according to taste): simple, of disorganised complexity and of organised complexity. Simple systems have a relatively small number of entities; complex systems have a very large number. The entities in the systems of disorganised complexity interact only weakly; those of organised complexity have entities that interact strongly. In the broadest terms – my language not his – Newton had solved the problems of simple systems and Boltzmann those of disorganised complexity. The biggest research challenges, he argued, were those of systems of organised complexity and more of these were to be found in the biological sciences than the physical. How right he was and it has only been after some decades that ‘complexity science’ has come of age – and become fashionable. (I was happy to re-badge myself as a complexity scientist which may have helped me to secure a rather large research grant.)

There is famous management scientist, no longer alive, called Peter Drucker. Such was his fame that a book was published confronting various business challenges with the title: ‘What would Peter Drucker say now?’. Since to my knowledge, no one has updated Warren Weaver’s analysis, I am tempted to pose the question ‘What would Warren Weaver say now?’. I have used his analysis for some years to argue for more research on urban dynamics – recognising cities as systems of organised complexity. But let’s explore the harder question: given that we understand urban organised complexity – though we haven’t progressed a huge distance with the research challenge – if Warren Weaver was alive now and could invest in research on cities, could we imagine what he might say to us. What could the next game changer be? I will argue it for ‘cities’ but I suspect, mutatis mutandis, the argument could be developed for other fields. Let’s start by exploring where we stand against the original Weaver argument.

We can probably say a lot about the ‘simple’ dimension. Computer visualisation for example can generate detailed maps on demand which can provide excellent overviews of urban challenges. We have done pretty well on the systems of disorganised complexity in areas like transport, retail and beyond. This has been done in an explicit Boltzmann-like way with entropy maximising models but also with various alternatives – from random utility models via microsimulation to agent-based modelling (ABM). We have made a start on understanding the slow dynamics with a variety of differential and difference equations, some with roots in the Lotka-Volterra models, some connected to Turing’s model of morphogenesis. What kinds of marks would Weaver give us? Pretty good on the first two: making good use of dramatically increased computing power and associated software development. I think on the disorganised complexity systems, when he saw that we have competing models for representing the same system, he would tell us to get that sorted out: either decide which is best and/or work out the extent to which they are equivalent or not at some underlying level. He might add one big caveat: we have not applied this science systematically and we have missed opportunities to use it to help tackle major urban challenges. On urban dynamics and organised complexity, we would probably get marks for making a goodish start but with a recommendation to invest a lot more.

So we still have a lot to do – but where do we look for the game changers? Serious application of the science – equilibrium and dynamics – to the major urban challenges could be a game changer. A full development of the dynamics would open up the possibility of ‘genetic planning’ by analogy with ‘genetic medicine’. But for the really new, I think we have to look to rapidly evolving technology. I would single out two examples, and there may be many more. The first is in one sense already old hat: big data. However, I want to argue that if it can be combined with hi-speed analytics, this could be a game changer. The second is something which is entirely new to me and may not be well known in the urban sciences: block chains. A block is some kind of set of accounts at a node. A block chain is made up of linked nodes – a network. There is much more to it and it is being presented as a disruptive technology that will transform the financial world (with many job losses?). If you google it, you will find out that it is almost wholly illustrated by the bitcoin system. A challenge is to work out how it could transform urban analytics and planning.

I have left serious loose ends which I won’t be able to tie up but which I will begin to elaborate the challenges with four further blog posts in coming weeks: (1) Competing models; (2) Big data and hi-speed analytics; (3) Block chains in urban analysis; (4) Applying the science to urban challenges.

Alan Wilson

March 2016

33: On writing

Research has to be ‘written up’. To some, writing comes easily – though I suspect this is on the basis of learning through experience. To many, especially research students at the time of thesis writing, it seems like a mountain to be climbed. There are difficulties of getting started, there are difficulties of keeping going! An overheard conversation in the Centre where I work was reported to me by a third party: “Why don’t you try Alan’s 500 words a day routine?” The advice I had been giving to one student – not a party to this conversation – was obviously being passed around. So let’s try that as a starting point. 500 words doesn’t feel mountainous. If you write 500 words a day, 5 days a week, 4 weeks a month, 10 months a year, you will write 100,000 words: a thesis, or a long book, or a shorter book and four papers. It is the routine of writing that achieves this so the next question is: how to achieve this routine? This arithmetic, of course, refers to the finished product and this needs preparation. In particular, it needs a good and detailed outline. If this can be achieved, it also avoids the argument that ‘I can only write if I have a whole day or a whole week’: the 500 words can be written in an hour or two first thing in the morning, it can be sketched on a train journey. In other words, in bits of time rather than the large chunks that are never available in practice.

The next questions beyond establishing a routine are: what to write and how to write? On the first, content is key: you must have something interesting to say; on the second what is most important is clarity of expression, which is actually clarity of thought. How you do it is for your own voice and that, combined with clarity, will produce your own style. I can offer one tip on how to achieve clarity of expression: become a journal editor. I was very lucky that early in my career I became first Assistant Editor of Transportation Research (later TR B) and then Editor of Environment and Planning (later EP A). As an editor you often find yourself in a position of thinking ‘There is a really good idea here but the writing is awful – it doesn’t come through’. This can send you back to the author with suggestions for rewriting, though in extreme cases, if the paper is important, you do the rewriting yourself. This process made me realise that my own writing was far from the first rank and I began to edit it as though I was a journal editor. I improved. So the moral can perhaps be stated more broadly: read your own writing as through an editor’s eyes – the editor asking ‘What is this person trying to say?’.

The content, in my experience, accumulates over time and there are aids to this. First, always carry a notebook! Second, always have a scratch pad next to you as you write to jot down additional ideas that have to be squeezed in. The ‘how to’ is then a matter of having a good structure. What are the important headings? There may be a need for a cultural shift here. Writing at school is about writing essays and it is often the case that a basic principle is laid down which states: ‘no headings’. I guess this is meant to support good writing so that the structure of the essay and the meaning can be conveyed without headings. I think this is nonsense – though if, say a magazine demands this, then you can delete the headings before submission! This is a battle I am always prepared to fight. In the days when I had tutorial groups, I always encouraged the use of headings. One group refused point blank to do this on the basis of their school principle. I did some homework and for the following week, I brought in a book of George Orwell’s essays, many of which had headings. I argued that if George Orwell could do it, so could everybody, and I more or less won.

The headings are the basis of the outline of what is to be written. I would now go further and argue that clarity, especially in academic writing, demands subheadings and sub-subheadings – a hierarchy in fact. This is now reinforced by the common use of Powerpoint for presentations. This is a form of structured writing and Powerpoint bullets, with sequences of indents, are hierarchical – so we are now all more likely to be brought up with this way of thinking. Indeed, I once had a sequence of around 200 Powerpoint slides for a lecture course. I produced a short book by using this as my outline. I converted the slides to Word, and then I converted the now bullet-less text to prose.

I am a big fan of numbered and hierarchical outlines: 1, 1.1, 1.1.1, 1.1.2, 1.2,…..2, 2.1, 2.1.1, 2.1.2, etc. This is an incredibly powerful tool. At the top level are say 6 main headings, then maybe six subheadings and so on. The structure will change as the writing evolves – a main heading disappears and another one appears. This is so powerful, I became curious about who invented it and resorted to google. There is no clear answer, and indeed it says something about the contemporary age that most of the references offer advice on how to use this system in Microsoft Word! However, I suspect the origins are probably in Dewey’s Library classification system – still in use – in effect a classification of knowledge. Google ‘Dewey’s decimal classification’ to find its Nineteenth Century history.

There are refinements to be offered on the ‘What to ….’ and ‘How to ….’ questions. What genre: an academic paper, a book – a text book? – a paper intended to influence policy, written for politicians or civil servants? In part, this can be formulated as ‘be clear about your audience’. One academic audience can be assumed to be familiar with your technical language; another may be one that you are trying to draw into an interdisciplinary project and might need more explanation. A policy audience probably has no interest in the technicalities but would like to be assured that they are receiving real ‘evidence’.

What next? Start writing, experiment; above all, always have something on the go – a chapter, a paper or a blog piece. Jot down new outlines in that notebook. As Mr Selfridge said, ‘There’s no fun like work!’ Think of writing as fun. It can be very rewarding – when it’s finished!!

Alan Wilson

December 2015

32: DNA

The idea of ‘DNA’ has become a commonplace metaphor. The real DNA is the genetic code that underpins the development of organisms. I find the idea useful in thinking about the development of – the evolution of – cities. This development depends very obviously on ‘what is there already’ – in technical terms, we can think of that as the ‘initial conditions’ for the next stage. We can then make an important distinction between what can change quickly – the pattern of a journey to work for instance – and what changes only slowly – the pattern of buildings or a road network. It is the slowly changing stuff that represents urban DNA. Again, in technical terms, it is the difference between the fast dynamics and the slow dynamics. The distinction is between the underlying structure and the activities that can be carried out on that structure.

It also connects to the complexity science picture of urban evolution and particularly the idea of path dependence. How a system evolves depends on the initial conditions. Path dependence is a series of initial conditions. We can then add that if there are nonlinear relations involved – scale economies for example – then the theory shows us the likelihood of phase changes – abrupt changes in structure. The evolution of supermarkets is one example of this; gentrification is another.

This offers another insight: future development is constrained by the initial conditions. We can therefore ask the question: what futures are possible – given plans and investment – from a given starting point? This is particularly important if we want to steer the system of interest towards a desirable outcome, or away from an undesirable one – also, a tricky one this, taking account of possible phase changes. This then raises the possibility that we can change the DNA: we can invest in such a way as to generate new development paths. This would be the planning equivalent of genetic medicine – ‘genetic planning’. There is a related and important discovery from examining retail dynamics from this perspective. Suppose there is a planned investment in a new retail centre at a particular location. This constitutes an addition to the DNA. The dynamics then shows that this investment has to exceed a certain critical size for it to succeed. If this calculation could be done for real-life examples (as distinct from proof-of-concept research explorations) then this would be incredibly valuable in planning contexts. Intuition suggests that a real life example might be the initial investment in Canary Wharf in London: that proved big enough in the end to pull with it a tremendous amount of further investment. The same thing may be happening with the Cross Rail investment in London – around stations such as Farringdon.

The ‘structure vs activities’ distinction may be important in other contexts as well. It has always seemed to me in a management context that it is worth distinguishing between ‘maintenance’ and ‘development’, and keeping these separate – that is, between keeping the organisation running as it is, and planning the investment that will shape its future.

The DNA idea can be part of our intuitive intellectual toolkit, and can then function more formally and technically in dynamic modelling. The core insight is worth having!!

Alan Wilson

December 2015

31: Too many journals, too many conferences?

Journals and conferences are certainly proliferating. I receive e-mails weekly inviting me to submit papers to new journals and daily inviting me to sign up for conferences. These are virtually all related to commercial profit-making enterprises rather than from, say, learned societies. (Though some learned societies make a large slice of their income from journals.) Inevitably this leads to a pecking order of journals which are long-established and those with high impact factors being important and supporting the idea of being a ‘top’ journal. I have some experience, of long ago, of being part of launching new journals, both from commercial publishers. I was the Assistant Editor of Transportation Research when it was first published in 1967 and Editor of Environment and Planning for its launch in 1969. They have grown, clearly being successful in their markets and meeting a demand. Transportation research now has Parts a, b, c, d and f and Environment and Planning has A, B, C and D. The 1969 volume of E and P had 2 issues, the 1970 volume, 4. Even E and P A now has 12 issue per annual volume. These are each journals which cover a broad field. Perhaps many of the new competitors are more niche oriented, perhaps not. Many of the long-established and prestigious journals continue to grow through addition of specialist marques – notably Nature for instance. Part of the growth is fuelled by the opportunities for on-line journals, and then related to this, to offering open access – something which is now required by research councils for example. The former offers low entry costs for the publisher and the latter a business model for which the author pays rather than the subscriber – though there may still be print editions.

Are there too many journals? At best, the current landscape is confusing for academic contributors. The new journals have to be filled and notwithstanding peer review, it must be tempting to publish papers which are not adding much to the corpus of knowledge. Indeed, a test of this is the very high proportion of published papers that end up with zero or maybe one citation. And where do editors find the army of referees that are needed? Of course, one reason for the increase is the growth of universities and the corresponding growth in academic staff. Most of these want to publish – for both good reasons and also – still good I guess – as a basis for promotion. So perhaps the number of journals matches this growth, and perhaps the pool of potential referees grows proportionately. Intuition suggests otherwise, but this would make a good research project for information scientists. Maybe it has been done?

This is still not answering the question: are there too many? In one sense, perhaps not – if this number of opportunities is necessary to support the publishing ambitions of the research community. But this leaves the problem of sorting out the real value but perhaps crowd sourcing solves this: some papers will rise to the top of the pile via citation counts. I thought when I started writing this piece that the answer to the question that I pose for myself would be that there are too many – but now I feel more neutral. Perhaps the market should be left to sort things out. We don’t have any other option in practice anyway!!

But what about conferences? Learned Society conferences of long-standing provide promotional opportunities, especially for early-career staff. There are problems: the societies need fee-paying attendees and most staff can only claim expenses if they can say that they have had a paper accepted for a conference. This leads to huge numbers of papers and many parallel sessions, so that even in large conferences, the attendance in a parallel session may be in single figures. And of course, the primary function of conferences is probably ‘networking’ – so even when there are many parallel sessions in play, the coffee shops and bars will be full of truants chatting! Conference organisers might be advised to allocate more time for this rather than cramming the programme.

So again, we probably have to say: let the market sort things out – but with some riders – some things could be improved. a colleague of mine banned her young researchers from presenting papers at conferences that had already been presented at previous conferences. Perhaps conference organisers could take up the policies of many journal editors by asking for confirmation that a paper has not been previously presented or published. But of course there would be ways of steering round that. The big quality test for a conference – beyond the location, the networking and the facilities – should be: what is being announced that is really new. The final confirmation of the Higgs boson for example!

Alan Wilson

December 2015