Development, Epidemiology, China; Course Reviews

I’ve taken a couple of these online courses now, and seen my wife take a few more, and its definitely been interesting and worthwhile.

PH207x: Health in Numbers: Quantitative Methods in Clinical & Public Health Research from edX Harvard was the most technical, the most rigorously graded, and generally the most recognizably similar to the undergraduate courses of my science degree. The organizers arranged for all enrolled students to have time-limited copies of Stata, the commercial stats processing package, and work in Stata made up a lot of the assessment. The student body had a lot of expertise to share, either from a medical or mathematical background. The lecturers and TAs were also well engaged, with responses to student questions and discussion being incorporated back into the material.

14.73x The Challenges of Global Poverty was an interesting lecture series by two genuine academic superstars of development economics, Abhijit Banerjee and Esther Duflo at MIT. The course is more or less an extension of their book Poor Economics, which advocates systematic use of randomised controlled trials in economics. It shares lessons from the research for the worlds very poor across issues like food, jobs, risk and access to even simple financial services. It’s a great book, and I wouldn’t have read it so thoroughly, let alone worked through the extra lecture material, without the structure of a course to fit it around. I’m glad I did work through it, as it included things like an excellent walkthrough of the famous Acemoglu et al paper The Comparative Origins of Colonial Development.
There was an electronic copy of Poor Economics available as part of the course, the reader was a little clunky and didn’t work on mobile devices, but given it was all free complaining seemed churlish. A heartening number of the key papers they mentioned are also freely downloadable. Assessment in this course was multi-choice usual based on a fairly strict reading of the course material. This could be frustrating for certain questions; the reputation of two-handed economists has a truth to it, and it could be seen here as hedging in the questions with qualifiers like might, could or should, as well as in the answers with multi-choice. This works in a short answer or essay question, where the student has a chance to explain their reasoning, but not multi-choice.
This is a translation of an existing undergraduate course, but I got the feeling the assessment was dumbed down: if I were paying for an education at MIT I would expect more too. There was engagement from staff but it didn’t seem to be a huge focus. The huge, global and very diverse student body seemed to overwhelm the usability of the edX discussion boards, which had moments of interest but were mostly dominated by typical Internet discussion dross.
Marie Hicks, far from a thoughtless cheerleader of the medium, has suggested MOOCs can be a “new way that we get our research out into the wild, taken seriously, and used as part of larger intellectual, social, and economic debates”. Banerjee and Duflo have seized on that very successfully. Rehabilitating the lecture as a piece of public performance and education is a heartening feature of the early twentieth century, and a course like this gives rather meatier content than the aperitifs at TED.

STSCHINA-001 Science, Technology and Society in China I sits slap across the middle of a swathe of my interests. Naubahar Sharif does a rapid fire tour of philosophy of science and engineering and history of science in China before discussing innovation systems in more detail. A lot of the MOOC discussion is resolutely US-centric, but this is run out of the Hong Kong University of Science and Technology. It’s also a short course format – it was only three weeks long, with two followup courses forming a trilogy about the normal semester length. I found this length easier to fit around other ongoing commitments. There are excellent reasons for regional studies to be pursued outside the focus region, but I do wonder how, eg, the chance to learn about China from teachers in China will change the field.
This course also had the challenge of examining humanities material in an online and massive student body. They chose peer graded short essays, two of which I slapped up on this blog. I thought this was a good compromise given the premises of MOOCs, and good on them for trying a different format. I didn’t get much engaging discussion out of the forums for this course, but the assignment marking process did let me see other angles on the topic. A lot of people seemed to complain about it though – it might have been a double shock to someone from a non English speaking background and used to multi choice or short problem questions.

There’s lots of things to speculate about around online courses, their open spirit, the dot-com venture capital hype and greed, the institutions about to truck crash, the young and not so young postdocs already hammered by the structure of the academic labour market, and I’d like to ruminate on all that too, in another post. From a purely personal and selfish perspective, though, these were courses I couldn’t do locally, on fascinating and important ideas, that pull material together in a way a teacher can and a book usually doesn’t. +1 would experience birth of new pedagogical genre again.

Big Powerful New Data

Power and language are both crucial currents for innovation. Two alternative tools for macro analysis of an economy’s innovative capacity and output then suggest themselves. Firstly, a power-centric analysis of changes in the economy, from in physical and political senses. Secondly, linguistic analysis of mass printed and digital material produced in an economy, from standalone and comparative perspectives. These techniques can complement one another, given that shifts in power and language also interact. Power-centric analysis of technology is a technique introduced by Russell et al in “The Nature of Power: Synthesizing the History of Technology and Environmental History”. An example of linguistic analysis of the economy is the R-word index run by The Economist, where the frequency of the word recession is used as an indicator of recession.

In a power analysis of the economy, energy flows and transitions are modelled qualitatively or quantitatively. Using this lens, we may note that the rise of the Internet has accompanied surging electrical power needs in large relatively centralized datacentres, with cloud computing being the current extreme of this. At the same time it has disintermediated middlemen such as travel agents. The move of labour – and spending of biochemical energy – from a travel agent in an office to the consumer at home or on a smartphone in turn requires increased electricity requirements for mobile phone towers and households. Given this analysis we can get insight into Google’s investment and research into alternative energy and distributed generation technologies such as solar photovoltaics. We might also note that, globally, the Internet mostly runs on coal. Combining the physical energy analysis with political analysis, we can see where innovation actors are constrained by energy and whether shifts in power are dominated by local or foreign actors, be they wind power entrepeneurs or multinational oil companies.

A focus on physical power can yield quantitative metrics of joules and watts that are not available to more structural approaches such as the system of innovation model. It focuses on facts about the economy that are fairly readily available for most countries, and also in comparative form. Though power analysis does include the labour market and its use of biochemical energy, this focus on economic output may make analysis of innovation capacity relatively indirect. How much did the energy use of a US mathematician change over the twentieth century, except as a consumer of productivity tools, such as computers, available to all professionas? This is a technique pioneered by historians, and it may speak most clearly in retrospect, requiring extrapolations to deduce capacity which are more prone to subjective policy hobby horses.

The linguistic approaches strengths and weaknesses seem to complement power analysis. By focusing on words, it will tend to weight research and development activity more strongly, such as use of terms in journal articles or social media. One weakness of linguistic analysis is that mass corpuses of content must be available to do “big data” style analysis. A developing economy, particularly in the poorest parts of the world, may not produce enough readily available searchable content to discover meaningful shifts and opportunities. Relying on the linguistic approach too heavily in a poor developing country may skew policy too much to theoretical research and ignore useful innovations happening on the ground but not on Twitter.

The innovation systems approach may have a weakness that the initial categories of organization (university, R&D lab, etc) constrain future analysis, missing trends which cut across traditional organizations. In this way both power and linguistic analysis may show up perspectives that do not emerge as readily in the otherwise more comprehensive innovations systems approach, and thereby supplement it.

Invention As A Hub

In a linear model of innovation, innovation is imagined to proceed through an orderly sequence of steps, from pure scientific research, to applied science, formulation as a technology, then developing and scaling up distribution of that technology as a product. One alternative model might be that of a “techno-social hub”. In a techno-social hub model, science, applied science, capital provision, product development and the exchange of products and services in the market are connected to each other through a media of technology and social processes. This can be represented graphically as a techno-social hub node connected by a single edge to nodes representing research, applied research, and so on. These nodes are similar but not identical to the stages in the linear model of innovation.

The techno-social model is an improvement on the linear model, as it distinguishes different factors in innovation without unrealistically segregating those factors. It represents that once a technological or process innovation is made, influence doesn’t flow in a straight line, but feeds back to different parts of society via the artifact or social change. For example, the development and use of the Newcomen steam engine in factories in 18th century Britain opened up the possibility of applied research and prototypes of steam trains by the early 19th and the capital provision required to build railway networks. The steam engine also spurred pure research in thermodynamics and was an influence on the psychological theories of Freud.

Operationally this model recognises the importance of institutions and organisations that support each aspect of innovation, such as universities for basic research and markets for exchange and use. By emphasizing the links between different stages it might direct policy makers and people in the field to the importance of good communications amongst organisations, via physical co-location, libraries, journal publication, less formal collaboration over the Internet, and so on. It recognises that, in William Gibson’s phrase, “the street finds its own use for things”, and that research and capital should be able to dynamically react to new uses of a technology.

A disadvantage of the model may be underemphasizing the links between closely related areas, such as basic and applied research. By placing technology at the centre of the model, it tends to technological determinism. The social aspect of the techno-social may also be too broad a category to effectively operationalise for setting innovation policy. Overall, however, the techno-social hub model avoids the constraints of the linear model at the cost of being slightly harder to say, and draw.