You can download that version (2.0) here. Below is a summary of the main ideas.
It describes a) some fundamental aspects of the world that make it complex and inherently hard to predict (p. 9-21), and b) some problems with political institutions and the education and training of those who control (and influence) them that compound this inherent complexity (mainly p. 84-94).
For example, very few in Parliament, Whitehall, or the media have any training in statistics. They therefore do not understand terms such as ‘normal variation’ (bell curve) or ‘variance’ which makes it impossible for them to make informed decisions about some things for which they are responsible, or to understand scientific discussions. Issues such as ‘how financial models contributed to the 2008 crisis’ or ‘intelligence and genetics’ cannot be understood in any depth without some basic statistical knowledge.
Courses such as Politics, Philosophy and Economics (and economics in general) do not train political leaders well. They encourage superficial bluffing, misplaced confidence (e.g. many graduates leave with little or no idea about fundamental issues concerning mathematical models of the economy (cf. p.21-5 & 214-229)), and they do not train people to make decisions in complex organisations. Ministers are selected from MPs but MPs are not selected for their ability to devise policy, prioritise, manage complex organisations, or admit and fix errors. In the absence of effective training, many default to gimmicks and attempts to manipulate the media. In the choice between ‘to be’ or ‘to do’, insiders tend to choose the former because the system incentivises behaviour that is contrary to the public interest.
Much thinking and discussion in Westminster is either a) vague ‘dinner party’ speculations about the distant future, or b) gossip about the daily crisis – amazingly little involves concrete operational planning to get from A to Z. Most media commentary on politics overstates the extent to which news derives from ‘plans’ (‘strategy’ being the most abused word) and understates the extent to which news derives from panic driven by chaos exacerbated by lack of operational grip. In Government, there is constant panic but little urgency. (A few footnotes deal with specific issues concerning Westminster. E.g. footnote 191 & 200 (media); 198 & 199 (Westminster dysfunction).) ‘We do not have a problem with “too much cynicism” – we have a problem with too much trust in people and institutions that are not fit to control so much.’
The essay describes why these failings are increasingly dangerous. It sketches some of the most important scientific, technological, economic, and military trends that can both enhance and harm the world (e.g. energy technologies, space science (e.g. hypersonic space planes to reduce dramatically the cost per kg into orbit, the search for ‘earth twins’, experiments with quantum communication from space platforms), genetic engineering and synthetic biology, machine intelligence, digital fabrication, algorithmic trading, drone swarms, cyberwar). See Section 7 for a discussion of various geopolitical trends (particularly p. 125-33). (A brilliant essay (here) by one of the 20th Century’s best mathematicians describes these issues connecting science, technology, decisions, and political institutions.) It also discusses changes to the scientific process itself and prospects for 1) large scale collaborations (such as PolyMath) and 2) better coordination of expert attention, brilliantly described by Michael Nielsen in Reinventing Discovery.
The essay also describes some ways to limit these problems by improving education and training and creating new institutions. It touches on various ways to improve government including: institutionalised Red Teams, ‘information markets’ (already being used by innovative companies), experiments in crowdsourced policy-making, the use of prizes / Grand Challenges as used by DARPA (and which produced the Spitfire in the 1930s), and the integration of scientific advice and robust evaluations of new ideas (to avoid reinventing ‘square wheels’ and wasting money).
Instead of trying to solve problems centrally and manage complex projects, Whitehall ought to reconsider what goals it incentivises centrally while decentralising decisions about methods. Other fields have developed empirical design rules and quantitative models (such as aircraft engineering). Whitehall needs to devise rules that encourage evidence-based policy where feasible and decentralised decision-making as a default mode. It also needs new methods to regulate, monitor, and when necessary intervene in complex systems: e.g. financial markets (high speed ‘algorithmic trading’, involving automated conflict at the microsecond scale, is likely to spark crises, cf. p120, 126. ). This requires the development of artificial immune systems (i.e. systems that produce robust defence via evolved decentralised solutions) and equivalents to products being developed to deal with manufacturing failure such as statistical stress modelling software.
We need (civilian) British versions of DARPA (pursuing ‘high risk, high return’ projects that markets won’t fund (i.e. failure is normal) and funding existing labs rather than setting up its own labs), the Santa Fe Institute, and TALPIOT (all operating outside Whitehall HR and EU procurement rules). From professional sports to the intelligence world, people are developing new training programmes, such as those based on online games: politics and education need to do the same in order to raise performance substantially. One of the most promising experiments is Tetlock’s Good Judgment Project which involves a systematic attempt to improve training of political decision-makers so their predictions are more accurate. It is telling that Tetlock’s groundbreaking research on the accuracy of predictions by ‘political experts’ is ignored in Westminster (cf. p. 84ff). In economics, physicists are devising new approaches to modelling and prediction, such as ‘agent-based models’ (cf. p.118ff and Endnote). (See mainly Section 6, p. 94-102, and footnote 227 for some Westminster-specific ideas about what to do, such as appointing non-MPs as Secretaries of State and scrapping Whitehall’s HR rules to open institutions up to people who have experience of dealing with very complex problems and organisations.)
Overall, the essay is an attempt to sketch some ideas for what the Nobel-winning physicist Murray Gell Mann called an ‘Odyssean’ education – an education that starts with the biggest questions and problems and teaches people to understand connections between them (‘integrative thinking’). Universities need new inter-disciplinary courses. For example, in March 2014 Stanford announced new undergraduate degrees such as Computer Science and English. It would be great if Oxford created alternatives to PPE such as ‘Ancient and Modern History, Maths for Presidents, and Coding‘. Instead of bluffing through essays on competing views of macroeconomics, future leaders should be rigorously trained to understand and apply probabilistic reasoning to problems – and know when the situation is so uncertain there is no useful help from the sciences. New courses could embed students with leaders managing complex projects. Schools need to change their curricula to provide the building blocks for such courses. (This should be an addition to – not a replacement for – traditional subjects and disciplines.)
Part of Section 6 (p. 62-83) sketches some ideas about education policy, some of which lay behind various policies enacted in the Department for Education 2010-14, which could provide a basis for further decentralisation of education so that as little as possible is controlled by Westminster and Whitehall. Both are fundamentally unfit to have the degree of power that they do and even if they were not unfit they still should not control all they do. For example, more money needs to be spent on our research universities and the basic science budget. There must be a determination to be a world-leader in basic science – not just a focus on applications. (The decision in this parliament by BIS to divert funding from pure maths to statistics, on the spurious grounds that the latter is more important for applications, was a mistake. The entire computer industry emerged from work in the 1930s on the logical foundations of mathematics – perhaps the most esoteric and apparently least ‘practical’ branch of mathematics.) The broken PhD pipeline needs repair. There should be far greater effort to make the environment around research Universities friendly to start-up businesses. Section 6 also touches on some projects that I worked on at the Department for Education that could contribute to the overall ‘Odyssean’ goal (e.g. Professor Mark Warner’s Cambridge project to redo the 16-18 physics curriculum; funding MEI to turn the ideas of Fields Medallist Tim Gowers into a ‘Maths for Presidents’ course) though it should go without saying that the merits of such specific projects are entirely separate from the merits of my ideas.
If English state schools are to improve substantially, it will require good Academy chains to integrate: a) more demanding and interesting maths and science curricula (such as the Gowers and Warner projects), including hard skills in modelling and problem-solving, including statistics such as conditional probability (the inclusion of ‘natural frequencies’ in the new GCSE is a step forward); b) more focus on essay writing and serious projects; c) the use of frequent testing to see if pupils are learning, particularly ‘Concept Inventory‘ tests developed and used increasingly (and successfully) in US undergraduate physics courses by the likes of Nobel-winner Carl Wieman, and (soon) in Professor Warner’s Cambridge University physics 16-18 project (these tests show whether students have actually understood fundamental concepts or the teaching was a waste of time); d) teacher hiring, firing, pay, and training; e) a learning feedback loop (idea-experiment-learn-embed-new idea) inside the chain that is connected to a broader external network (including the research community) which allows incremental improvements in performance based on solid information about what works.
One of the most extraordinary aspects of English politics and education is the lack of structured, disciplined thought about what works and doesn’t work, and how to build reliable systems that allow improvement in performance. (Most of those with power in the English education system are much more interested in appearing to be ‘on the side of the poor and less able’ than they are in raising standards, so many policy debates are really just exercises in moral exhibitionism. Oxbridge power structures are much more interested in appeasing political forces than they are in raising standards, and they deliberately suppress the views of their own leading academics in order to avoid political controversy.) Most activity in Whitehall occurs without asking ‘who, somewhere in the world, has already solved this problem?’, and people can be remarkably resentful if one asks this question. A ‘learning feedback loop’ would bring a scientific approach to education that would allow the system overall to deliver high performance even though most of the people involved are (by definition) of roughly average ability. Many of the changes made 2010-14 were aimed at allowing such a ‘learning feedback loop’ to develop. Without it, education will remain a patchy cottage industry dominated by ‘house price rationing’ and avoidable problems.
The essay also suggests a fundamental principle for the reorientation of British policy. It states: ‘After 1945, Dean Acheson quipped that Britain had failed to find a post-imperial role. It is suggested here that this role should focus on making ourselves the leading country for education and science…’
I know from experience of Whitehall (spending rounds, budget discussions with the Treasury etc) that it would be easy* to make significant overall savings (tens of billions), spend a large fraction of those savings on education and science (of course combined with policy changes), and still be able to save taxpayers money overall. Such a change could make us better educated, more prosperous, safer, and better able to lead the world in various fields. It would be a big improvement on the depressing spectacle of watching politicians thrash around with no priorities and fundamentally little idea about what to do other than try to stay a step ahead of the media with badly implemented gimmicks and avoid blame for our institutionalised dysfunction, while tweaking (usually ineffectually) the bureaucracy and trashing the opposition.
My advice to those trying to get things done in Westminster is: focus, ‘know yourself’ (face errors), think operationally, work harder than others, don’t stick to the rules, and ask yourself ‘to be or to do?’.
(* By ‘easy’, I mean that the task is not intellectually hard nor would it require very rare management skills: the scale of waste remains vast, partly because Whitehall works on hidden dodgy accountancy. (E.g. MOD’s procurement is disastrous (the aircraft carrier project remains shambolic). ‘High speed rail’ is a huge waste of money. Welfare reform has barely scratched the surface of waste and fraud. Energy policy wastes large amounts that could be spent on basic research instead. There remains a lot of low hanging fruit in Whitehall savings. Negotiation of contracts is appalling – e.g. officials consciously enforce EU Frameworks that they know in advance will prevent them from withholding payment in the event of failure.) The main difficulty arises from the lack of suitable people in positions of power and the need for overturning existing power bases, which is always politically tricky. The most likely way it will happen is, as usual, by the exploitation of crises. Of course there is also a feedback loop: the institutionalised dysfunction of the state makes it hard for any political force to make significant changes – and the more the priority of insiders is to remain insiders (‘to be’, not ‘to do’), the harder it is to make changes.)
Footnote: genes and IQ
Only a tiny fraction of the essay dealt with genetics and IQ though the media focused on that.
There were various responses. One came from Professor Steve Jones, a geneticist, who later apologised. Unfortunately, he wrote an article in the Telegraph (15/10/2013) attacking how I used the research on ‘heritability’. He based his article not on my actual essay but on false media reports. He made similar comments to Polly Toynbee who wrote a column on the same day (this column completely confused what I had said about ‘heritability’ and claimed that ‘wealth is more heritable than genes’, which is meaningless if the word ‘heritability’ is given its scientific definition).
In my essay, I had used the standard definition of heritability and explicitly warned about a common misunderstanding (‘[Heritability] is a population statistic – it does not mean that for every individual x% of one’s IQ score is accounted for by genes’, p.196). I also warned that the media routinely confuses such debates by getting these things wrong. Unfortunately, many media reports, ironically, reported my account of the research along the lines of ‘Cummings says x% of a child’s performance is predetermined by its genes’ alongside furious denunciations. I replied to Professor Jones and Polly Toynbee here (Telegraph, 15/10/2013).
Professor Jones graciously apologised for his mistake in assuming the accuracy of media reports and suggested that I add the text that appears at the bottom of my article.
Most commentary repeated factually wrong ideas about what people thought I’d written about genes. Some thought I was giving my own views about genes and IQ: in fact, I did not try to give my own views – I do not have my ‘own’ views on such technical subjects – but merely tried to summarise the scientific consensus. The world-leading expert on this, Professor Robert Plomin (King’s), told the media that I had not misrepresented the science.
There were some interesting blogs – for example:
Edge piece by Timo Hannay here.
Blog by Dr. James Thompson here.
Matt Ridley blog here.
Article in Prospect here.
Nick Pearce (IPPR) blog here.
Duncan Brown blog here.
For the first time since it appeared I am going through it and collecting errors pointed out to me. I will keep a tally below and update it some time…
In Section 2 I shift between energy and power without correcting the units, which means some of them are wrong. (Thanks to @michael_nielsen)