Bodmer 2.0
Make science great again
In the early 1980s the internet connected hundreds of computers and the Royal Society was the science establishment in the UK. Walter Bodmer’s eponymous report in 1985 was radical in proposing something should change and that serious attention should be paid to the public dimension of science. That was widely mischaracterised as promoting a deficit conception which needed to be remedied when in fact it recommended a lot of practical changes to engage politics, industry and media communication. School education already had a sympathetic government.
The Royal Society has pursued an ambitious project to follow up Bodmer and update it for the current times, a vision of science for society (Bodmer 2.0). But science has changed, not just in the digital systems which support dissemination, computation and observation which have changed the practice of science but also the organisation. There are now separate academies for engineering, mathematical and medical sciences, a development the report acknowledges only in oblique references to working with other national academies on a small number of matters.
Bodmer 2.0 follows up the sections of the original, in public, policy, education, industry and academic strands, but in doing so it neglects the urgency of science. So as well as reviewing what progress has been made on the recommendations made in the mathematical sciences, this is a review of emergency science. In the UK that keeps throwing up challenges around unknowns, risk and uncertainty, all of which are intrinsic to science but not comfortable for society. Rather than making science great again as the report implies we can think about trustworthiness for society.
Education
Since the 1980s, participation has changed: education was not compulsory to 16 but now all students must take science as long as that. Indeed the Royal Society is now firm that the gap is in not requiring any balance across the curriculum taken to 18. Yet science has moved on and in the 1990s the acronym STEM was adopted, adding the ‘M’ for mathematics to science, engineering and technology (SET, as it then was). and these are integrated in applications after formal schooling. Yet the vision and the curriculum remain focused on separated science subjects without integration of computing, so delivery of computer science qualifications and mathematical science as integrated data-driven education for the modern world are an addendum.
People need to use computers to do things, and to understand how automation supports prosperity. Unfortunately, an AI iceberg of public distrust founded on scepticism of political and economic motivations has grown, in contrast to the AI Council vision in their roadmap in 2021. They had expected that everyone needed to know how to discriminate what was good for them in their work, not just their personal life, as adoption included public and corporate procurement. Sadly research on mathematical needs where some relied on deep understanding at a level above the application used is being supplanted by the group whose technical competence was insufficient to use anything more than heuristics for the blackboxes they worked with.
Media
The Bodmer report could see a trust gap in science around health and the environment, but that persisted even as the issues changed. BSE and GM crops, the MMR vaccine were speculation in the media and partisan politics instead of serious reporting of evidence. The Science Media Centre (SMC) was established twenty years ago to inject independent scientific comment and enhance the debate but that model struggles with industry AI research. While there is critical support on general matters of health, specialised issues are not so well served, and media still make a horrible mess of statistics. There is also too much reliance on the one institution to organise response to stories, so the SMC has a reputation for being too close to the science establishment, on contentious issues like ME, and funding for science.
The Bodmer report was unfairly criticised for promoting a deficit approach to public engagement, but there was a lack of strategy still evident today. Popular maths talks appear on youtube (e.g. 3B1B) with much more sophistication than the BBC which hosts science as entertainment not dialogue, and criticism of its impartiality and false balance is unresolved. Similarly the new report has not noticed the James Lind Alliance and says public priorities for science are not known, they are certainly neglected but also ignored. The new report does mention the need for better standards of science to promote trust, but this requires scientists to raise the standard of rigour, in contrast to the current peer review model which rewards novelty.
Politics
Science has come into formal government structures over the intervening decades, so that chief scientists are present in departments and lead networks for advice. At the centre there is a council for science and technology to advise the prime minister, including topics for finance and society. But press conferences often frame policy much more than evidence, so that expedience seems to matter more than curiosity which has recently led to STEM recruitment targets for the civil service fast stream. A change in science context means more than technology being adopted directly and modelling is able to be the policy development process. However, that change has been itself a political pronouncement than a change and support structures like the Alan Turing Institute have come up short in transforming public policy.
The strain on public trust is perhaps best highlighted as citizens do their own research, by which they mean independently cross reference evidence (there is not nearly enough appreciation of the rigour required for robust empirical research). That may be remediable by more engagement but the foundation of public policy is that different arms of the state make judgements independently. The concern is they should be free of political influence in matters of justice, for example, but this separation of powers is technically weak and the expert evidence process remains deficient despite recommendations for reform. Repeated state failure to appreciate science and systems is covered up before being unravelled in public inquiries as administrators responded to challenge by defiantly abusing their authority.
Industry
The Royal Society has always struggled with industry engagement, from its origins as a club for gentlemen scientists to its staunchly academic volunteer governance. But it is surprising that it is so weak on understanding the role of industry in local communities: despite concern about training of technician roles it does not articulate the local labour market role of apprenticeships. And the appreciation of regulation supporting industry is not developed to the strength of the science which has emerged around evidence, particularly in health. The ‘what works’ model has been so effective it has been taken up in other places and is also part of the emerging AI security regime where is it clear that grand claims obfuscate frailties and inequity. Indeed a bigger challenge is that science, especially industry sponsored results, are not scrutable to people applying it in government (although this demand is fairly new).
The revolution in science processing itself (metascience) is overlooked entirely and innovation now about global data processing in distributed computing. This blindness was also seen when the UN expert group produced an interim report about AI which did not mention access to data at all and also speaks to the incoherence of the Turing AI science model. Private sector domination of AI research speaks to defending an incumbent advantage in certain domains but also monopoly on data and scale of computing power. Sadly the science for society does not talk about the role of institutions in shaping the objectives of the research agenda, nor even how the partnership with other parts of the system will operate indirectly.
Emergency
As a starting point, following the sections of the original report was a good foundation, but it neglected a shift as emergency science became security for society. The role of SAGE in the recent pandemic is salient, but its international significance in coordinating and releasing science openly ought to be celebrated. That model has adapted over several crises, some local, and about foreign interference, but the importance of expert academic and technical advice is established. CSAs have a role in the development of the national register of risks, including how risk itself is conceptualised and it continues to develop. And it still took a year for new insights to cohere in the roadmap out of lockdown in 2021 so investment in methods, networks of expertise and sleeping programmes is needed and may also improve value for money.
The incentives of emergency science, lining up a national effort for the public good, saw the UK produce almost all of the pandemic science bar vaccine development in 2020. These incentives to produce results and share them openly reflect science for society in contrast to the current system emphasising personal prestige and competition for priority and recognition. There are occasionally meek appeals to ethics in the new report without explaining what that means, that standards should apply to scientific practice not just trustworthy communication. Scientists can make science great again but this appeal from the establishment smacks of desperation and the hope to build on the statistics code of practice risks spreading that attention thin.
Mathematical Sciences for Society
A lot has improved about science integration since the Bodmer report, with general institutions founded soon afterwards being replaced with specific functions. The acute problem persists much more as access to specific science expertise at the right time, seen across all the sections above. Perhaps the most prominent challenge now is metascience, applying the evidence approach to the processes of scientific activity, which has been dominated by social scientists talking about it all. The contrast with mathematical sciences is striking, being deficient generally and chronically all over the place and active resistance to improvement, including arguments about forming the new national academy. Developments in mathematical science have been so sporadic and fragmented they have depended on individuals and philanthropic support, which institutions have sadly no prospect of continuation.
Science needs to progress to reproducible practices, most notably using documented code to do analysis remotely despite the resistance of established academics. Poor understanding of conditional probability and population strata continues to handicap popular discussion of important policy around public health and badly needs to be taught universally in schools. Many professionals need to pick up skills in working with evidence and communicating it effectively through contemporary media channels, and the mathematical sciences community needs to work with them; similarly the use of modelling in policy development and evaluation. There is an opportunity for data systems supporting emergency response to be enhanced with the support of technical professionals already working on this, but this needs serious discrimination of the distinction between policy and politics (we must also do better).