Home page

Dominic Cummings gave evidence in front of the Science and Technology Committee chaired by Greg Clark, the other day. This concerned, mainly, ARIA (Advanced Research & Invention Agency). APEurope Pool specialists have now had time to review the transcript of this session. The observations are set out below.

Many fundamental truths

Anyone with even a modicum of experience with the UK, largely academic research community would agree with Dominic Cummings observations relating to bureaucracy, very slow peer reviews and latency, in general, involving the recording of advances and indeed, obtaining advances. His statements that the UK cannot solve all such problems by itself and neither can an agency since a whole bunch of things need to change in the overall system, are undoubtedly correct. He referred to a diagram which was not possible to see in detail and his promise to post it on his blog, so far, has not been carried out. He seemed to be referring to the entropy of knowledge or the need to understand that the differentials of novel and useful composite objects or concepts depends on the alignment of the components within the system. In other words there is a need for a freedom to explore different system or subsystem configurations. If the knowledge entropy process is padded out with administrative procedures applied to "manage" progress of elements, then getting anywhere becomes difficult.

If this is what he is referring to then Cummings is right to point at peer reviewed output. Peer reviews have become labyrinths within which contributors often try to ensure their own past recorded work and concepts are reflected in whatever is being reviewed. In science and economics this is an issue, less so in technology and engineering. Because many researchers consider their status to be linked to the number of peer-reviewed "papers" they have pumped out and their contribution to the state of knowledge depends upon any such acknowledgements appearing in the references of peer reviewed papers. This causes a bias in output and long delays in the completion of some peer review processes.

Cummings seemed to refer to ARIA as a sampling entity operating across the knowledge entropy space; if this is what he means, it makes a lot of sense.

On how this initiative came about Cummings explained that it was a "deal" agreed with the Prime Minister, on the basis of 4 stipulations made by himself:
  • Getting BREXIT done
  • Doubling the science budget
  • Create some ARPA-like entity
  • Gain support to change how Whitehall and the Cabinet Office works
The majority of the replies provided are highly repetitive because the questions were insufficiently forensic on the issues of what exactly ARIA will do.

An old concept

ARPA (Advanced Research Projects Agency) was established by the Eisenhower administration in 1958 as a panic reaction to the successful launch of the Sputnik satellite by Russia.

However, in the UK the National Research & Development Corporation (NRDC) had been established ten years before in 1948 as a non-departmental government body, formed initially to exploit products developed during World War II by the Defence Research Establishments. These included carbon fibre, asbestos-plastic composites and in semi-conductor technology.

Later it supported new developments such as the hovercraft. On the agricultural front it supported the development of the potato harvester launched in 1959 and which had saved farmers more that £1 billion by 1969. In the 1970s, Rothamsted Research scientists discovered three pyrethroids suitable for use as insecticides, namely permethrin, cypermethrin and deltamethrin. These were subsequently licensed as NRDC 143, 149 and 161 respectively, to companies and manufactured for use worldwide

In 1953 the UK government helped establish a NRDC for the Indian Goverment which is still in operation. It has been transferring technologies to private Indian companies with a significant impact on this economy, now for over 65 years.

A developer of British potato harvested who gave a lecture on its development in 1966 at the Cambridge University School of Agriculture, described the many "cock ups" and wrong ideas involved in its development but since the purpose was clear researchers learnt on-the-job through relativey short test, assess, adjust and re-test cycles so as to end up with a highly successful product.
There are dangers in attempting to base ARIA on ARPA because the environment during the Eisenhower period was one when all funding depended on Congressional approval. Experience with the Congressional funding cycles shows that under US funded initiatives there was a problem of monitoring and evaluation getting out of hand with evaluation exaggerating results. Retrospective cost-benefit analyses quite often included intangibles as benefits. As early as the mid-1950s people who reported back problems on some overseas project were treated like whistle blowers and they lost their contracts. Reports to Congress always needed to be glowing. This is why in later years USAID began to add performance indicators leading to project managers spending more time of filling our monitoring and evaluation reports than getting on with the job.

It is therefore important which referring to any government-funded initiatives such as ARPA and later DARPA to look more closely at the benefits which can be traced to ARPA and DARPA projects. A significant problem was the notion that the quantity of currency units could ensure benefits; this is very convenient for researchers because they benefit from successful applications for funding. As Dominic Cummings stated in, response to some questions, the actual required outlays for some successful initiatives were quite small.

A danger with stripping away the counter-productive procedures is that whatever replaces this should contain people who know how to carry basic research through an associative notion to experimentation and then useful results to proof of concept and prototyping to an innovative product or service. This technology push also requires the capability of carrying out needs identification and lines of enquiry based running in the opposite direction. One of the benefits of DARPA was that it introduced an operational "island" approach free from state bureaucracy and corporate executive interference. As a result many initiatives were successful but also, over time it helped build up a community of first rate project managers. This was possible because of the duration of projects and the significant funding involved.

One of the problems with the DARPA-type model is that the outsourcing of defence projects became a way to subsidize US industry, especially in aerospace and had very little to do with research. European strategic programme (EEC) for research and development in information technologies (ESPRIT-1 and ESPRIT-2) allocated 2.35 billion Euros. The "evaluation" of ESPRIT-1 results was based on the opinions of the companies who had benefited from ESPRIT-1 funding, so naturally phase 2 was far bigger. Most of the languages and systems developed were overtaken by the normal progress in technology and by more useful languages and the semiconductor segment, the biggest allocation, came up with nothing. Here we see the impact of allowing the corporate executives have a say in R&D and applied R&D.

Towards the end of the session, Cummings referred to Education Department parochialism in the context of talking about lessons from China. Having left the EU it is true what Cummings stated concerning collaborative programmes continuing in any case, between leading researchers, without the red tape. In any case several projects include as eligibility criteria the number of collaborators from different Member States that need to be involved. This resulted in a sort of rent-a-collaborator business to make up the numbers just to become eligible for funding.

However, on the parochialism front, the UK has a serious problem in relation to bridging the working prototype to commercial product or service development, the "valley of death". This problem will increase for the UK in particular. This is because what gave the USA a significant advantage up to 2000 was the size of its internal market in terms of population and per capita income. Therefore for venture capital developments there were rated as lower risk than in the UK with a small market. In 1984-1986 market projections for global-networked devices, by an internal project managed by a temporary agent at the Information Technology and Telecommunications Task Force (ITTTF) at the European Commission including mobile telephones (note this was before the W3 and iPhones existed) projected that China and India would become the main consumer and development centres for these technologies by around 2010. The logic applied was the size of the market. The ITTTF projections were based on a mobile telephone costing 100 ECU (European Currency Units) and then applying the learning curve coefficient (Wright's Law) linked to the rising rate of sales in China associated with the steady decline in unit costs, resulting from a combination of Moore's Law on semiconductor stuffing and Wright's Law, and a rise in per capita incomes . This proved to be completely correct. Huawei, which was set up as a mutual, in the mid 1980s when this projection was made, is an example of how successful this process is. The US accusation of subsidy of this company's operations giving it a competitive advantage is completely unfounded.

However, this reality presents a problem for the UK. We have nowhere near this size of domestic market i.e. in population terms alone the UK is just 25% of the USA's population and less that 5% of the population of China. As a result global venture capital will regard UK applied development to incur relatively higher risks and the "valley of death" will become morre difficult to cross. Research at SEEL-Systems Engineering Economics Lab on this issue has explained that in some application such as AI, the velocity of transactions can compensate for population numbers in reducing the unit costs of products and services. However, the decision by the UK government to follow the USA in getting rid of Huawei 5G units further reduces the ability of the UK to benefit from this added advantage in relation to advanced networked applications. Therefore one of the mitigation factors to compensate for our tiny population has been lost, placing our future economic status in precarious state of affairs, a result of an unjustified paranoia emanating from the USA. The UK government's cavalier approach to development and trade and the "free market", in the light of this reality, is an overly romantic and naive approach because, in strategic terms, this creates too many dependencies for national success on agents who have no particular alliegence to the interests of the UK population. This has been the problem creating the current vaccines supply uncertainty and no one quite knowing what will occur; an unacceptible state affairs.

In terms of the role of science and technology needing to become a strategic plank of Number 10. This particular context and reality placed a question by Mark Logan into context. This was,
"...on the industrial strategy, what would be your views of in terms of that buttressing of our science and tech prowess?"

Surprisingly, Dominic Cummings stated that he had spent more time talking about industrial policy and the productivity challenges than he did about all political and communications issues put together. If this is the case what were his conclusions because the answer to Logan's question is quite important. Cummings stated that Rishi Sunak is supportive of much of R&D objectives and that Cameron and Osborne did little on this front. The fact remains that for 12 years quantitative easing was on auto pilot leading to falling investment, productivity and real wages and leaving most without any savings.

On the front of economic research it has to be admitted that macroeconomic theory, be this Keynesianism or monetarism, is extremely crude. As established by SEEL, these approaches have no microeconomic foundations and they have no components that support innovation and increased productivity. Indeed work at SEEL has explained why monetarism does not work and how it undermines any asdvantages to be gained from technical innovationin, very logical terms. As can be seen by all macroeconmic policies have failed and the failure of economists to free themselves from a incestuous peer-reviewed world of mutual admiration is running the country into the ground. This only reconfirms the suffocating parochialism the dominates the approach of policy decision makers.

Something like 80% of economic growth is the result of learning, operational competence derived from the accumulation of tacit knowledge gained from the repetitive application of tasks. Associated with this the accumulation of an evidence base from accumulating explicit knowledge in the form of data to be applied in decision analysis leading to advances in innovation, productivity and real growth. Within the constituency such a system can gain wings if the levels and content of general education are appropriate to today's world'; but they are not. This is explained at the end of the previous article in this series, concernng technical education and the innovative approaches to education by the pioneers, Fredrick Sanderson of Oundle and Thomas McNeill of Portsmouth

It is extremely significant, and of concern, that these fundamental realities concerning our application of flawed macroeconomic policies and the population enduring a deficient system of education were not really referred to. This, in reality, made both the questions or the answers in this session somewhat academic.