Architecting Moonshots

“We need a dream-world to discover the features of the real world we think we inhabit.”

― Paul Karl Feyerabend, Against Method

This is an effort to consolidate my thinking. I will explore different dimensions of large scale programmes for technical and scientific R&D. My quest is driven by general interest and day to day challenges at the Alan Turing Institute. Below, you’ll find resources and references on moonshots, but the list is by no means comprehensive or final. I will not be drooling over the Apollo project.

These are some questions I explore around mission-led innovation:

• What is a moonshot?
• What makes a good moonshot challenge?
• Why are certain moonshots more successful than others?
• How do we compare moonshots?
• What is the appropriate timeframe for the successful design and delivery of a moonshot?
• How does one make a successful roadmap of technical quests?
• What is the appropriate size, budget and duration for various challenges?
• How can we incentivise more moonshot projects?
• Who is responsible for funding and delivering moonshots?
• How are moonshots governed?
• Are moonshots lame?
• A “What Works Centre” for moonshots
• Civic Technology moonshots

What is a moonshot?

Whether you came through Cumming’s blogpost, BARPA, Japan’s new moonshot programme, the Mazzucato wave of mission-led innovation or Loonshots, you will be interested in ambitious and groundbreaking research & innovation programmes. I’ll use the term moonshot to mean a subspecies of mission-oriented innovation which is:

“A private or public, typically ambitious, exploratory and groundbreaking in nature initiative, often cross-disciplinary, targeting a concrete problem/challenge, with a large impact, a well-defined timeframe, and clearly defined (societal or technological) goals targets and progress monitored along predefined milestones.”

A moonshot is essentially a task force for social and technological research and development and requires a creative, directed, and sustained fusion of science, engineering and entrepreneurship. Sir Patrick Vallance and Professor Dame Nancy Rothwell have identified seven central principles that they believe are needed to guide a moonshot. A moonshot should:

i) Excite and inspire the public, academia, and industry

ii) Help solve an important societal issue

iii) Be truly disruptive and groundbreaking

iv) Focus on areas where the underpinning science is at a stage to make a significant breakthrough feasible

v) Be specific and well-defined in what it sets out to achieve, with a clear timeframe for completion

vi) Take advantage of areas where the leading country (in this case the U.K.) is or is poised to be a world leader

vii) Generate significant additional benefits.

The observatory on mission-oriented initiatives by the Joint Institute for Innovation Policy (JIIP) has mapped 137 ongoing mission-oriented Research & Innovation (R&I) initiatives in 32 countries. These include efforts from governments, international organisations, civil society, and businesses tackling societal and technological challenges in climate change, energy, food, health, security and transportation.

Some initiatives have their origins in societal or economic problems, such as pollution, energy and water security. Others aim to catalyse ongoing efforts, for example, develop a vaccine, map the human genome or strengthen industrial competitiveness. Moreover, some initiatives aim towards accelerated technical and scientific advancements while others target more systemic challenges. These vary according to the policy instruments that are mobilised, the technological and scientific challenges addressed and the markets that are targeted, among other factors.

Fisher et al. (2018) distinguish between Accelerator Missions (a.k.a Moonshots) i.e. well-defined programmes whose accomplishment relies on accelerated scientific and technological advancements such as accelerating the development of a solar-powered aircraft and Transformer missions aimed at system’s transformation that require not only research and innovation but also changes in regulation, behaviour & even new markets. A combination of accelerators and transformers is needed for successful mission-oriented policy.

Examples include the Solar Impulse, the Estonian E-mobility Programme, the Korean Brain Initiative, the E.U. Fuel Cells and Hydrogen Initiative, the U.S. SunShot Initiative, the London Cleaning-up Air initiative, the Delta Programme (Dutch floods protection system), the U.S. Cancer Moonshot and the Indian Electric Mobility Plan among others.

These cases make clear that mission-oriented R&I policy is shifting from mere knowledge production and technical breakthroughs towards bigger socio-technical missions such as tackling climate change. Directionality and intentionality (i.e. series of connected transformations of socio-technical systems in a similar direction) differentiate moonshots from other types of R&I activities. Moonshots typically have well-articulated goals, timelines and milestones and may predominantly focus on knowledge application (while also having targets related to knowledge production).

From Wicked Problems to Grand Challenges

“There is, then, no such thing as the food crisis. Similarly, there is no such thing, in isolation, as the population crisis, the urbanisation crisis, the pollution crisis, the armaments crisis, the oil crisis, the energy crisis, the fertiliser crisis, the resources crisis, the water crisis, the soil crisis, the fish crisis, the technology crisis or the trade crisis. Each of these crises acts on the others.

While it may be useful to focus attention on them one at a time, none of them can be solved unless the others are taken into account. This hydra-headed world crisis is difficult to comprehend…

— Jon Tinker. The Green Revolution is over. New Scientist, 1974”

The world is full of sticky problems and wicked challenges. The Club of Rome introduced the term “world problematique” to describe the situation in which we are no longer confronted by discrete problems but rather by an “intricate and dynamic maze of situations, mechanisms, phenomena, and dysfunctions, which, even when they are apparently disjointed, interfere and interact with one another, creating a veritable problem system”.

This means that no purely technical, economic, or legal measures can bring substantial advances. “Entirely new approaches are required… and a new reorganisation will involve a supreme effort of understanding, imagination, and political and moral resolve.” (Commentary by The Club of Rome Executive Committee on The Limits to Growth. New York, Universe Books, 1973).

Unpacking the dynamics of problem spaces is crucial for deciding whether a moonshot approach is required, assessing their success and for moonshot designers to take a system’s perspective. This could also be a good meta-problem for the field, i.e. how to develop tools to discover problems and evaluate ideas (as described by Brett Victor).

I’d say a good moonshot challenge has the potential to change the lives of dozens of millions of people for the better; encourages new combinations of disciplines, technologies and industries; has multiple, bottom-up diverse solutions; presents a clear case for technical and scientific developments that would otherwise be 5–7x more difficult for any actor or group of actors to tackle; and acknowledges the principles of deep transitions i.e. how socio-technical systems are a part of our economies, cultural frameworks, social interactions and polities.

What makes a successful moonshot

Several factors can make or break moonshots, and I will focus on two: technical and political feasibility.

Casadevall and Fang argue that one difference between successful and unsuccessful scientific moonshots is “the extent to which the basic science underlying the goal is understood”. Some successful examples include:

• the Manhattan Project in the 1940s (based on a solid understanding of nuclear fission since the 1930s),
• the Moon landing in the 1960s (based on Newtonian physics and advances made German ballistic missiles in the 1940s),
• AIDS therapies in the 1980s (relied on decades of research on retroviruses and antiretroviral treatments going back to early 20th century),
• and the sequencing of the human genome in the 1990s which relied on our understanding of DNA heredity and the variety of methods we could use to sequence it.

Projects like President Nixon’s war on cancer in the 1970s are an exception to the above, as it didn’t manage to deliver a cure. However, in the past 40 years, we’ve made huge progress in our understanding of cancer and we have managed to reduce death rates for major cancers like breast, uterine and gastric as well as improve survival times. This indicates that complex systemic challenges take a lot of time and require sub-systems and behaviours to shift simultaneously (and that perhaps now is a better time for a cancer moonshot?).

High-level political support is another critical factor for the success of moonshots, especially the ones led by governments. Examples include the German Energiewende and the Brazilian Inova Renewable Energy which have received sustained political support. JIIP also highlights that especially for collective action challenges like climate change, coordination at the national, European and international level reinforce a moonshot’s political feasibility (examples include the E.C. climate package and the Clinton Climate initiative)

Who is funding moonshots?

Most moonshots are delivered in a top-down way and funded by governments. In recent years more players have entered the moonshot ecosystem. These include initiatives like the Moonshot factory at Google X, Tesla and projects by private citizens like Boyan Slat’s Ocean Clean-up. Companies are also sponsoring prizes. Qualcomm, for example, is sponsoring a $10 million X Prize to develop a portable wireless device that can diagnose a range of health conditions.

Moreover, some venture capitalists invest in startups that are tackling grand challenges. Examples include the Founders Fund and Flagship Pioneering. The latter has invested in a company called Essentient, which claims that their “technology could provide enough protein for everyone on the planet in an area the size of Rhode Island”. Accelerators like Y Combinator have developed the model of “Requests for Startups” for example in Carbon Removal Technologies. Last but not least, Deep Science Ventures characterise grand scientific challenges and fund entrepreneurial researchers to work on them.

Moreover, foundations like the Rockefeller foundation with its Food System Vision Prize, the Wellcome Trust with Wellcome Leap and philanthropists like Bill and Melinda Gates foundation are increasingly shaping the field.

The funding ecosystem -while rich- it has gaps and is fragmented. It could benefit from greater collaborative structures and pooling of funding and responsibilities within countries and internationally. This collaboration can be operationalised and funded through various models which include (Mission Innovation, 2019):

• Bi-lateral/ Multi-lateral joint calls led by various research councils
• Mutual opening agreements of R&D programmes where R&I entities from one country can participate in calls by another country. For example, the European Commission and the U.S. National Institutes of Health agreed for the Horizon 2020 Programme to fund entities established in the U.S. and vice versa.
• Collaborative Platforms that support international research collaborations (e.g. EUREKA)
• Corporate and public-private transnational R&D where private-sector set up corporate transnational R&I collaborations (e.g. the BMW Group and PSA Peugeot Citroën joint R&I centres)

Moonshots over time

Successful moonshot programmes have a long impact horizon, but they are also time-bound to ensure focus and directedness. There are cases where earlier moonshots can ripple into the future and produce outcomes that could not have been imagined/planned when the projects started.

For example, the developments of the Manhattan project later found their way into civilian nuclear power and radioisotopes for medical use, and the 60’s space program improved weather forecasting via satellite observations and gave us the GPS. This is perhaps an argument for critics of highly technical not directly socially relevant R&I programmes which can be impactful 40 years down the line.

How do you select challenges?

This question refers to who’s doing the futuring and whose values and wellbeing are prioritised in the design of moonshot programmes. Even though there are some obvious grand challenges determined by national agendas (see U.K.’s industrial strategy) or international goals (see the U.N.’s sustainable development goals), overall there is no general consensus on the relative priority of various problems. As noted in OECD’s report on the Social Indicator Development Programme: “Commonality of social concerns among Member countries tends to be greatest at the highest level of generality, diminishing as the definition becomes more specific.”

The scope of a mission influences the process of mission formation. Mission programmes are usually developed top-down and at the highest level of government (examples are the Mission-Innovation initiative and Obama’s Brain Initiative). Academia also has a long history of selecting and prioritising technical and scientific grand challenges that feed into R&I policy and may use a gamut of innovative crowdsourcing techniques, such as these developed by William J Sutherland. There is lots of good thinking behind what is a good problem and some notable examples include:

• cause reports examining particular challenges or opportunities — such as reforming the criminal justice system, preventing pandemics and tackling dementia- that philanthropists can get behind,
• institutes developing rigorous approaches on how motivated actors can do good more effectively,
• Problem briefs that describe a problem, evaluate its importance and suggests ways that philanthropic funders can help solve it,
• guides for picking important problems based on their neglectedness, importance, and tractability,
• guides on how to best prioritise causes.

(I definitely want to see new ways of shaping agendas using social media, incorporating the insights of savvy and interested citizens or scholars outside of academia and even scoping Millenium Prize-like problems across multiple disciplines and intersections of fields.)

Roadmapping moonshots

This is one of the more crucial strategic aspects of moonshots and involves the systematic breakdown of grand challenges into addressable (preferably measurable) intermediate sub-challenges with clear goals and envisioned outputs. A roadmap should have appropriate incentivisation and engagement mechanisms in place (we will examine this the next section) and a clear timeline.

A good example of a roadmap comes from Xprize who has pioneered the use of impact roadmaps for their challenges. In summary, the document paints a picture of the preferred future state of the world which establishes an aspirational vision to guide the analysis; it then systematically works backwards to understand what is needed to achieve this future; it examines the obstacles and explores the existing and emerging remedies for these obstacles; through this analysis, they arrive at some grand challenges and then identify several breakthroughs that if achieved, can overcome the grand challenges.

Another good example of breaking down challenges and identifying dependencies comes from 50 Breakthroughs that has identified the most critical science and technology breakthroughs needed to achieve the SDGs.

In principle, this process also requires a good map of knowledge across fields, the identification of gaps or opportunities as well as tracking/forecasting of the migration of research concepts and promising methodologies and technologies. There is a lot of room to develop new tools to help us with road mapping. Systems like Hanalyzer, extract and integrate data from the literature to create a knowledge network about genes and their interactions and then reason over this knowledge graph to help biologists form new hypotheses. Nutonian and DataRobot explore the hypothesis space, autonomously select the most promising ones and devise experiments to test them. There is also room to more systematically explore both idea and solution spaces, as described by Adam Marblestone in the field of neuroscience.

How do you incentivise moonshots?

There are various ways to design and deliver moonshots. Successful moonshots can create opportunities for work across disciplinary boundaries, establish networks of engaged and excited partners and have clearly identified entry and exit points for those involved. They successfully align the incentives of those involved.

Edler (2017) describes a variety of innovation policy instruments that I think could also be mobilised in the context of moonshots, especially those led by governments. These broadly fall into three categories: enhancing the innovation supply, jointly enhancing supply and demand and enhancing innovation demand. The overall aims range from increasing R&D skills, to improving systemic capability, complementarity and even discourse.

Science-Practice has expanded these ideas on what I find a useful way to think about innovation supply. The proposed incentivisation mechanisms include prizes, workshops, grants, procurement, competitions, hackathons and incubators, which aim to build, expand or support an existing community of innovators.

Image by Science Practice

As far as I know, if one was to design a new moonshot, there is not enough evidence to choose between appropriate instruments and incentivisation mechanisms.

How do you manage and govern moonshots?

Mission-oriented R&I initiatives have very clearly identified governance bodies that are responsible for the missions. This structure varies across missions and countries. For example, the Cancer Moonshot and Japan’s strategy for robotic technology for elderly care are governed by several ministries.

In complex initiatives such as Energiewende governance is more distributed and polycentric: it’s coordinated by the Federal Ministry for Economic Affairs and Energy, and it’s implemented by many other ministries and interest groups, cooperatives, banks, and individuals at the European, federal, state, and municipal level. Other initiatives, like the Chinese Deep-Sea Workstation, involve major public-private partnerships. Unfortunately, in most missions, there is no citizen involvement.

Moreover, there is a lot of work on non-bureaucratic and creative management practices. Moonshots need to be managed through an agile and adaptive process as they may run over several years and involve hundreds of organisations and individuals. A lot of thinking has gone into appropriate funding structures, less so into creating “attractors” for organisational and systemic collaborations.

Several innovation portfolio approaches can be used in the context of running moonshots to ensure that projects and programmes are funded, and managed in a much more connected way. Given the scale of moonshots, insights from megaproject management can also be helpful. Megaproject management refers to usually highly complex projects that have budgets in the billions of dollars.

(I found CERN to have a somewhat nerdy method to organise 2000 scientists and to effectively coordinate and monitor hundreds of sub-projects involved in each detector.

“The LHC accelerator itself was monitored through a methodology called ‘earned value management’, which is suitable for large complex projects that are centrally funded and managed. For the detectors, constructed in a distributed manner under central (but also, to a certain extent distributed local) management, such a system was not practicable. Therefore, the detector collaborations were asked to define rather detailed milestones and a plot was invented where the number of milestones to be achieved to complete the project on time was plotted against time.

On the same plot, the number of milestones actually achieved was also plotted against time, in a different colour. As a management tool, this methodology will never make it into textbooks on project management, and it could of course, only have been invented by a physicist. The plot became known as the Cashmore plot. It illustrates that the divergence observed between achieved and planned milestones leads to ‘re-baselining’ and to ever better-understood planning.

(It should be emphasised that the detailed schedules underlying these plots were indeed the real tools to plan and monitor these complex projects, but these were not suited for ‘high-level’ status reports.)

Are moonshots lame?

In some ways, I think they…are. There are several critical arguments that I agree with:

• We can actually do better than moonshots. Complex challenges like cancer and climate change demand a much more careful and expansive approach compared to single moonshot like missions.
• Moonshots can be dangerous when presented as panaceas to problems. Many contemporary issues cannot be addressed simply through better and more directed science and technology. We should also leave space for open-ended, curiosity led research whose goal may not be evident from the beginning.
• The term moonshot creates hype and obfuscates the need for detailed plans/ explanations which fundamentally distracts us from the systemic nature of the problems facing society.
• The role of R&I isn’t to just produce new bombastic technologies but to also implement/maintain existing technology, engage with the public and gain public trust (which are mostly missed by moonshot-like programmes)

A “What Works Centre” for Moonshots

A more systematic understanding of the successes and failures of these initiatives is needed to avoid pitfalls and ensure we are using fit-for-purpose R&I mechanisms.

One may use different vectors of comparison to understand what works and why, such as these used by JIIP, which include: the objectives, the elaboration process, technical and political feasibility, governance, resources, policy mix, relationship with other initiatives, and the strengths, weaknesses, threats and opportunities of various mission-oriented initiatives. Juan Mateos Garcia takes a more data-driven approach to prototype key indicators that aim to help R&I policymakers understand:

What are the levels of activity and funding in this mission field?

How have the levels of activity evolved over time?

What is the disciplinary breakdown of the mission field?

How has the disciplinary breakdown of the mission field evolved over time?

What are the levels of interdisciplinarity in the mission field?

What is the distribution of outcomes in the mission field?

What actors are active in the mission field, and what is their ‘novelty’?

What is the diversity of technological trajectories in the mission field and how it is evolving over time?

Having this knowledge and evidence can help those responsible for thinking and shaping moonshot initiatives. Also, there is a lot of institutional dark matter that is not captured in papers or databases so I would invite people —academics, practitioners, funders and public labs etc— to share more publicly what works and what doesn’t.

The big gaps

Moonshots can be a useful mechanism to focus the energies of various actors towards bigger societal missions, especially those that benefit from significant developments in science and technology —almost all, one may argue.

Moonshots don’t have to be just technocratic in nature. It’s really important to imagine more participatory versions of ambitious socio-technical programmes. These “moonshots” should go beyond the dipoles of public/private and expert/non-expert. There can be civic-technology moonshots that build open digital infrastructures for the public interest; systems, software, hardware and governance structures that ensure equitable access to opportunities; and technologies that reconcile the environment, human ‘progress’ and redistribution.

Also, the idea of scale needs to be perceived beyond “scaling up”, i.e. making initiatives bigger in terms of funding and/or resources. We need scaling out (i.e. finding synergies across projects and expanding impact across systems) and replicating (i.e. using existing R&I and reproducing its principles). And of course, we need good maintenance and care of it all.

Moonshot designers don’t have to mimic the language, institutions and/or R&I initiatives developed somewhere else and shaped by rivalrous wartime rhetoric, but rather take only what’s relevant to the context they are operating in. Governments should also consider strengthening the ecosystems *around* moonshots —including people, capabilities and sociotechnical infrastructure and imagination.

I want to see many more technically ambitious, directed and interdisciplinary moonshots that are fit for the complexities and social realities of the 21st century and can get us faster to a safe and just post-carbon world.

I’m very hopeful!

Thankful to Hushpreet Dhaliwal and Logan Graham for bouncing off ideas and resources.