Speculative technologies is a nonprofit research organization based in the US.
Speculative Technologies was founded by Ben Reinhardt.
Kanjun Qiu and Adam Marblestone are on the board of directors.
Speculative Technologies has been funded by Patrick Collison, Malcolm Handley, Protocol Labs, Schmidt Futures, The Alfred P. Sloan Foundation and a number of smaller donors.
Physical research is expensive — we expect a single five year program to cost approximately $10M. If these programs are actually high-risk (say, a 10% chance of success) it will cost $70M over at least five years to run enough programs to have a 50% chance of a single success.
These numbers are large in an absolute sense, they are cheap compared to other research funding: 0.2% of the NSF’s budget, 0.04% of the philanthropic dollars spent on science per year, and 0.001% of the amount spent by the US government on R&D.
In version 1.0 we are depending on donations from foundations and individuals. In the long run, we’re exploring several options that could bring in more consistent funding and hopefully make us autocatalytic. While it is far from our core goal, the best way to get some of our technology into the world will be through licensed IP or spun-out startups. Another approach is to build a research consortium similar to the Semiconductor Research Corporation.
Government-funded research has resulted in many breakthroughs and we are certainly not trying to replace it! However, like any institutional structure, government research has constraints. Specifically, government research in a democracy faces a fundamental tension: taxpayers want governments to be responsible with their money, but great research requires speculative bets. We can take risks without leaving the taxpayer on the hook for it.
There’s no “natural” role for governments to do research. While governments funding the majority of research is just “the way things work” today, that has only been the case for less than a century.
More broadly, no single institution can support all eventually-good ideas. Institutional pluralism is important so that ideas that fall outside of that institution’s constraints have a home.
One might expect an organization that creates valuable technology to be structured around capturing that technology’s value. There are many examples of startups or incubators that have successfully created technologies and brought them to market. Profit also creates a tight feedback loop and makes organizations auto-catalyzing.
Despite these advantages, we are structured as a nonprofit because it is structurally hard to capture value from the systems-research style work we do and by setting out to capture the value we create, we believe we would ultimately hamstring the technologies.
Furthermore, even if DARPA could capture a 1% fraction of the value of the entire internet, it would not have a positive return. We believe that brutal honesty is the best policy and that we would be deceiving investors if we promised them anything beyond a low probability of modest returns.
It’s true that both corporations and academia do a lot of research on materials and manufacturing. There are entire departments devoted to them!
However, the sort of research work that is happening isn’t the kind of work that will unlock big technological shifts:
- Academic work focuses on only the most interesting and novel part of the problem: showing proofs-of-principle, usually in one component of a larger system.
- Industrial work primarily focuses on improvements within an existing system or business model.
Breakthroughs in materials and manufacturing require creating new systems that can scale. The work to make materials and manufacturing systems scale can require as much research as inventing the system in the first place, but doesn’t lead to exciting papers (and thus, goes against academic incentives). At the same time, new materials and manufacturing processes often require new business models and have worse performance than existing systems out of the gate. Both of these properties make earnest efforts to scale and use unproven systems unappealing to most corporations.
So while there is plenty of research on materials and manufacturing processes, there is a critical gap that we are trying to fill.
There are many similarities between Focused Research Organizations (FROs) and private ARPA-style programs: both focus on things that don’t fit into academia or startups, both leverage empowered leaders, both aim towards clear goals.
However, there are key differences between “ARPA shaped ideas” and “FRO-shaped ideas:
- Clear goal from day one Tight coupling between work-streams from day one
- Strong conviction about the right approach at the project level early on
- Most work done under one organizational roof, outsourcing routine aspects to contractors but not widely distributing the fundamental research questions.
- Quickly ramp up to $10-50m range to focus on scalable execution not achievable in existing organizational settings.
- ARPA shaped ideas
- No clear goal from day one, but a clear goal is the end goal
- Projects are initially parallel with coupling increasing over time
- Unclear about the right approach at the project level early on
- Naturally leverages existing organizations and forms networks among them to develop new system level ideas, approaches or concepts.
Materials and manufacturing may seem removed from the world’s most important problems: war, poverty, inequality, climate change, sickness, and death. But materials and manufacturing are upstream of all these problems. The time and money that we pour into therapeutics and carbon removal are not a complete waste, but without entirely new ways to rearrange atoms, materials to contain fission and fusion power, and new ways to make technology that breaks us out of our hyper-optimized and fragile paradigms, there are diminishing returns to work that directly attacks big problems.
The history of technology is dominated by second-order effects: we solved many huge problems not by working on them directly, but by creating powerful technologies – the Haber-Bosch process, steel, electricity, refrigeration, and beyond. More broadly, new frontiers obsolete old problems, break us out of zero-sum games, and create human agency.
There are three major differences:
- We have different mandates: ours is to build the future through new materials and manufacturing processes; theirs is to “prevent strategic surprise.” While DARPA’s mandate is broad, it still needs to justify all the work it does in terms of national defense.
- We answer to different stakeholders:DARPA is still a government agency. Despite having lower bureaucratic overhead than most parts of the US government, DARPA is still answerable to the Department of Defense, Congress, and the American taxpayer more broadly. Those stakeholders demand a certain set of justifications. We have our own stakeholders: our donors, employees (and we like to think) humanity more broadly. These stakeholders still demand justifications, but a different kind.
- We have different constraints: DARPA has a budget that is orders of magnitude larger than ours. As a result, they’re able to take on more expensive work. At the same time, we have many fewer constraints on how fast we can move and the sorts of “institutional moves” we can make. There’s nothing stopping Speculative Technologies from writing a check the day we decide to fund a project or giving money to an unincorporated team working out of a garage if they’re doing good work.
The problems we’re addressing are not new. The realization that something is amiss with our innovation ecosystem is more than a decade old. Tyler Cowen published his book “The Great Stagnation” in 2011. That same year Founders Fund adopted their now-famous tagline “We were promised flying cars, instead we got 140 characters.”
At the time though, a widely held view was that high-growth startups were the solution. After all, Silicon Valley had begun to disrupt media, shopping, transportation, and many other industries. Nominally, they did this by building technology so it was natural to assume that high-growth startups would soon do the same for materials, manufacturing, and research more generally.
One why now is that enough time has passed to make it clear that we need more than ambitious startups to improve the innovation ecosystem. By and large, the story of ambitious, researchy startups building general-purpose technologies is that they either go out of business, are acquired for talent, or pivot towards profitable niches. It has become clear that we need something beyond startups. Additionally, recent events have exposed the cracks in many systems:
- There are many talented people trained in the hard sciences or engineering disciplines who want to do hard, impactful, research work but (reasonably) opt out of academia and its tiny number of professor positions compared to the postdocs who want to fill them, bureaucracy, publish or perish, and more. People have realized that they could be doing so much more than data science or SaaS startups.
- COVID and Ukraine-driven supply chain disruptions and geopolitical tensions have made it clear that there is a lot of important work to be done in materials and manufacturing to make more robust systems.
- The pandemic also revealed that relatively small interventions could have a large impact on our sclerotic research system. Fast Grants, premised on the simple idea that a fast turnaround time on grant decisions could unlock research that wouldn’t otherwise happen, was able to get hundreds of COVID-related projects off the ground that would otherwise have been stuck in limbo.
To some extent, the answer is also “vibes.” Many people have realized that the innovation ecosystem and old institutions more generally are not working as well as they once did. While perhaps not the consensus position, there’s a growing perception that we need new institutions.
We’re working with excellent program managers who have ambitious visions and deeply understand how research works while being intimately familiar with the limitations of the current system.
We’ve assembled a group of top-tier advisors and donors who understand what we’re going for and will both push us to be pragmatic but not to deviate from the broader mission.
Ben, Speculative Technologies’ founder has worked across the research ecosystem: PhD in academia working with NASA’s Chief Technologist on tractor beams for space robots, visiting researcher at NASA, computer vision at a unicorn startup trying to build drastically new technology, starting a startup, and venture capital. He’s written a heavily-cited analysis of how DARPA works that is directly influencing how Speculative Technologies works.