The Orthogonal Bet: How to fund R&D that is for the public good?

Samuel Arbesman:
So Ben, great to be talking with you. It's awesome.

‍

Ben Reinhardt:
It's always great to be talking with you.

‍

Samuel Arbesman:
Thank you, likewise. So Speculative Technologies, which I, full disclosure, I'm advising as well, just finished its first Brains Accelerator Demo Day, its first cohort of program... And so maybe you can talk about what was, first of all, what is Speculative Technologies, at least a little bit. Then we can go more into how that came about, but also what was the accelerator program? What was this Brain's Demo Day and everything like that? And just go from there.

‍

Ben Reinhardt:
I would describe us as a industrial lab for public goods focused on materials and manufacturing technologies. So to unpack that slightly-

‍

Samuel Arbesman:
That's right. There's a lot of terms you have. I want to hear more.

‍

Ben Reinhardt:
Yes. A lot of technology research ends up being public goods, whether we like it or not. The work that we do is not quite ready to be a product. We do technology research, but with the very explicit focus on making useful things, and so we are explicitly saying, "We're a nonprofit. Our goal is to get technology into the world, not necessarily to capture its value."

‍

Ben Reinhardt:
And then we're focused on materials and manufacturing with the hypothesis that materials and manufacturing are the stuff that we make stuff out of and the ways that we turn that stuff into other stuff. So, it's at the root of our entire technological civilization, and then it's also the technology work that tends to fall into this valley of death where it doesn't quite fit into other institutions.

‍

Samuel Arbesman:
This kind of knowledge, whether it's technological knowledge or kind of scientific research into technologies and materials and manufacturing, these things are oftentimes public goods. I mean, which is acknowledged is a public good, and so therefore, it makes sense to encapsulate or wrap around it some sort of nonprofit organization or allow people to move that research forward without necessarily trying to put it within a more traditional startup.

‍

Samuel Arbesman:
The idea behind it is, and going back to this valley of death, it's hard to do this in between the lab and the startup. But also in some of these kinds of things, they're often enabling technologies to create huge amounts of value, and so therefore, I assume there's a certain ideological component, which is these things should be available for everyone because they will unlock huge amounts of potential. Is that the right way to think about it?

‍

Ben Reinhardt:
Yes, that's absolutely right. The paradigmatic story I like around this is just the thought experiment of imagine that Bell Labs, which invented the transistor, had then tried to capture all the value of building it. Because of the nature of Bell Labs, had an open license on the transistor so people could run with it and eventually build computers.

‍

Ben Reinhardt:
If Bell Labs were like, "Nope, we're the only ones who are going to do anything with that." We would have very fancy underwater relay systems, but we may not have even gotten the computer and all the things that came of transistors. So, it's both that ideological piece and a practical approach.

‍

Samuel Arbesman:
From your perspective, Xerox PARC, when people talk about that of they always say, "Oh, they invented all the ideas around the graphical user interface and object-oriented program and just the personal computer and many of its features." Many people say, "Oh, they did all this stuff, but they really weren't able to capture any of the value." And then it went to, I guess, obviously Microsoft Windows and the Macintosh, because Steve Jobs visited very famously.

‍

Samuel Arbesman:
From your perspective, is that a success case? Where this is actually a success story, where they developed a whole bunch of different things and then allowed it to spread throughout the world?

‍

Ben Reinhardt:
Yeah, I think that it is, and I think that that's kind of the nature of technology. I think what we're trying to do is embrace that, where you see that story repeated over and over and over again, where one organization will invent a thing, but then other organizations will be the ones to productize it and capture that value. It's perhaps controversial, but I would assert that trying to jealously capture that value, keep it for yourself, can end up actually killing it.

‍

Ben Reinhardt:
Again, you could imagine a world where Xerox actually went and created these massive lawsuits against Microsoft and Apple. I think that we'd be living in a poorer world for it.

‍

Samuel Arbesman:
I guess two things there. One is that, so I guess there is the counterfactual release with Xerox, which is, and they did make a lot of money on the laser printer, and yet we still have laser printer from lots of different... And so is that saying that as long as you capture some value from certain of your things, then it's okay?

‍

Samuel Arbesman:
Or is the idea that when you try to maybe prematurely productize things, if it's within a single individual organization, there's almost this failure of imagination of how it can be used, and so therefore it needs to be spread and so therefore it can be used all different places. How do you think about all that?

‍

Ben Reinhardt:
Definitely the latter. There's both that failure of imagination and perhaps an even more insidious thing, which is that if you can't imagine how a thing will make the organization a lot of money, then you kill projects. And you see this over and over again where if the MO of the organization is, "We need to do this research in order to make products that we can then make a lot of money off of," then you almost back propagate these expectations about where the research will lead.

‍

Ben Reinhardt:
It might be possible to have an organization where you give things enough time to cook or you're sufficiently unopinionated about what will lead to really amazing technologies. I've seen over and over again is that when you have those pressures, whether it's from public shareholders or whether it's from investors, then you're always needing to justify research in terms of how will it make us money soon?

‍

Ben Reinhardt:
How can we imagine this? As opposed to being able to ask the question, how could this be amazing and impactful without needing to specify exactly the business model that will allow you to capture the value of that impact.

‍

Samuel Arbesman:
Got it. Which I mean I feel like on the one hand that could argue in favor of a certain amount of undirectedness and idiosyncratic interestingness-based research, which I am a fan of. But I know at least in some of the things you're working on, it's not necessarily focused on value, but it's still very directed.

‍

Samuel Arbesman:
There is this very clear sense of a directionality, and maybe that's the right way to lead into the Brains Accelerator program of what you were trying to do in terms of actually helping create these kind of research programs.

‍

Ben Reinhardt:
I think that's absolutely right, and look, a way that I've actually been thinking about it recently is that I still think it's very important to have feedback loops with the world. So you have these feedback loops that make sure that the research is aiming towards something that useful, and one of the most powerful feedback loops we have is capitalism.

‍

Ben Reinhardt:
So having that feedback loop of do people want to buy the thing that you're making, but that is not the only feedback loop that one can have. The ideas with the Brains program is to enable coordinated research programs that are driving towards this usefulness and this impact, but don't make sense as a project in an academic lab nor as a startup.

‍

Samuel Arbesman:
One of my favorite essay/blog posts that you wrote is, and let me get the title, it's, "When should an idea that smells like research be a startup?" In many cases, I think your argument is, most of the time fundamental research is actually not necessarily good for a startup, and when you try to shoehorn it into a startup-like thing, there is this incentive mismatch, whether it's a coordinated research program or some other kind of thing. And how would you think about where research fits into the startup ecosystem?

‍

Ben Reinhardt:
Part of writing that article was me grappling with that question without actually coming to a hard answer. I think that startups are really good at making products and scaling those products and building businesses around them. Where research fits in is almost everything before that maybe shouldn't be a startup, and in part it depends on how quickly you can get to that product.

‍

Ben Reinhardt:
So one of the reasons that I think software is both amazing and confusing is because you can iterate on it so quickly that you can actually do software research on a timescale that allows you to get to a product within the time bounds you have in a startup. For most other domains, that's not really the case because it just takes so much more resources and time to get things to be products.

‍

Samuel Arbesman:
So then this cohort, the people who went through this Brains accelerator, so they were thinking about these coordinated research programs, which could be more like ARPA style, they could be more [inaudible 00:08:32] style. Give me a sense of the vast sweep of both the topics, the goals, and even the types of structures that participants were thinking about that could actually be a fit for building these things.

‍

Ben Reinhardt:
Just to give you a sense across the spectrum, so we have on the ARPA-style program, and one of the fellows was working on a program to create a new system for developing and verifying clinical endpoints for testing drugs to drastically increase the rate at which drugs succeed in clinical trials. So if you have better endpoints, you can know not to do a clinical trial for more drugs more quickly. This needs to be an ARPA program because it requires buy-in from many different drug companies and many different labs testing many different endpoints.

‍

Ben Reinhardt:
On the other end of the spectrum, you have one program to build a high throughput material screening system that wants to be an FRO, because you need everybody in the same room building this same machine and integrating it tightly with software. And then orthogonal to that, there was another program. The goal is to drastically decrease the time and costs of building nuclear power plants by bringing in construction techniques that have already been approved out in other industries but have not yet been regulatorily approved for nuclear power.

‍

Ben Reinhardt:
The goal of this program is do the work to get the data to show the nuclear regulation commission that this construction technique is viable for nuclear power plants so that other organizations that are building power plants can then start incorporating these techniques.

‍

Samuel Arbesman:
So that one's almost like doing the lifting to allow the import/export of ideas to actually work smoothly of there's this really set of interesting set of construction ideas or manufacturing ideas or techniques, technologies, but they're only in this other space and in order to get them here, we need to import them according regulatory things.

‍

Ben Reinhardt:
Exactly.

‍

Samuel Arbesman:
Oh, that's fascinating. So you had this demo day. Who are they demoing to and what is the goal for these fellows afterwards? Because in startup demo days it's like, "We're going to raise money and we're going to build a startup."

‍

Samuel Arbesman:
And for this it can sometimes be clear, like building a nonprofit or organizing things or maybe getting a role at DARPA in order to actually become a program manager. Is it all of these different kinds of things? Is it also similarly bespoke of I'm going to pair you with these people to actually give this thing the best chance it will have to succeed?

‍

Ben Reinhardt:
In large part, it was the two things that you mentioned, in the sense of getting a program funded or getting someone a role at an organization where they could go and execute on that program, whether it's at one of the government ARPAs, whether it's at a large foundation that runs programs like this. That's the primary goal. The secondary goal though is to just get well-connected people excited about the thing to help the fellow make it happen.

‍

Samuel Arbesman:
The success of the program, you're trying to help catalyze them because they are interesting, misfit, I'm using misfit in the best possible sense-

‍

Ben Reinhardt:
I love [inaudible 00:11:46] a misfit.

‍

Samuel Arbesman:
They don't quite fit in other different organizations, so you are trying to make them work. So, I love this idea of within two years whether or not they're actually running these things.

‍

Ben Reinhardt:
The reason that we're not paying attention to the success or failure of the programs and then their eventual impact is, one, just the timescale is so, so long. And then, two, it's an open question of what does success for one of these programs look like.

‍

Ben Reinhardt:
With a startup, it's very clear what success looks like. You have a giant exit, put a number on that. With these coordinated research programs because there's not a scale or value that you can measure them on, they could do some work that then 20 years later ends up being really core to something else.

‍

Samuel Arbesman:
It's diffuse, illegible. You're playing the long game, which is great because you want to have a space in the world of research or progress in general to allow for those things to actually marinate and simmer and actually just be done. Related to these coordinated research programs, you mentioned this as a spectrum, I also view this as a pretty high dimensional space of the types of institutional forums. I don't really have a sense of what the new types of forums should be.

‍

Samuel Arbesman:
Is that also one of the things as these fellows are exploring things is trying to get them to think about this high dimensional space of maybe there are these weird new types of structures that allow people to either do research or just do other things?

‍

Ben Reinhardt:
Absolutely. My meta theory of change and Brains' meta theory of change is that we need many, many new kinds of institutions. We've almost collapsed, especially in our research ecosystem. You have academia and then startups and they've basically eaten all of the different kinds of work. So the theory is that we need many new institutional forums.

‍

Ben Reinhardt:
And so throughout the course of the Brains program, we work with the fellows to say, "What is the right institutional forum for this idea that you're working on?" And obviously, we're to some extent limited by what we can convince people to fund. I think that it is a mechanism for getting new institutions out into the world.

‍

Samuel Arbesman:
Evolutionarily, maybe we've weeded out all these other weird ones, maybe not. Was there ever a time that you felt like there were lots of different types of research or institutional forms, and we've lost that or has there never really been. And there's a need for actively injecting the system with an increase in temperature and willingness to experiment?

‍

Ben Reinhardt:
In the past, there were more forums and things were a little bit more fluid. I think we didn't call them, it wasn't a formalized thing where there was like, "Ah, yes, this kind of institution and this kind of institution." But if you look at the history of research organizations in the US, there were weird things. There were research collectives, and-

‍

Samuel Arbesman:
What exactly is a research collective just out of curiosity?

‍

Ben Reinhardt:
It's like an R&D organization that was a co-op. So, you have all the people doing research voting on what they would do research on.

‍

Samuel Arbesman:
So in of the funding, then we have all these ideas for all these new types of institutional structures, and some of them are good, some are not. We probably need a lot more structures. We need new types of research programs, but if there's not similar efflorescence of scientific funding structures, then none of these things will happen. Is that another foundational problem as well that needs to be addressed?

‍

Ben Reinhardt:
I think so. To some extent, in the same way that Pacific institutional structures have sort of taken over most of the work of research, the government has taken over a lot of the role of funding research. Other organizations have started to back off and say, "Oh, well, that's the government's job." That then forces all that research into this bureaucracy and this situation where it's explicitly calling out what structures people need to be using.

‍

Samuel Arbesman:
Is there a need then for almost an accelerator program, not for creating coordinated research programs, but for getting funders of all different types, whether it's individual philanthropists, foundations, specific types of government agencies to actually think more broadly? Is there something there too?

‍

Ben Reinhardt:
I don't know if you've been paying attention to the Renaissance Philanthropy. It is a new thing started by Tom Kalil, who until recently was at Schmidt Features, former White House OSTP leader, who's been in the research world for a very long time.

‍

Ben Reinhardt:
My understanding of this organization is that their goal is to really activate a lot of science and technology, philanthropy, and the idea that there are many philanthropists, especially high net worth individuals who don't have an institutionalized funding organization. So they don't have a foundation the way that Bill Gates has a foundation, and they're interested in funding science and technology, but they're very busy.

‍

Ben Reinhardt:
And it's much harder to evaluate a proposal for a scientist or technology research than it is for soup kitchen, where it's like, oh, you can be like, "How well does the soup kitchen give out soup?" And so their goal is to really help those people find projects that they're interested in funding.

‍

Samuel Arbesman:
Taking a step back, so the speculative technologies obviously is one of these kinds of non-traditional research organizations, and I definitely have a sense, and I feel you've shared the sense of you have a great deal of ideas around the possible space of scientific research should be done. And the current world is a narrowing of it, but to start a research organization is a highly non-trivial kind of thing. What led you to that?

‍

Ben Reinhardt:
Two words, frustration and satisfaction, in the sense that I feel like I really did march through the different institutions that I thought were responsible for enabling this kind of work that I want to see in the world where it is this very use focused, very engineering heavy research that's focused on getting technology to a point where it can actually get out into the world. I was in academia, and I saw, all right, well, we're going to do the interesting thing and publish a paper about it and then move on.

‍

Ben Reinhardt:
Worked briefly at NASA. I went and worked at a startup that raised way too much money. I started my own startup to try to do this and was laughed out of... Found it very hard to raise money.

‍

Samuel Arbesman:
Politely chuckled out of the room as opposed to laughed out of the room.

‍

Ben Reinhardt:
Yes, politely chuckled out of the room, exactly, exactly, politely chuckled out of the room. Especially when I tried to be very explicit about this is going to take research, which is going to take several years. We don't know how to do this. I worked briefly at a VC firm thinking we could change that. Realized that VCs have their LPs, so they're institutionally constrained. That's a very long way of saying eventually I felt forced to do this, as the got to do it yourself sort of thing.

‍

Samuel Arbesman:
Then what has the reception been like, and is it, this is a very bimodal distribution in terms of how people think about these kinds of things and receive speculative technologies?

‍

Ben Reinhardt:
Yeah, that's exactly it. It's incredibly bimodal distribution, where there some people who are like, "I don't get it. Isn't this what we have startups and universities for?" But then there are other people, this is especially true among researchers and technologists who are like, "Yes, please keep doing what you're doing. This is so important."

‍

Ben Reinhardt:
And frankly, that's one of the reasons I keep going is because I run into people who really, really get it. And part of my job and the goal is to help people who are in the former category move into the latter category. It's tough because if you haven't been there, it's not a very legible problem.

‍

Ben Reinhardt:
Where on its surface it really does seem like, "Oh, we have a pretty good system." The government funds university labs who develop technology to a point where they can then spin them out into startups, which then get VC funding, build products and get amazing technology into the world. What's hard to get a sense of is all the ideas that have died because they don't fit into that step-by-step model. Then the hypothesis is that the world will be way more abundant and wonderful if we're able to support those things that right now are falling into the cracks.

‍

Samuel Arbesman:
Create other paths, like say, "Here's a whole menu of ways of getting these things into the world." As opposed to, "Here's the one path that everybody's... It's well trodden and is paved and is very nice." But there are these other paths and some of them are... And you're clearing the brush, and to really mangle a metaphor.

‍

Samuel Arbesman:
But I think you have a pretty clear view of what this longer term and not clear like, "Oh, this is exactly what the world's going to look like." But you have a direction and a vision of what this world should be like, which is driving, I think, a lot of the things around speculative technology. What does it look like for each of us? I don't know if it's a touchstone of it's Star Trek, it's Iain Banks' culture, it's whatever, or is it something, solar punk, whatever it is?

‍

Ben Reinhardt:
I mean, I think if we're going to point to one single work of science fiction, I would definitely point to Ada Palmer's Terra Ignota series, but I think what it entails, it entails this drastically multipolar world where people have a huge amount of agency to live almost like various different science fiction scenarios.

‍

Ben Reinhardt:
Where it's like there are some people who want to live in something like the culture, and so they get to live in something like that. There are people who want to live in something that's more like Star Trek, and so they live like that. I think the big thing for me is I would describe as exploration, agency and capability where we use technology to enable people to do more things in the world.

‍

Ben Reinhardt:
That means very different things for different people. For some people, it means creative expression. For some people, it means exploring the universe. For some people it means just hanging out with their families. I think that's the ability for people to do any of those.

‍

Samuel Arbesman:
To create the preconditions for this kind of combinatorial, open-end of possibilities, of different ways, and giving everyone the technological means, the energetic means, all the different structures to allow them to do all these different things.

‍

Ben Reinhardt:
There's the freedom from things and the freedom to things. There is freedom from disease and concern about food and labor, and then there's the freedom to things, which is the freedom to invent and go where you want when you want, and explore new knowledge and explore out into the universe. Maybe the last thing I would say is that this is the world that I want to exist.

‍

Ben Reinhardt:
I don't think that this is actually the most probable world that we get to. It's not the default thing. So, the thing that I'm pushing for is it will take a lot of hard work to get there and a lot of pushing back against various trends.

‍

Samuel Arbesman:
So as long as you can shift those probabilities a little bit at the margin, then you're making that future a little bit more possible. And that is the goal, ultimately. The vision shifting of the role of speculative technologies is interesting and complimentary to the way in which you see a lot of startups envisioning their impact, where it's like many startups are tech companies.

‍

Samuel Arbesman:
It's like it is a huge vision of that, of their specific thing is going to change the world. Versus saying, "No, the impact is going to be small but steady over a very long period of time in the way you reorient a battleship," or whatever it is. And that in turn will make this specific future that much more possible, and maybe we need more startups to think that thing as well. Maybe this is the only thinking that is available outside of that. I don't know. I think there's a lot of things we can think about philosophically what that means.

‍

Ben Reinhardt:
We're a startup in everything but cap table, in the sense that we are a new, small organization that is trying to do something radically new. The only reason we're not taking investment is because did a Monte Carlo simulation of whether we could... I don't lie.

‍

Samuel Arbesman:
Is that real?

‍

Ben Reinhardt:
Oh, yeah, yeah, I can send that to you.

‍

Samuel Arbesman:
You actually did a simulation.

‍

Ben Reinhardt:
Yeah, yeah.

‍

Samuel Arbesman:
Oh, that's amazing. Oh, that's fantastic.

‍

Ben Reinhardt:
It's in a Jupyter Notebook. I'll send it to you.

‍

Samuel Arbesman:
That's fantastic.

‍

Ben Reinhardt:
I did a simulation of whether I could tell investors without lying to them that we'd be a good investment, and the answer is no because of this public goods hypothesis. So other than that, we act like a startup in the sense that we move fast, all these other things.

‍

Samuel Arbesman:
Well, I think the idea of a startup is a lot broader than a tax code designation of for-profit, nonprofit, 501(c)(3), whatever it is.

‍

Ben Reinhardt:
That's basically a venture capitalist.

‍

Samuel Arbesman:
It's a certain set of ideas around changing the world in some massive way, whether it's in the short term, long term. There's multiple different ways of doing that, but bringing about some sort of positive vision of the future.

‍

Ben Reinhardt:
Exactly.

‍

Samuel Arbesman:
Awesome. This is probably a great place to end. This is fantastic.

‍

Ben Reinhardt:
Sweet.

‍

Samuel Arbesman:
Thank you so much, Ben. This was awesome.

‍

Ben Reinhardt:
Thank you, Sam.

‍