Go offline with the Player FM app!
Focusing on Research with Adam Marblestone [Idea Machines #33]
Manage episode 275328183 series 2470122
A conversation with Adam Marblestone about his new project - Focused Research Organizations.
Focused Research Organizations (FROs) are a new initiative that Adam is working on to address gaps in current institutional structures. You can read more about them in this white paper that Adam released with Sam Rodriques.
Links
Transcript
[00:00:00]
In this conversation, I talked to Adam marble stone about focused research organizations. What are focused research organizations you may ask. It's a good question. Because as of this recording, they don't exist yet. There are new initiatives that Adam is working on to address gaps. In current institutional structures, you can read more about them in the white paper that Adam released recently with San Brad regens.
I'll put them in the show notes. Uh, [00:01:00] just a housekeeping note. We talk about F borrows a lot, and that's just the abbreviation for focus, research organizations.
just to start off, in case listeners have created a grave error and not yet read the white paper to explain what an fro is. Sure. so an fro is stands for focus research organization. the idea is, is really fundamentally, very simple and maybe we'll get into it. On this chat of why, why it sounds so trivial.
And yet isn't completely trivial in our current, system of research structures, but an fro is simply a special purpose organization to pursue a problem defined problem over us over a finite period of time. Irrespective of, any financial gain, like in a startup and, and separate from any existing, academic structure or existing national lab or things [00:02:00] like that.
It's just a special purpose organization to solve, a research and development problem. Got it. And so the, you go much more depth in the paper, so I encourage everybody to go read that. I'm actually also really interested in what's what's sort of the backstory that led to this initiative. Yeah. it's kind of, there's kind of a long story, I think for each of us.
And I would be curious your, a backstory of how, how you got involved in, in thinking about this as well. And, but I can tell you in my personal experience, I had been spending a number of years, working on neuroscience and technologies related to neuroscience. And the brain is sort of a particularly hard a technology problem in a number of ways.
where I think I ran up against our existing research structures. in addition to just my own abilities and [00:03:00] everything, but, but I think, I think I ran up against some structural issues too, in, in dealing with, the brain. So, so basically one thing we want to do, is to map is make a map of the brain.
and to do that in a, in a scalable high-speed. Way w what does it mean to have a map of the brain? Like what, what would, what would I see if I was looking at this map? Yeah, well, we could, we could take this example of a mouse brain, for example. just, just, just for instance, so that there's a few things you want to know.
You want to know how the individual neurons are connected to each other often through synopsis, but also through some other types of connections called gap junctions. And there are many different kinds of synopsis. and there are many different kinds of neurons and, There's also this incredibly multi-scale nature of this problem where a neuron, you know, it's, it's axon, it's wire that it sends out can shrink down to like a hundred nanometers in [00:04:00] thickness or less.
but it can also go over maybe centimeter long, or, you know, if you're talking about, you know, the neurons that go down your spinal cord could be meter long, neurons. so this incredibly multi-scale it poses. Even if irrespective of other problems like brain, computer interfacing or real time communication or so on, it just poses really severe technological challenges, to be able to make the neurons visible and distinguishable.
and to do it in a way where, you can use microscopy, two image at a high speed while still preserving all of that information that you need, like which molecules are aware in which neuron are we even looking at right now? So I think, there's a few different ways to approach that technologically one, one is with.
The more mature technology is called the electron microscope, electromicroscopy approach, where basically you look at just the membranes of the neurons at any given pixel sort of black or white [00:05:00] or gray scale, you know, is there a membrane present here or not? and then you have to stitch together images.
Across this very large volume. but you have to, because you're just able to see which, which, which pixels have membrane or not. you have to image it very fine resolution to be able to then stitch that together later into a three D reconstruction and you're potentially missing some information about where the molecules are.
And then there's some other more, less mature technologies that use optical microscopes and they use other technologies like DNA based barcoding or protein based barcoding to label the neurons. Lots of fancy, but no matter how you do this, This is not about the problem that I think can be addressed by a small group of students and postdocs, let's say working in an academic lab, we can go a little bit into why.
Yeah, why not? They can certainly make big contributions and have to, to being able to do this. But I think ultimately if we're talking about something like mapping a mouse brain, it's not [00:06:00] going to be, just a, a single investigator science, Well, so it depends on how you think about it. One, one, one way to think about it is if you're just talking about scaling up, quote, unquote, just talking about scaling up the existing, technologies, which in itself entails a lot of challenges.
there's a lot of work that isn't academically novel necessarily. It's things like, you know, making sure that, Improving the reliability with which you can make slices of the brain, into, into tiny slices are making sure that they can be loaded, onto, onto the microscope in an automated fast way.
those are sort of more engineering problems and technology or process optimization problems. That's one issue. And just like, so Y Y Can't like, why, why couldn't you just sort of have like, isn't that what grad students are for like, you know, it's like pipetting things and, doing, doing graduate work.
So like why, why couldn't that be done in the lab? That's not why [00:07:00] they're ultimately there. Although I, you know, I was, I was a grad student, did a lot of pipetting also, but, But ultimately they're grad student. So are there in order to distinguish themselves as, as scientists and publish their own papers and, and really generate a unique academic sort of brand really for their work.
Got it. So there's, there's both problems that are lower hanging fruit in order to. in order to generate that type of academic brand, but don't necessarily fit into a systems engineering problem of, of putting together a ConnectTo mapping, system. There's also the fact that grad students in, you know, in neuroscience, you know, may not be professional grade engineers, that, for example, know how to deal with the data handling or computation here, where you would need to be, be paying people much higher salaries, to actually do, you know, the kind of industrial grade, data, data piping, and, and, and many other [00:08:00] aspects.
But I think the fundamental thing that I sort of realized that I think San Rodriquez, my coauthor on this white paper also realized it through particularly working on problems that are as hard as, as clinic Comix and as multifaceted as a system building problem. I th I think that's, that's the key is that there's, there's certain classes of problems that are hard to address in academia because they're system building problems in the sense that maybe you need five or six different.
activities to be happening simultaneously. And if any, one of them. Doesn't follow through completely. you're sort of, you don't have something that's novel and exciting unless you have all the pieces putting, you know, put together. So I don't have something individually. That's that exciting on my own as a paper, Unless you, and also three other people, separately do very expert level, work, which is itself not academically that interesting.
Now having the connectome is academically [00:09:00] interesting to say the least. but yes, not only my incentives. but also everybody else's incentives are to, to maybe spend say 60% of their time doing some academically novel things for their thesis and only spend 40% of their time on, on building the connectome system.
Then it's sort of, the probability of the whole thing fitting together. And then. We see everyone can perceive that. And so, you know, they basically, the incentives don't align well, for, for what you would think of as sort of team science or team engineering or systems engineering. yeah. And so I'm like, I think, I think everybody knows that I'm actually like very much in favor of this thing.
So, I'm going to play devil's advocate to sort of like tease out. what I think are. Important things to think about. so, so one sort of counter argument would be like, well, what about projects? Like cert, right? Like that [00:10:00] is a government yeah. Led, you should, if you do requires a lot of systems engineering, there's probably a lot of work that is not academic interesting.
And yet, it, it, it happens. So like there's clearly like proof of concepts. So like what what's like. W why, why don't we just have more things like, like certain for, the brain. Yeah. And I think this gets very much into why we want to talk about a category of focused research organizations and also a certain scale, which we can get into.
So, so I think certain is actually in many ways, a great example of, of this, obviously this kind of team science and team engineering is incredible. And there are many others, like LIGO or, or CBO observatory or the human genome project. These are great examples. I think the, the problem there is simply that these, these are multibillion dollar initiatives that really take decades of sustained.
government involvement, to make it happen. And so once they get going, and [00:11:00] once that flywheel sort of start spinning, then you have you have it. And so, and so that, that is a nonacademic research project and also the physics and astronomy communities, I think have more of a track record and pipeline overall.
perhaps because it's easier, I think in physical sciences, then in some of these sort of emerging areas of, of, you know, biology or sort of next gen fabrication or other areas where it's, it's, there's less of a, a grounded set of principles. So, so for CERN, everybody in the physics basically can agree.
You need to get to a certain energy scale. Right. And so none of the theoretical physicists who work on higher energy systems are going to be able to really experimentally validate what they're doing without a particle accelerator of a certain level. None of the astronomers are gonna be able to really do deep space astronomy without a space telescope.
and so you can agree, you know, community-wide that, This is something that's worth doing. And I think there's a lot of incredible innovation that happens in those with focus, research organizations. We're thinking about a scale that, [00:12:00] that sort of medium science, as opposed to small science, which is like a, you know, academic or one or a few labs working together, Or big science, which is like the human genome project was $3 billion.
For example, a scope to be about $1 per base pair. I don't know what actually came out, but the human genome has 3 billion basis. So that was a good number. these are supposed to be medium scale. So maybe similar to the size of a DARPA project, which is like maybe between say 25 and. A hundred or $150 million for a project over a finite period of time.
And they're there. The idea is also that they can be catalytic. So there's a goal that you could deliver over a, some time period. It doesn't have to be five years. It could be seven years, but there's some, some definable goal over definable time period, which is then also catalytic. so in some ways it will be more equivalent to.
For the genome project example, what happened after the genome project where, the [00:13:00] cost of genome sequencing through, through new technologies was brought down, basically by a million fold or so is, is, is, how George Church likes to say it, inventing new technologies, bringing them to a level of, of readiness where they can then be, be used catalytically.
whereas CERN, you know, It's just a big experiment that really has to keep going. Right. And it's also sort of a research facility. there's also permanent institutes. I think there's a, is a, is a, certainly a model that can do team science and, and many of the best in the brain mapping space, many of the sort of largest scale.
connectomes in particular have come either from Janelia or from the Allen Institute for brain science, which are both sort of permanent institutes, that are, that are sort of, nonacademic or semi academic. but that's also a different thing in the sense that it's, it takes a lot of activation energy to create an Institute.
And then that becomes now, a permanent career path rather than sort of focusing solely on what's the shortest path to. To some [00:14:00] innovation, the, the, the permanence. So, so the, the flip side of the permanence is that, I guess, how are you going to convince people to do this, this, like this temporary thing, where.
I think, someone asked on Twitter about like, you know, if it's being run by the government, these people are probably going to get, government salaries. So you're, you're getting a government salary, without the like one upside of a government job, which is the security. so like what, what is the incentive for, for people to, to come do this?
Yeah. And I think, I think it depends on whether it's government or philanthropic, philanthropic fro Faros are also definitely. An option and maybe in many ways more flexible, because the, you know, the government sort of has to, has to contract in a certain way and compete out, you know, contracts in a certain way.
They can't just decide, the exact set of people to do something, for example. So, so the government side has. Both a huge [00:15:00] opportunity in the sense that I think this is a very good match for a number of things that the government really would care about. and the government has, has, has the money, and resources to do this, but philanthropic is also one we should consider.
but in any case, there are questions about who and who will do Froy and, and why. and I think the basic answer though, it, it comes down to, it's not a matter of, of cushiness of the career certainty. it's, it's really, these are for problems that are not doable any other way. this is actually in many ways, the definition is that you're only going to do this.
if this is the only way to do it, and if it's incredibly important. So it really is a, it's a medium scale moonshots. you would have to be extremely passionate about it. That being said, there are reasons I think in approximate sense why one might want to do it both in terms of initiating one and in terms of sort of B being part of them. [00:16:00]
so one is simply that you can do science. that is for a fundamental purpose or, or, or, pure, purely driven toward your passion to solve a problem. and yet can have potentially a number of the affordances of, of industry such as, industry competitive salaries, potentially. I think the government, we have to ask about what the government can do, but, but in a certain philanthropic setting, you could do it another aspect that I think a lot of scientists find.
Frustrating in the academic system is precisely that they have to. spend so much work to differentiate themselves and do something that's completely separate from what their friends are doing, in order to pay the bills basically. So, so if, if you don't eventually go and get your own appealing, you know, Tenure track job or, or so on and so forth.
the career paths available in academia are much, much fewer, and often not, not super well compensated. And, and [00:17:00] so there are a number of groups of people that I've seen in sort of, if you want critical mass labs or environments where they're working together, actually, despite perhaps the. Incentive to, to, differentiate where they're working, does a group of three or four together.
and they would like to stay that way, but they can't stay that way forever. And so it's also an opportunity if you, if you have a group of people that wants to solve a problem, to create something a little bit like, like a seal team. so like when, when I was, I'm not very generally militaristic person, but, when I was a kid, I was very obsessed with the Navy seals.
But, but anyway, I think the seal team was sort of very tight knit. kind of a special forces operation that works together on one project is something that a lot of scientists and engineers I think want. and the problem is just that they don't have a structure in which they can do that. Yeah.
So then finally, I think that, although in many cases maybe essentially built into the structure fro is make sense. We can [00:18:00] talk about this as, as nonprofit organizations. these are the kinds of projects where, you would be getting a relatively small team together to basically create a new industry.
and if you're in the right place at the right time, then after an fro is over, you would be in the ideal place to start. The next startup in an area where it previously, it's not been possible to do startups because the horizons for a venture investment would have been too long to make it happen from the beginning.
Well, that's actually a great transition to a place that I'm still not certain about, which is what happens. After it fro, cause you, you said that it, that it's a explicitly temporary organization. And then, how do you make sure that it sort of achieves its goal, right?
Like, because you can see so many of these, these projects that actually sound really great and they like go in and possibly could do good work and then somehow it all just sort of diffuses. [00:19:00] so, so have you thought about how to sort of make sure that that lives on. Well, this is a tricky thing as we've discussed, in a number of settings.
So, in a, like to maybe throw that question back to you after I answer it. Cause I think you have interesting thoughts about that too, but, but in short, it's, it's a tricky thing. So, so the fro. Is entirely legal focused there isn't, there's no expectation that it would continue, by default and simply because it's a great group of people, or because it's been doing interesting work, it's sort of, it is designed to fulfill a certain goal and it should be designed also from the beginning to have a, a plan of the transition.
Like it could be a nonprofit organization where it is explicitly intended that at the end, assuming success, One or more startups could be created. One or more datasets could be released and then a, you know, a much less expensive and intensive, nonprofits, structure could be be there to [00:20:00] host the data and provide it to the world.
it could be something where. the government would be using it as a sort of prototyping phase for something that could then become a larger project or be incorporated into a larger moonshot project. So I think you explicitly want a, a goal of a finite tune to it, and then also a explicit, upfront, deployment or transition plan, being central to it much more so than any publication or anything.
Of course. At the same time. there is the pitfall that when you have a milestone driven or goal focused organization, that the funder would try to micromanage that and say, well, actually, not only do I care about you meeting this goal, but also I really care that by month six, you've actually got exactly this with this instrument and this throughput, and I'm not going to let you buy this other piece of equipment.
Unless, you know, you show me that, you know, [00:21:00] and that's a problem that I think, we sometimes see with, externalized research models, like DARPA ARPA models, that try to. achieve more coordination and, and, and goal driven among otherwise, somewhat uncoordinated entities like contractors and, and universities that, that are working on programs, but then they, they, they, they achieve that coordination by then, managing the process and, with an fro, I think it will be closer to.
You know, if you have a series, a investment in the startup, you know, you are reporting back to your investors and, and they, they, at some level care, you know, about the process and maybe they're on your board. but ultimately the CEO gets to decide, how am I going to spend the money? And it's extremely flexible to get to the goal.
Yeah. Yeah. The, the micromanage, like [00:22:00] figuring out how to avoid, Micromanagement seems like it's going to be really tricky because it's sort of like once you get to that amount of money, I like, have you, have you thought about, like how, like, if you could do some kind of like actually, well, I'll, I'll give her the, the, the, the, the, the thing that the cruxy thing is like this, I think there's a huge amount of trust that needs to happen in it.
And what I'm. like I constantly wonder about is like, is there this like fundamental tension between the fact that, especially with like government money, we really do want it to be transparent and well-spent, but at the same time, in order to sort of do these like knowledge frontier projects, sometimes you need to do things that.
Are a little weird or like seem like a waste of money at the time, if you're not like intimately connected. and so there's, there's this sort of tension [00:23:00] between accountability and, Sort of like doing the things that need to get done. I agree with that and Efros, we're going to navigate that. Yeah.
I agree with that. And I think it relates to a number of themes that you've touched on and that we've discussed with, which has sort of, has to do with the changing overall research landscape of, in what situations can that trust actually occur, you know, in bell labs, I think there was a lot of trust. throughout, throughout that system.
And as you have more externalized research, conflicting incentives and so on it, it's, it's hard. It's hard to obtain that trust. startups of course, can align that financially, to a large degree. I think there are things that we want to avoid. so one of the reasons I think that these need to be scoped as.
Deliverables driven and roadmaps, systematic projects over finite periods of time, is to avoid, individual [00:24:00] personalities, interests, and sort of conflicting politics, ending up. Fragmenting that resource into a million pieces. So, so I think this is a problem that you see a lot with billion dollar scale projects, major international and national initiatives.
Everybody has a different, if you say, I want this to be, to solve neuroscience, you know, and here's $10 billion. Everybody has a different opinion about what solved neuroscience is. And there's also lots of different conflicting personalities and, and leadership there. So I think for an fro, there needs to be an initial phase, where there's a sort of objective process of technology roadmapping.
And people figure people understand and transparently understand what are the competing technologies? What are the approaches? What, what are the risks? And you understand it. and you also closely understand the people involved. but importantly, the people doing that roadmapping and sort of catalyzing the initial formation of that [00:25:00] fro need to have a somewhat objective perspective.
It's not just funding my lab. It's actually, you, you want to have vision, but you, you need to. Subjected to a relatively objective process, which, which is hard because you also don't want it to be a committee driven consensus process. You want it to be active, in, in a, in a systematic, analysis sense, but, but not in a, everyone agrees and likes it, you know, emotionally sense.
and so that, that's a hard thing. but you need to establish it's that trust upfront, with, with the funder, And that's a hard process and it gets a hard process to do as a large government program. I think DARPA does it pretty well with their program managers where a program manager will come in and they will pitch DARPA on the idea of the program.
there'll be a lot of analysis behind it and, but then once, once they're going, that program manager has tremendous discretion, and trust. To how they actually run that [00:26:00] project. And so I think you need something like a program manager driven process to initiate the fro and figure out is there appropriate leadership and goals and our livable as reasonable, Yeah, that seems the way, at least the way that it's presented in the paper, it, it feels a little bit chicken and egg in that.
so with DARPA, DARPA is a sort of permanent organization that brings in program managers. And then those programmers program managers then go, start programs, whereas, The look at fro it seems like there's this chicken and egg between like, you sort of, you need someone spearheading it. It seems like, but then it, you sort of like, it, it seems like it will be very hard to get someone who's qualified to, to spearhead it, to do that before you have funding, but then you need someone spearheading it in order to get that [00:27:00] funding.
yeah. Like, yeah. How, how are you thinking about. Cracking that that's, that's sort of the motivation for me behavior over the next year or two, is that I'm trying to go out and search for them. And, a little bit of it is from my own creativity, but a lot of it is going out and talking to people and try and understand what the best ideas.
Here would be, and who are the networks of, of human beings behind those ideas, and trying to make kind of a prioritized set of borrows. Now, this kind of thing would have to be done again, I think to some degree, if there was a, larger umbrella program that someone else wanted to do, but, I'm both trying to get a set of, of exemplary.
And representative ideas and people together, and try to help those people get funding. You know, I think there can be a stage process. I agree that, in the absence of a funder showing [00:28:00] really strong interest, people committing, to really be involved is difficult, because it is a big change to people's normal.
Progression through life to do something like that. but just like with startups, to the extent that you can identify, someone who's. We spiritually just really wants to do this and we'll kind of do anything to do it, the sort of founder type, and also teams that want to behave like that. that's obviously powerful, and also ideas where there's a kind of inevitability, where based on scientific roadmapping, it, it just has to happen.
There's no way, you know, for neuroscience to progress unless we get better. Connectomics and I think we can go through many other fields where, because of. The structures we've had available and just the difficulty of problems now, where arguably Faros are needed in order to make progress in fields that people really care about.
So, so I think you can get engagement at the level of, of discussion, and, and, and starting to nucleate [00:29:00] people. But, but there is a bit of a chicken and egg problem. In the sense that it's, it's not so much as here's an fro, would you please fund to me it's we need to go and figure out where there might be Faros to be had, and then who is interested in those problems as well to, to fund and support those things.
So, yeah. So I guess to recap what I see your process that is, is that you're going out, you're sort of really trying to. Identify possible people possible ideas, then go to funders and say, here, like sort of get some, some tentative interest of like, okay, what, which of these things might you be interested in if I could get it to go further and then you'll circle back to.
the, the people who might be interested in sort of say like, okay, I have someone, a funder who's potentially interested. Can we [00:30:00] sort of like refine the idea? and then sort of like, like you will drive that loop hopefully to, Getting a, an fro funded that's right. And there's, there's further chicken and egg to it.
that has to be solved in the sense that, when you go to funders and you say, why, you know, I have an idea for an fro. We also need to explain what an fro is, right? in a way that both, engages people in creating these futuristic models, which many people want to do, While also having some specificity of, of what we're looking for and what, what, what we think is as possible.
So, and then the same on, on the, on the side of, of scientists and engineers and entrepreneurs all over the world, who, you know, have the ideas certainly, but most of those ideas have been optimized to hit, the needs of existing structures. So, so we are, we are trying to, I think, broker between those, And [00:31:00] then start prototyping a few.
but the, you know, the immediate thing I think is to make, w Tom Coolio has referred to a catalog, a Sears catalog of moonshots. and so we're trying to make a catalog of, of moonshots that fit the fro category. but that sounds like the perfect name for this podcast, by the way. the cataloging mood child, like, you're kind of kind of cataloging moonshots and ways to get moonshots and yeah, absolutely.
Yeah. and so I guess another sort of, thing that I've seen, and I'm not sure, it's almost like for people like a lot of people who like really want. Who like sees something as inevitable and they really want to get it done. In sort of like the current environment we're recording in October, 2020.
there's. There's sort of this perception that capital is really cheap. [00:32:00] you know, there's a lot of venture capitalists there. They're pretty aggressive about funding and one could make an argument that, if it's, if it, it really is going to be inevitable and it really is going to start a new industry.
Then that is exactly where venture capital funding should come in. And I do see this a lot where people, you know, it's like they have this thing that they really want to see exist and they, you know, come out of the lab and it started a company that's sort of extremely common. so. I guess, like, what almost would you say to someone who you see doing this that you think maybe should do an fro instead?
Yeah, that's a great question. I mean, I think it's a complicated question and obviously, you know, we got to see VC also, you know, obviously VC backed, you know, innovation is, is, is one of, if not sort of the key, [00:33:00] Things that is driving technology right now. So, so I'm in no way saying that fro is, are somehow superior to two startups, in any generalized way.
So I think that things that can be startups and are good as startups should be startups and people, if you have an idea that could be good for a startup, I think you should go do it. Generally speaking. But, there, there are a few considerations, so yeah. So I think you can divide it into categories where VCs, no, it's not a good idea for startups.
And therefore won't talk to you, in cases where VCs don't always know whether or not it's good for a startup or whether there's a way that you could do it as a startup, but it would involve some compromise that is actually better not to make, even potentially for the longterm. economic prospects of, of an area.
So things that can happen, would be, if you have something that's basically meant to be a kind of platform technology or which you [00:34:00] need to develop a tool or a platform in order to explore a whole very wide space of potential applications. maybe you have something like a new method of microscopy or something, or a new way to measure proteins in the cell or things like that, that, you know, you could target it to a very particular, if you want product market fit application, where you would be able to make the most money on that and get the most traction, the soonest.
Yeah. Sometimes people call this, you know, the, the, the, the sort of Tesla Roadster, equivalent. You want to guys as quickly as you can to the Tesla Roadster. And I think generally, what people are doing with, with that kind of model, where you take people that have science, to offer, and you say what's the closest fastest you can get, to a Tesla Roadster that lets you it lets you build, get, get revenue and start, start being financially sustainable and start building a team, to go further.
generally that's really good. and generally we need more scientists to learn how to do that. it'd be supported to do that, but, [00:35:00] sometimes you have things that really are meant to be. either generalized platforms or public goods, public data, or knowledge to underlie an entire field. And if you work to try to take the path, the shortest path to the Roadster, you would end up not producing that platform.
You would end up, producing something that is specialized to compete in that lowest hanging fruit regime, but then in the, in, in doing so you would forego the more general larger. Larger thing. And, you know, Alan Kay has, has the set of quotes, that Brett Victor took is linked on his website. and I think Alan K meant something very different actually, when he said this, but he's, he refers to the dynamics of the trillions rather than the billions.
Right. and this is something where in, and we can talk about this more. I'd be curious about your thoughts on that, but something like the transistor. You know, you, you could try to do the transistor as a startup. and maybe at the time, you know, the best application for transistors would have been [00:36:00] radios.
I don't think like that. I think it was, it was guiding a rockets. Yeah. So you could have, you could have sort of had had a transistors for rockets company and then tried to branch out into, becoming Intel. You know, but really, given the structures we had, then the transistor was allowed to be more of a, a broadly, broadly explored platform.
yeah, that, that progressed in a way where we got the trillions version. And I worry sometimes that even some startups that have been funded at least for a seed round kind of stage, and that are claiming that they want to develop a general platform are going to actually struggle a little bit later.
when investors, you know, see that, see that they would need to spend way more money to build that thing. then the natural shortest path to a Roadster, or another words the Roadster is, is, somehow illusory. yeah. Yeah, this [00:37:00] is, this is a. Sort of like a regime that I'm really interested in and a, just on the transistor example, I've, I've looked at it.
So just the, the history is that it was developed at bell labs, in order to prevent a T and T from being broken up, bell labs had to, under strictly licensed a bunch of their innovations, including the transistor William Shockley went off and, Started, chocolate semiconductor, the traders eight then left and started, Fairchild and then Intel.
And, believe that that's roughly the right history. but the, the really interesting thing about that is to ask the question of like, one, what would have happened if, bell labs had exclusive license to the transistor and then to what would have happened if they had like exclusively licensed it to, Shockley semiconductor.
And I think I would argue in both of those situations, you don't [00:38:00] end up. Having the world we have today because I fell labs. It probably goes down this path where it's not part of the core product. and so they just sort of like do some vaguely interesting things with it, but are never incentivized to like, you know, invent, like the, the planner processing method or anything.
Interesting. yeah. Yeah. And so I guess where I'm. Go. And then like at the same time, the interesting thing is like, so Shockley is more, akin to like doing a startup. Right. And so it's like, what if they had exclusive license to it? And the, what I would argue is actually like that also would've killed it because, you have like, they had notoriously bad management.
And so if you have this, this company with. And like the only reason that, the trader could go and start a Fairchild was because they, that was, that was [00:39:00] an open license. So this is actually a very long way of asking the question of, if F borrows are going to have a huge impact, it seems like they should default to.
Really being open about what they create from like IP to data. but at the same time, that sort of raises this incentive problem where, people who think that they are working on something incredibly valuable, should want to do a startup. And then. And so there, and then similarly, even if they'd be like that sort of couldn't be a thing, they would want to privatize as much of the output of an fro.
and so which. Maybe necessary in order to, to get the funding to make it happen. So I guess like, how are you thinking about that tension? That was a very long winded. Yeah. [00:40:00] Yeah. Well, there's, there's a lot, a lot there, I think, to loop back to you. So, so I think, right, so, so this idea that we've talked a bit about as sort of default openness, so, so things that can be open for maximum impact should be open.
there are some exceptions to that. So, so if, And it's also has to do partly with how you're scoping the problem. Right? So, so rather than having an SRO that develops drugs, let's say, because drugs really need to be patentable, right. In order to get through clinical trials, we're talking about much more money than the fro funding, you know, to do the initial discovery of a target or something.
Right. So to actually bring that to humans, you know, you need to have the ability to get exclusive IP. for downstream investors and pharma companies that that would get involved in that. so there are some things that need to be patented in order to have to have their impact. but in general, you, you want, I think fro problems to steer themselves to things where indeed.
it can be maximally open and maybe, maybe you, you provide [00:41:00] a system that can be used to, to, or underlie the discovery of a whole new sets of classes of drugs and so on. But you're not so much focused on the drugs themselves. Now, that being said, right. if I invest in an SRO, and I've enabled this thing, right.
It kind of would make sense for the effort, you know, maybe three of the people of, of, of, of 15 in the fro will then go and start a company afterwards that then capitalizes on this and actually develops those drugs or what have you, or it takes it to the next stage. And gosh, it would really make sense if I had funded in fro.
that's, those people would like to take me as a sort of first, first, first refusal to get a good deal on, on investing in this startup, for example. Right. so I think there are indirect network-based, or potentially even legal based, structure, structure based ways to both incentivize the investors and, But it's, it's a weaker, admittedly weaker, incentive financially than, [00:42:00] than, than the full capture of, of, of something.
But then, but then there's, I think this gets back to the previous discussion. So which is sort of the trillions rather than the billions. So if you have something where maybe there are 10 different applications of it, Right in 10 different fields. you know, maybe, maybe we have a better way to measure proteins and based on this better way to measure proteins, we can do things in oncology and we can do things in Alzheimer's and we can do things in a bunch of different directions.
We can do things in diagnostics and pandemic surveillance, and so many fields that one startup, It would be hard even to design, to start, if that could capture all of that value just as it would have been hard to design sort of transistor incorporated. Right. Right. given that, I think there's, there's a lot of reason to.
To do an fro and then explore the space of applications. Use it as a means to explore a full space in which you'll then get [00:43:00] 10 startups. so if I'm the investor, I might like to be involved in all 10 of the new industry, right. And the way to do that would be to create a platform with which I can explore, but then I have a longer time horizon.
Cause I have to first build the thing. Then I have to explore the application space and only then. do I get to invest in a specific verticals, right? Yeah. I think the, the two sort of tricky questions that I, I wonder about what that is one. So you mentioned like, Oh, there's 15 people in an fro, three of them go off to start a startup.
What about those other 12 people? Like, I, I assume that they might be a little bit frustrated if, if that happens, Yeah, because like, like they, they did, they did help generate that value in it. It sort of gets into two questions of like capturing, like sort of kicking back, value generated by research in general, but like, yeah, it could, it could, it could be all 15 people, you know, we saw something [00:44:00] similar with open AI, you know, in a way, for example, converting, you know, into, into a, for profit or at least a big arm of it being, being the for-profit, and keeping all the people.
Right. So you, you, you, you can imagine, just blanket converting. but yeah, I think, I think it's sort of, In the nature of it, that these are supposed to be things that open up such wide spaces that there's, there's sort of enough for everyone, but no, no, no one person necessarily one startup would completely capture.
And I think that's true for clinic Comix too, for example. Right. So if you had really high throughput clinical, connectomics just, just to keep going on this example, that's a great example of perfect. It's a good thing as a good example. It's not. Depending on the details, whether this is exactly the first fro or not.
I think it's totally, totally other issue, but, but. Connectomics there's potentially applications for AI and you know, how, how the neurocircuits work, and sort of fundamental, funding. Mental is a brain architecture and intelligence. although there's a bunch of ranges of the sort of uncertainty of exactly what that's going to be.
So it's hard to sort of [00:45:00] know it until you see the data. There's also potentially applications for something like drug screening, where you could put a bunch of different, Kind of some CRISPR molecules or drug perturbations on, on a, on a brain and then look at what each one does to their, the synopsis or, and look at that in a, in a brain region specific way and sort of have ultra high, but connect to them based drug screening.
Neither of those are things you can start a start up until you have connected. Right. working. but so anyway, so maybe three people would start an AI company and maybe those would be the very risk tolerant ones. and then three would start at, you know, a crisper drug company and, and, and, three would just do, do fundamental neuroscience with it and, take those capabilities and, and, and go, go back into the university system or so on and yeah.
And start using that. Yeah. And the, the sort of the other related to. like creating value with it. there's, there's a little like uncut discomfort that like even I have [00:46:00] with, say like philanthropic or government funding, then going to fund a thing that proceeds to make a couple of people very wealthy.
Which like, and like, there's very much arguments on both sides, right. Where it's like, it'll generate a lot of good for the world. and, and all and, and such. so, so like, I guess what would you say? I guess like, as a, as a, like, if I were a very wealthy philanthropist and I'm like, do it, like, you know, it's like, I'm just giving away money so that these people can.
Yeah, the company is a complicated thing. Right? How much, how many further rich people, you know, did the Rockefeller foundation, you know, investing in the basics of molecular biology or things like that ended up generating? I mean, I think that, I think you, I think in some way the government does want to end up is they want the widely distributed benefit.
And I think everything that should be an SRO should have widely distributed benefits. It shouldn't just [00:47:00] be a kind of, A startup that just, just enhances one, one person. It should be something that really contributes very broadly to economic growth and understanding of the universe and all that. But it's almost inevitable.
I think that, if you create a new industry, you're gonna, you know, you're gonna, you're gonna feel it going to be some more written about rich successful people in that industry. And they're probably going to be some of the people that were involved. Early and thinking about it for the longest and waiting for the right time to really enter it.
And so, yeah, that's a really good point. I guess the, then the question would be like, how do you know, like, like what are, what are sort of a, the sniff test you use to think about whether something would have broadly distributed benefits? That's a great question. Cause it's like connect to them. It seems like fairly clear cut or, or generating sort of like a massive data set that you then open up. Feels very [00:48:00] clear. Cut. it's. We we've talked before about that, like fro is, could like scale up a process or build a proof of concept of, of a technology.
and it, it seems like that it's less clear cut how you can be sure that those are going to, like if they succeed. Yeah. I mean, there are a few different frames on it, but I mean, I think one is, FRS could develop technologies that allow you to really reduce the cost of having some. Downstream set of capabilities.
so, you know, if, just to give you an example, right? If, if we had, much lower costs, gene therapies available, right? So, so sometimes when drug prices are high, you know, this is basically it's recouping these very large R and D costs and then there's competition and, and, and profit and everything involved.
you know, there was the marching squarely situation and, you know, there's a bunch of, sort of. What was that? there was, remember the details, but there [00:49:00] was some instance within which, a financially controlling entity to sort of arbitrarily bumped drug prices way high, right. A particular drug. and then w was, you know, was regarded as an evil person then, and maybe that's right.
but anyway, there are some places I think, within the biomedical system where you can genuinely reduce costs for everyone. Right. and it's not simply that I, you know, I make this drug and I captured a bunch of value on this drug, but you know, it's really, it should be available to everyone and I'm just copying there.
There's genuine possibility to reduce costs. So if I could reduce the cost of, of the actual manufacturing of. The viruses that you use for gene therapy, that's a, that's a process innovation. that would be, you could order as a magnitude drop the cost of gene therapy. If you could figure out what's going on, in the aging process and what are the real levers on a single, you know, biological interventions that would prevent multiple age related diseases that [00:50:00] would massively drop the cost.
Right? So those, those are things where, Maybe even in some ways it would be threatening, to some of, some of the pharma companies, you know, that, that work on specific age related diseases, right? Because you're going to have something that, that replaced, but this is, this is what, you know, things that are broad productivity improvements.
And I think economists and people very broadly agree that, that the science and technology innovations, For the most part. although sometimes they can be used to in a way that sort of, only benefits, a very small number of people that generally speaking there's a lot you can do, with technology that will be extremely broadly shared in terms of benefit, right?
Yeah. Yeah. I mean, I, I do actually, like I agree with that. I'm, I'm just, I'm trying to represent as much skepticism as, as possible. Definitely. I know you agree with that.
And actually, another thing that I have no idea about which I'm really interested in is as you're going and sort of creating this, [00:51:00] this moonshot catalog. how do you tell the difference between people who have these really big ideas who are like hardcore legit? but like maybe a little bit crazy.
And then people who are just crackpots. Yeah, well, I don't claim to be able to do it in every field. and, and I think there's a reason why I've, I'm not trying to do a quantum gravity, fro you know, both, both, because I don't know that that's, you know, I think that's maybe better matched for just individual.
Totally. Open-ended Sunday, you know, fun, brilliant people for 30 or 40 year long period to just do whatever they want. Right. Yeah. For quantum gravity, rather than directed, you know, research, but, But also because there's a class of problem that I think requires a sort of Einsteinian type breakthrough in fr fro is, are not, not perfect for that in terms of finding people.
I mean, I, I find that, there's a lot of pent up need for, this is that's my preliminary feeling. and you can see there's a [00:52:00] question of prioritizing, which are the most important, but there's a huge number of. Process innovations or system building innovations that are needed across many, many fields.
And you don't need to necessarily have things that even sound that crazy. There are some that just kind of just make sense, you know, are, are very simple. You know, we here, here in our lab, we have this measurement technology, but we, you know, we can only have the throughput of one cell, you know, every, every few weeks.
And if we could build the system, we could get a throughput of, you know, A hundred thousand cells, you know, every month or something. Right. there are some, there's some sort of ones that are pretty obvious, or where there's an obvious inefficiency. In kind of, how things are structured. Like every, every company and lab that's that's modeling fusion reactors, and then also within the fusion reactor, each individual component of it, like the neutrons in the wall versus the Plaza and the core, those are basically modeled with different.
Codes many of which are many [00:53:00] decades old. So there's sort of an obvious opportunity to sort of make like a CAD software for fusion, for example, you know, that the, the, it doesn't, it's not actually crazy. It's actually just really basic stuff. In some cases, I think they're ones where we'll need more roadmapping and more bringing people together to really workshop the idea, to really have people that are more expert than me say, critique each other and see what's.
Really going on in the fields. and I also rely on a lot of outside experts. if I have someone comes with an idea, you know, for, for energy, you know, and I'm talking to people that are like former RPE program managers or things like that, that, that know more of the questions. so I think we can, we can, we can do a certain amount of, of due diligence on ideas and.
and then there are some that are, that are really far out. you know, we both have an interest in atomically precise manufacturing, and that that's when, where we don't know the path I think, forward. and so that's maybe a pre fro that's something where you [00:54:00] need a roadmapping approach, but it's maybe not quite ready to, to just immediately do an fro.
Yeah, no, that's, you sort of hit on a really interesting point, which is that. when we think of moonshots, it's generally like this big, exciting thing, but perhaps some of the most valuable is will actually sound incredibly boring, but the things that they'll unlock will be. Extremely exciting. yeah, I think that's true.
And, and you have to distinguish there's there's boring. Right? So, so I think there's, there's some decoupling of exactly how much innovation is required and exactly how important something is. And also just how much brute force is required. So I think in general, our system might under weight, the importance of brute force.
And somewhat overweight the importance of sort of creative, individual breakthrough thinking. at the same time, there are problems where I think we are bottlenecked by thinking I'm like really how to do something, not just to [00:55:00] connect them of brain, but how do you actually do activity map of entire brain?
You actually need to get a bunch of physicists together and stuff to really figure out what's, you know, there's a level of thinking that is not very non-obvious similarly for like truly next gen fabrication. You really, really, really need to do the technology roadmapping approach. And that's a little different than the fro.
And in some cases there may be a, as we discussed, I think in the past, there was sort of a, a continuum potentially between DARPA type programs or programs that would start within the existing systems and try to catalyze the emergence of ideas and discoveries. And then fro is, which are a bit, a bit more cut and dry.
And in some cases, even you could think of it as boring. but just very important. how do we prevent Faros from becoming a political football? because you see this all the time where, you know, a Senator will say, well, like I'll sponsor this bill, as long as we mandate that.
50% of the work has to happen in my particular state or [00:56:00] district. and, and I imagine that that would be counterproductive towards the goals of . so do you, do you have any sense of like how to, how to get around that
probably much easier in philanthropic setting than governments? Although I think I'm overall, I'm, I'm sort of optimistic that, if. If the goals are made very clear, the goal is disruptive, you know, multiplicative improvements in scientific fields. that's the primary goal. It needs to be managed well.
so it's not either about the individual peoples, if you want academic politics and also that it doesn't, doesn't become about sort of, you know, districts, congressional districts, or all sorts of other things. I think there's a certain amount of complexity, but the other, the other thing is. I think there's really amazing things to be done in all sorts of places and by all sorts of people that are not necessarily identified as, as the biggest egos or the largest cities also, although certainly there are hubs that [00:57:00] matter.
yeah. Cool. I think so. I think those are all like the actual questions I have. Is there anything you want to talk about that we have not touched on? Yeah, that's a good question. I mean, how does this fit into two things that you're thinking about, in terms of your overall analysis of the research system, then, do you think this, what is this leave unsolved as well?
if, even if we can get some big philanthropic and government, donors. Yeah. So, so there are sort of two things that I. see it not covering. And so the, the first that you you've sort of touched on is that there are, some problems that still like don't fit into academia, but are not quite at the point where they're ready to be at fro.
And so, they need, the, the like mindset of the fro without. Having this sort of, cut and dryness [00:58:00] that you need to sort of plunk down, like have the confidence to plunk down $50 million. so, so we need sort of a, a, what I would see as a sustainable, way of. Getting to the point of fro type projects.
And as you know, I'm spending a lot of time with that. and then sort of a, the other thing that I've realized is that when, when people, we sort of have these discussions that are like research is broken, I think what we're actually talking about is, is sort of two really separate phenomenon. So, what we've been talking about, like Efros, Are really sort of sitting in like the Valley of death where it's like helping bridge that.
but I think that at the same time, there there's like what I would call like the, the Einstein wouldn't get in any funding problem, which is, as you alluded to there, there are some of these things, like some of the [00:59:00] problems with research that we talk about are just about, The sort of conformity and specialization of really idea based exploratory, like completely uncertain research.
And that's also really important, but I, I think it's what we don't do is, is, is sort of like separate those two things out and say like, these are both fall under the category of research, but are in fact. Extremely different processes. They require very different solutions. Yeah. Actually let me, let me, since you mentioned that, and since we are here together on the podcast, I agree with that and I, I have some things to say about that as well.
So, so I think that the fro is indeed only address, or are designed to address this issue of sort of system building. problems that have a sort of catalytic nature and are a particular kind of pre-commercial stage. Right? So in some ways, [01:00:00] even though I'm so excited about borrows and how much they can unlock, because I think that this is one of two or three categories that has been, you know, under emphasized by current systems or has systems currently have struggled with it.
there are these others. So, so I think that. The, the supporting the next Einstein and people that may have also have just be cognitively socially in any other number of ways, just different and weird and not good at writing grants. You know, not good at competing. Maybe not even good at graduating undergrad.
Yeah. You know, I'm running a lab who are, are brilliant and because the system now. Has proliferate in terms of the number of scientists. it's very competitive and, and there is a, there's a lot of need to sort of filter people based on credentials. So there's this sort of credential there's people that don't fit with perfectly with credentials or with a sort of monoculture of who is able to get NSF grants and go through the university system and [01:01:00] get the PhD and all those different Alexey goosey has this nice blog post is oriented toward biomedical, but saying basically that in order to get through the system, you need to do 10 or 15 things simultaneously.
Well, and also be lucky. And maybe we want to be looking for some people that are only able to do three of those things about, but are orders of magnitude better than others, then there's people even who have done well with those things, but still don't have the funding or sort of sustained ability, to, to pursue their own individual ideas over decades.
even if they do get tenure or something, because the grant system is based on peer review and is, is sort of filtering out really new ideas, for whatever reason, There's kind of the broader issue that Michael Nielsen has talked about, which is sort of the idea that too much funding is centralized in a single organizational model.
So particularly the NIH, the NIH grant is kind of hegemonic as, as, as a structure and as a peer review mechanism. then I think we need more [01:02:00] DARPA stuff. We probably need more darker agencies for other problems. Even though I've, I've sort of said that I think Rose can solve some problems that DARPA DARPA will struggle with.
Likewise, DARPA walls solve problems that fro may struggle with. particularly if there's a very widely distributed expertise across the world that you need to bring together in a, some transient, interesting way, for a little bit more discovery oriented, perhaps in Faros and less deliverable oriented or team oriented.
And then there's even bigger things we need, you know, like we need to be able to create, you know, a bell labs for energy, you know, or sort of something even bigger than fro. so yeah, I think the thing that you're, you're getting at that I is, is sort of simple, but under done is actually analyzing like what the activity is and what.
How to best support it. Yep. Which is instead of just saying [01:03:00] like, ah, there's some research let's give some money to the research and then magical things will happen actually saying like, okay, like, like how does this work? Like what, and then what can we do for these, these specific situation? Yes. I think as you've identified.
Like there's both on the one hand, there's the tendency to micromanage research and say, research has to do this, this with this equipment and this timescale it's entirely, this is sort of subject to milestone. And on the other hand is research is this magical thing. We have no idea. but just. Let other scientists, peer review each other, and just sort of give as much money to it as we can.
and then we see what happens. Right. And I think neither of those, is a, is a good design philosophy, right? Yeah. Yeah. And I think it involves people like thinking it's it's uncomfortable, but like, like thinking and learning about. How, how did you think then understanding how it could, how it could be different?
[01:04:00] How it's not a it's it's a system. Kevin has felt set, said it said it well. And so in some ways it's been designed, but really our scientific systems are something that has evolved into large degree. No one has designed it. It's not. Something that's designed to be optimal is it's a, it's a emergent property of many different people's incentives.
And, if we actually try to apply more design thinking, I think, I think that can be good as long as we're not over overconfident in saying that there's one model for everyone. Yeah. I think that the trick to, sort of fixing. Emergent systems is to like, basically like do little experiments, poking at them.
And that's, that's very much what I see getting fro is going okay. It's like, you're not saying, Oh, we should like dismantle the NSF and have it all be . Okay. Let's do a couple of these. See what happens. That's right. It's I think it's inherently a small perturbation and it it's. And I [01:05:00] think DARPA, by the way is a similar thing.
It's sort of dark. You wouldn't need DARPA. If everything else was already sort of efficient, right. Given that things are not perfectly efficient, Darko has all these, all these sort of this niche that it fills. I think similarly Faros, they can only exist. if you also have a huge university system and you also have companies that that doesn't make sense, otherwise it's, it's a perturbation, but as we, I think it's a perturbation in which you unlock a pretty big pressure stream sort of behind it when you open it up.
So. Excellent. Well, I think that's, that's actually a great place to close. I guess the last question would be, Like, if people are interested in, in Faros, especially like funding or running one, what is the best way for them to reach you? Well, they can, they can talk to me or they can talk to you.
my email has, is prominently listed on my website. Twitter is great. and that, yeah, I really interested in, people that have a kind of specificity [01:06:00] of, of, of what they want of, you know, here here's, here's what I would do, very specifically, but I'm also interested in talking to people that, See problems with the current systems and want to do something and want to learn about, other highly specific fro ideas that others might have, and how to enable those.
50 episodes
Manage episode 275328183 series 2470122
A conversation with Adam Marblestone about his new project - Focused Research Organizations.
Focused Research Organizations (FROs) are a new initiative that Adam is working on to address gaps in current institutional structures. You can read more about them in this white paper that Adam released with Sam Rodriques.
Links
Transcript
[00:00:00]
In this conversation, I talked to Adam marble stone about focused research organizations. What are focused research organizations you may ask. It's a good question. Because as of this recording, they don't exist yet. There are new initiatives that Adam is working on to address gaps. In current institutional structures, you can read more about them in the white paper that Adam released recently with San Brad regens.
I'll put them in the show notes. Uh, [00:01:00] just a housekeeping note. We talk about F borrows a lot, and that's just the abbreviation for focus, research organizations.
just to start off, in case listeners have created a grave error and not yet read the white paper to explain what an fro is. Sure. so an fro is stands for focus research organization. the idea is, is really fundamentally, very simple and maybe we'll get into it. On this chat of why, why it sounds so trivial.
And yet isn't completely trivial in our current, system of research structures, but an fro is simply a special purpose organization to pursue a problem defined problem over us over a finite period of time. Irrespective of, any financial gain, like in a startup and, and separate from any existing, academic structure or existing national lab or things [00:02:00] like that.
It's just a special purpose organization to solve, a research and development problem. Got it. And so the, you go much more depth in the paper, so I encourage everybody to go read that. I'm actually also really interested in what's what's sort of the backstory that led to this initiative. Yeah. it's kind of, there's kind of a long story, I think for each of us.
And I would be curious your, a backstory of how, how you got involved in, in thinking about this as well. And, but I can tell you in my personal experience, I had been spending a number of years, working on neuroscience and technologies related to neuroscience. And the brain is sort of a particularly hard a technology problem in a number of ways.
where I think I ran up against our existing research structures. in addition to just my own abilities and [00:03:00] everything, but, but I think, I think I ran up against some structural issues too, in, in dealing with, the brain. So, so basically one thing we want to do, is to map is make a map of the brain.
and to do that in a, in a scalable high-speed. Way w what does it mean to have a map of the brain? Like what, what would, what would I see if I was looking at this map? Yeah, well, we could, we could take this example of a mouse brain, for example. just, just, just for instance, so that there's a few things you want to know.
You want to know how the individual neurons are connected to each other often through synopsis, but also through some other types of connections called gap junctions. And there are many different kinds of synopsis. and there are many different kinds of neurons and, There's also this incredibly multi-scale nature of this problem where a neuron, you know, it's, it's axon, it's wire that it sends out can shrink down to like a hundred nanometers in [00:04:00] thickness or less.
but it can also go over maybe centimeter long, or, you know, if you're talking about, you know, the neurons that go down your spinal cord could be meter long, neurons. so this incredibly multi-scale it poses. Even if irrespective of other problems like brain, computer interfacing or real time communication or so on, it just poses really severe technological challenges, to be able to make the neurons visible and distinguishable.
and to do it in a way where, you can use microscopy, two image at a high speed while still preserving all of that information that you need, like which molecules are aware in which neuron are we even looking at right now? So I think, there's a few different ways to approach that technologically one, one is with.
The more mature technology is called the electron microscope, electromicroscopy approach, where basically you look at just the membranes of the neurons at any given pixel sort of black or white [00:05:00] or gray scale, you know, is there a membrane present here or not? and then you have to stitch together images.
Across this very large volume. but you have to, because you're just able to see which, which, which pixels have membrane or not. you have to image it very fine resolution to be able to then stitch that together later into a three D reconstruction and you're potentially missing some information about where the molecules are.
And then there's some other more, less mature technologies that use optical microscopes and they use other technologies like DNA based barcoding or protein based barcoding to label the neurons. Lots of fancy, but no matter how you do this, This is not about the problem that I think can be addressed by a small group of students and postdocs, let's say working in an academic lab, we can go a little bit into why.
Yeah, why not? They can certainly make big contributions and have to, to being able to do this. But I think ultimately if we're talking about something like mapping a mouse brain, it's not [00:06:00] going to be, just a, a single investigator science, Well, so it depends on how you think about it. One, one, one way to think about it is if you're just talking about scaling up, quote, unquote, just talking about scaling up the existing, technologies, which in itself entails a lot of challenges.
there's a lot of work that isn't academically novel necessarily. It's things like, you know, making sure that, Improving the reliability with which you can make slices of the brain, into, into tiny slices are making sure that they can be loaded, onto, onto the microscope in an automated fast way.
those are sort of more engineering problems and technology or process optimization problems. That's one issue. And just like, so Y Y Can't like, why, why couldn't you just sort of have like, isn't that what grad students are for like, you know, it's like pipetting things and, doing, doing graduate work.
So like why, why couldn't that be done in the lab? That's not why [00:07:00] they're ultimately there. Although I, you know, I was, I was a grad student, did a lot of pipetting also, but, But ultimately they're grad student. So are there in order to distinguish themselves as, as scientists and publish their own papers and, and really generate a unique academic sort of brand really for their work.
Got it. So there's, there's both problems that are lower hanging fruit in order to. in order to generate that type of academic brand, but don't necessarily fit into a systems engineering problem of, of putting together a ConnectTo mapping, system. There's also the fact that grad students in, you know, in neuroscience, you know, may not be professional grade engineers, that, for example, know how to deal with the data handling or computation here, where you would need to be, be paying people much higher salaries, to actually do, you know, the kind of industrial grade, data, data piping, and, and, and many other [00:08:00] aspects.
But I think the fundamental thing that I sort of realized that I think San Rodriquez, my coauthor on this white paper also realized it through particularly working on problems that are as hard as, as clinic Comix and as multifaceted as a system building problem. I th I think that's, that's the key is that there's, there's certain classes of problems that are hard to address in academia because they're system building problems in the sense that maybe you need five or six different.
activities to be happening simultaneously. And if any, one of them. Doesn't follow through completely. you're sort of, you don't have something that's novel and exciting unless you have all the pieces putting, you know, put together. So I don't have something individually. That's that exciting on my own as a paper, Unless you, and also three other people, separately do very expert level, work, which is itself not academically that interesting.
Now having the connectome is academically [00:09:00] interesting to say the least. but yes, not only my incentives. but also everybody else's incentives are to, to maybe spend say 60% of their time doing some academically novel things for their thesis and only spend 40% of their time on, on building the connectome system.
Then it's sort of, the probability of the whole thing fitting together. And then. We see everyone can perceive that. And so, you know, they basically, the incentives don't align well, for, for what you would think of as sort of team science or team engineering or systems engineering. yeah. And so I'm like, I think, I think everybody knows that I'm actually like very much in favor of this thing.
So, I'm going to play devil's advocate to sort of like tease out. what I think are. Important things to think about. so, so one sort of counter argument would be like, well, what about projects? Like cert, right? Like that [00:10:00] is a government yeah. Led, you should, if you do requires a lot of systems engineering, there's probably a lot of work that is not academic interesting.
And yet, it, it, it happens. So like there's clearly like proof of concepts. So like what what's like. W why, why don't we just have more things like, like certain for, the brain. Yeah. And I think this gets very much into why we want to talk about a category of focused research organizations and also a certain scale, which we can get into.
So, so I think certain is actually in many ways, a great example of, of this, obviously this kind of team science and team engineering is incredible. And there are many others, like LIGO or, or CBO observatory or the human genome project. These are great examples. I think the, the problem there is simply that these, these are multibillion dollar initiatives that really take decades of sustained.
government involvement, to make it happen. And so once they get going, and [00:11:00] once that flywheel sort of start spinning, then you have you have it. And so, and so that, that is a nonacademic research project and also the physics and astronomy communities, I think have more of a track record and pipeline overall.
perhaps because it's easier, I think in physical sciences, then in some of these sort of emerging areas of, of, you know, biology or sort of next gen fabrication or other areas where it's, it's, there's less of a, a grounded set of principles. So, so for CERN, everybody in the physics basically can agree.
You need to get to a certain energy scale. Right. And so none of the theoretical physicists who work on higher energy systems are going to be able to really experimentally validate what they're doing without a particle accelerator of a certain level. None of the astronomers are gonna be able to really do deep space astronomy without a space telescope.
and so you can agree, you know, community-wide that, This is something that's worth doing. And I think there's a lot of incredible innovation that happens in those with focus, research organizations. We're thinking about a scale that, [00:12:00] that sort of medium science, as opposed to small science, which is like a, you know, academic or one or a few labs working together, Or big science, which is like the human genome project was $3 billion.
For example, a scope to be about $1 per base pair. I don't know what actually came out, but the human genome has 3 billion basis. So that was a good number. these are supposed to be medium scale. So maybe similar to the size of a DARPA project, which is like maybe between say 25 and. A hundred or $150 million for a project over a finite period of time.
And they're there. The idea is also that they can be catalytic. So there's a goal that you could deliver over a, some time period. It doesn't have to be five years. It could be seven years, but there's some, some definable goal over definable time period, which is then also catalytic. so in some ways it will be more equivalent to.
For the genome project example, what happened after the genome project where, the [00:13:00] cost of genome sequencing through, through new technologies was brought down, basically by a million fold or so is, is, is, how George Church likes to say it, inventing new technologies, bringing them to a level of, of readiness where they can then be, be used catalytically.
whereas CERN, you know, It's just a big experiment that really has to keep going. Right. And it's also sort of a research facility. there's also permanent institutes. I think there's a, is a, is a, certainly a model that can do team science and, and many of the best in the brain mapping space, many of the sort of largest scale.
connectomes in particular have come either from Janelia or from the Allen Institute for brain science, which are both sort of permanent institutes, that are, that are sort of, nonacademic or semi academic. but that's also a different thing in the sense that it's, it takes a lot of activation energy to create an Institute.
And then that becomes now, a permanent career path rather than sort of focusing solely on what's the shortest path to. To some [00:14:00] innovation, the, the, the permanence. So, so the, the flip side of the permanence is that, I guess, how are you going to convince people to do this, this, like this temporary thing, where.
I think, someone asked on Twitter about like, you know, if it's being run by the government, these people are probably going to get, government salaries. So you're, you're getting a government salary, without the like one upside of a government job, which is the security. so like what, what is the incentive for, for people to, to come do this?
Yeah. And I think, I think it depends on whether it's government or philanthropic, philanthropic fro Faros are also definitely. An option and maybe in many ways more flexible, because the, you know, the government sort of has to, has to contract in a certain way and compete out, you know, contracts in a certain way.
They can't just decide, the exact set of people to do something, for example. So, so the government side has. Both a huge [00:15:00] opportunity in the sense that I think this is a very good match for a number of things that the government really would care about. and the government has, has, has the money, and resources to do this, but philanthropic is also one we should consider.
but in any case, there are questions about who and who will do Froy and, and why. and I think the basic answer though, it, it comes down to, it's not a matter of, of cushiness of the career certainty. it's, it's really, these are for problems that are not doable any other way. this is actually in many ways, the definition is that you're only going to do this.
if this is the only way to do it, and if it's incredibly important. So it really is a, it's a medium scale moonshots. you would have to be extremely passionate about it. That being said, there are reasons I think in approximate sense why one might want to do it both in terms of initiating one and in terms of sort of B being part of them. [00:16:00]
so one is simply that you can do science. that is for a fundamental purpose or, or, or, pure, purely driven toward your passion to solve a problem. and yet can have potentially a number of the affordances of, of industry such as, industry competitive salaries, potentially. I think the government, we have to ask about what the government can do, but, but in a certain philanthropic setting, you could do it another aspect that I think a lot of scientists find.
Frustrating in the academic system is precisely that they have to. spend so much work to differentiate themselves and do something that's completely separate from what their friends are doing, in order to pay the bills basically. So, so if, if you don't eventually go and get your own appealing, you know, Tenure track job or, or so on and so forth.
the career paths available in academia are much, much fewer, and often not, not super well compensated. And, and [00:17:00] so there are a number of groups of people that I've seen in sort of, if you want critical mass labs or environments where they're working together, actually, despite perhaps the. Incentive to, to, differentiate where they're working, does a group of three or four together.
and they would like to stay that way, but they can't stay that way forever. And so it's also an opportunity if you, if you have a group of people that wants to solve a problem, to create something a little bit like, like a seal team. so like when, when I was, I'm not very generally militaristic person, but, when I was a kid, I was very obsessed with the Navy seals.
But, but anyway, I think the seal team was sort of very tight knit. kind of a special forces operation that works together on one project is something that a lot of scientists and engineers I think want. and the problem is just that they don't have a structure in which they can do that. Yeah.
So then finally, I think that, although in many cases maybe essentially built into the structure fro is make sense. We can [00:18:00] talk about this as, as nonprofit organizations. these are the kinds of projects where, you would be getting a relatively small team together to basically create a new industry.
and if you're in the right place at the right time, then after an fro is over, you would be in the ideal place to start. The next startup in an area where it previously, it's not been possible to do startups because the horizons for a venture investment would have been too long to make it happen from the beginning.
Well, that's actually a great transition to a place that I'm still not certain about, which is what happens. After it fro, cause you, you said that it, that it's a explicitly temporary organization. And then, how do you make sure that it sort of achieves its goal, right?
Like, because you can see so many of these, these projects that actually sound really great and they like go in and possibly could do good work and then somehow it all just sort of diffuses. [00:19:00] so, so have you thought about how to sort of make sure that that lives on. Well, this is a tricky thing as we've discussed, in a number of settings.
So, in a, like to maybe throw that question back to you after I answer it. Cause I think you have interesting thoughts about that too, but, but in short, it's, it's a tricky thing. So, so the fro. Is entirely legal focused there isn't, there's no expectation that it would continue, by default and simply because it's a great group of people, or because it's been doing interesting work, it's sort of, it is designed to fulfill a certain goal and it should be designed also from the beginning to have a, a plan of the transition.
Like it could be a nonprofit organization where it is explicitly intended that at the end, assuming success, One or more startups could be created. One or more datasets could be released and then a, you know, a much less expensive and intensive, nonprofits, structure could be be there to [00:20:00] host the data and provide it to the world.
it could be something where. the government would be using it as a sort of prototyping phase for something that could then become a larger project or be incorporated into a larger moonshot project. So I think you explicitly want a, a goal of a finite tune to it, and then also a explicit, upfront, deployment or transition plan, being central to it much more so than any publication or anything.
Of course. At the same time. there is the pitfall that when you have a milestone driven or goal focused organization, that the funder would try to micromanage that and say, well, actually, not only do I care about you meeting this goal, but also I really care that by month six, you've actually got exactly this with this instrument and this throughput, and I'm not going to let you buy this other piece of equipment.
Unless, you know, you show me that, you know, [00:21:00] and that's a problem that I think, we sometimes see with, externalized research models, like DARPA ARPA models, that try to. achieve more coordination and, and, and goal driven among otherwise, somewhat uncoordinated entities like contractors and, and universities that, that are working on programs, but then they, they, they, they achieve that coordination by then, managing the process and, with an fro, I think it will be closer to.
You know, if you have a series, a investment in the startup, you know, you are reporting back to your investors and, and they, they, at some level care, you know, about the process and maybe they're on your board. but ultimately the CEO gets to decide, how am I going to spend the money? And it's extremely flexible to get to the goal.
Yeah. Yeah. The, the micromanage, like [00:22:00] figuring out how to avoid, Micromanagement seems like it's going to be really tricky because it's sort of like once you get to that amount of money, I like, have you, have you thought about, like how, like, if you could do some kind of like actually, well, I'll, I'll give her the, the, the, the, the, the thing that the cruxy thing is like this, I think there's a huge amount of trust that needs to happen in it.
And what I'm. like I constantly wonder about is like, is there this like fundamental tension between the fact that, especially with like government money, we really do want it to be transparent and well-spent, but at the same time, in order to sort of do these like knowledge frontier projects, sometimes you need to do things that.
Are a little weird or like seem like a waste of money at the time, if you're not like intimately connected. and so there's, there's this sort of tension [00:23:00] between accountability and, Sort of like doing the things that need to get done. I agree with that and Efros, we're going to navigate that. Yeah.
I agree with that. And I think it relates to a number of themes that you've touched on and that we've discussed with, which has sort of, has to do with the changing overall research landscape of, in what situations can that trust actually occur, you know, in bell labs, I think there was a lot of trust. throughout, throughout that system.
And as you have more externalized research, conflicting incentives and so on it, it's, it's hard. It's hard to obtain that trust. startups of course, can align that financially, to a large degree. I think there are things that we want to avoid. so one of the reasons I think that these need to be scoped as.
Deliverables driven and roadmaps, systematic projects over finite periods of time, is to avoid, individual [00:24:00] personalities, interests, and sort of conflicting politics, ending up. Fragmenting that resource into a million pieces. So, so I think this is a problem that you see a lot with billion dollar scale projects, major international and national initiatives.
Everybody has a different, if you say, I want this to be, to solve neuroscience, you know, and here's $10 billion. Everybody has a different opinion about what solved neuroscience is. And there's also lots of different conflicting personalities and, and leadership there. So I think for an fro, there needs to be an initial phase, where there's a sort of objective process of technology roadmapping.
And people figure people understand and transparently understand what are the competing technologies? What are the approaches? What, what are the risks? And you understand it. and you also closely understand the people involved. but importantly, the people doing that roadmapping and sort of catalyzing the initial formation of that [00:25:00] fro need to have a somewhat objective perspective.
It's not just funding my lab. It's actually, you, you want to have vision, but you, you need to. Subjected to a relatively objective process, which, which is hard because you also don't want it to be a committee driven consensus process. You want it to be active, in, in a, in a systematic, analysis sense, but, but not in a, everyone agrees and likes it, you know, emotionally sense.
and so that, that's a hard thing. but you need to establish it's that trust upfront, with, with the funder, And that's a hard process and it gets a hard process to do as a large government program. I think DARPA does it pretty well with their program managers where a program manager will come in and they will pitch DARPA on the idea of the program.
there'll be a lot of analysis behind it and, but then once, once they're going, that program manager has tremendous discretion, and trust. To how they actually run that [00:26:00] project. And so I think you need something like a program manager driven process to initiate the fro and figure out is there appropriate leadership and goals and our livable as reasonable, Yeah, that seems the way, at least the way that it's presented in the paper, it, it feels a little bit chicken and egg in that.
so with DARPA, DARPA is a sort of permanent organization that brings in program managers. And then those programmers program managers then go, start programs, whereas, The look at fro it seems like there's this chicken and egg between like, you sort of, you need someone spearheading it. It seems like, but then it, you sort of like, it, it seems like it will be very hard to get someone who's qualified to, to spearhead it, to do that before you have funding, but then you need someone spearheading it in order to get that [00:27:00] funding.
yeah. Like, yeah. How, how are you thinking about. Cracking that that's, that's sort of the motivation for me behavior over the next year or two, is that I'm trying to go out and search for them. And, a little bit of it is from my own creativity, but a lot of it is going out and talking to people and try and understand what the best ideas.
Here would be, and who are the networks of, of human beings behind those ideas, and trying to make kind of a prioritized set of borrows. Now, this kind of thing would have to be done again, I think to some degree, if there was a, larger umbrella program that someone else wanted to do, but, I'm both trying to get a set of, of exemplary.
And representative ideas and people together, and try to help those people get funding. You know, I think there can be a stage process. I agree that, in the absence of a funder showing [00:28:00] really strong interest, people committing, to really be involved is difficult, because it is a big change to people's normal.
Progression through life to do something like that. but just like with startups, to the extent that you can identify, someone who's. We spiritually just really wants to do this and we'll kind of do anything to do it, the sort of founder type, and also teams that want to behave like that. that's obviously powerful, and also ideas where there's a kind of inevitability, where based on scientific roadmapping, it, it just has to happen.
There's no way, you know, for neuroscience to progress unless we get better. Connectomics and I think we can go through many other fields where, because of. The structures we've had available and just the difficulty of problems now, where arguably Faros are needed in order to make progress in fields that people really care about.
So, so I think you can get engagement at the level of, of discussion, and, and, and starting to nucleate [00:29:00] people. But, but there is a bit of a chicken and egg problem. In the sense that it's, it's not so much as here's an fro, would you please fund to me it's we need to go and figure out where there might be Faros to be had, and then who is interested in those problems as well to, to fund and support those things.
So, yeah. So I guess to recap what I see your process that is, is that you're going out, you're sort of really trying to. Identify possible people possible ideas, then go to funders and say, here, like sort of get some, some tentative interest of like, okay, what, which of these things might you be interested in if I could get it to go further and then you'll circle back to.
the, the people who might be interested in sort of say like, okay, I have someone, a funder who's potentially interested. Can we [00:30:00] sort of like refine the idea? and then sort of like, like you will drive that loop hopefully to, Getting a, an fro funded that's right. And there's, there's further chicken and egg to it.
that has to be solved in the sense that, when you go to funders and you say, why, you know, I have an idea for an fro. We also need to explain what an fro is, right? in a way that both, engages people in creating these futuristic models, which many people want to do, While also having some specificity of, of what we're looking for and what, what, what we think is as possible.
So, and then the same on, on the, on the side of, of scientists and engineers and entrepreneurs all over the world, who, you know, have the ideas certainly, but most of those ideas have been optimized to hit, the needs of existing structures. So, so we are, we are trying to, I think, broker between those, And [00:31:00] then start prototyping a few.
but the, you know, the immediate thing I think is to make, w Tom Coolio has referred to a catalog, a Sears catalog of moonshots. and so we're trying to make a catalog of, of moonshots that fit the fro category. but that sounds like the perfect name for this podcast, by the way. the cataloging mood child, like, you're kind of kind of cataloging moonshots and ways to get moonshots and yeah, absolutely.
Yeah. and so I guess another sort of, thing that I've seen, and I'm not sure, it's almost like for people like a lot of people who like really want. Who like sees something as inevitable and they really want to get it done. In sort of like the current environment we're recording in October, 2020.
there's. There's sort of this perception that capital is really cheap. [00:32:00] you know, there's a lot of venture capitalists there. They're pretty aggressive about funding and one could make an argument that, if it's, if it, it really is going to be inevitable and it really is going to start a new industry.
Then that is exactly where venture capital funding should come in. And I do see this a lot where people, you know, it's like they have this thing that they really want to see exist and they, you know, come out of the lab and it started a company that's sort of extremely common. so. I guess, like, what almost would you say to someone who you see doing this that you think maybe should do an fro instead?
Yeah, that's a great question. I mean, I think it's a complicated question and obviously, you know, we got to see VC also, you know, obviously VC backed, you know, innovation is, is, is one of, if not sort of the key, [00:33:00] Things that is driving technology right now. So, so I'm in no way saying that fro is, are somehow superior to two startups, in any generalized way.
So I think that things that can be startups and are good as startups should be startups and people, if you have an idea that could be good for a startup, I think you should go do it. Generally speaking. But, there, there are a few considerations, so yeah. So I think you can divide it into categories where VCs, no, it's not a good idea for startups.
And therefore won't talk to you, in cases where VCs don't always know whether or not it's good for a startup or whether there's a way that you could do it as a startup, but it would involve some compromise that is actually better not to make, even potentially for the longterm. economic prospects of, of an area.
So things that can happen, would be, if you have something that's basically meant to be a kind of platform technology or which you [00:34:00] need to develop a tool or a platform in order to explore a whole very wide space of potential applications. maybe you have something like a new method of microscopy or something, or a new way to measure proteins in the cell or things like that, that, you know, you could target it to a very particular, if you want product market fit application, where you would be able to make the most money on that and get the most traction, the soonest.
Yeah. Sometimes people call this, you know, the, the, the, the sort of Tesla Roadster, equivalent. You want to guys as quickly as you can to the Tesla Roadster. And I think generally, what people are doing with, with that kind of model, where you take people that have science, to offer, and you say what's the closest fastest you can get, to a Tesla Roadster that lets you it lets you build, get, get revenue and start, start being financially sustainable and start building a team, to go further.
generally that's really good. and generally we need more scientists to learn how to do that. it'd be supported to do that, but, [00:35:00] sometimes you have things that really are meant to be. either generalized platforms or public goods, public data, or knowledge to underlie an entire field. And if you work to try to take the path, the shortest path to the Roadster, you would end up not producing that platform.
You would end up, producing something that is specialized to compete in that lowest hanging fruit regime, but then in the, in, in doing so you would forego the more general larger. Larger thing. And, you know, Alan Kay has, has the set of quotes, that Brett Victor took is linked on his website. and I think Alan K meant something very different actually, when he said this, but he's, he refers to the dynamics of the trillions rather than the billions.
Right. and this is something where in, and we can talk about this more. I'd be curious about your thoughts on that, but something like the transistor. You know, you, you could try to do the transistor as a startup. and maybe at the time, you know, the best application for transistors would have been [00:36:00] radios.
I don't think like that. I think it was, it was guiding a rockets. Yeah. So you could have, you could have sort of had had a transistors for rockets company and then tried to branch out into, becoming Intel. You know, but really, given the structures we had, then the transistor was allowed to be more of a, a broadly, broadly explored platform.
yeah, that, that progressed in a way where we got the trillions version. And I worry sometimes that even some startups that have been funded at least for a seed round kind of stage, and that are claiming that they want to develop a general platform are going to actually struggle a little bit later.
when investors, you know, see that, see that they would need to spend way more money to build that thing. then the natural shortest path to a Roadster, or another words the Roadster is, is, somehow illusory. yeah. Yeah, this [00:37:00] is, this is a. Sort of like a regime that I'm really interested in and a, just on the transistor example, I've, I've looked at it.
So just the, the history is that it was developed at bell labs, in order to prevent a T and T from being broken up, bell labs had to, under strictly licensed a bunch of their innovations, including the transistor William Shockley went off and, Started, chocolate semiconductor, the traders eight then left and started, Fairchild and then Intel.
And, believe that that's roughly the right history. but the, the really interesting thing about that is to ask the question of like, one, what would have happened if, bell labs had exclusive license to the transistor and then to what would have happened if they had like exclusively licensed it to, Shockley semiconductor.
And I think I would argue in both of those situations, you don't [00:38:00] end up. Having the world we have today because I fell labs. It probably goes down this path where it's not part of the core product. and so they just sort of like do some vaguely interesting things with it, but are never incentivized to like, you know, invent, like the, the planner processing method or anything.
Interesting. yeah. Yeah. And so I guess where I'm. Go. And then like at the same time, the interesting thing is like, so Shockley is more, akin to like doing a startup. Right. And so it's like, what if they had exclusive license to it? And the, what I would argue is actually like that also would've killed it because, you have like, they had notoriously bad management.
And so if you have this, this company with. And like the only reason that, the trader could go and start a Fairchild was because they, that was, that was [00:39:00] an open license. So this is actually a very long way of asking the question of, if F borrows are going to have a huge impact, it seems like they should default to.
Really being open about what they create from like IP to data. but at the same time, that sort of raises this incentive problem where, people who think that they are working on something incredibly valuable, should want to do a startup. And then. And so there, and then similarly, even if they'd be like that sort of couldn't be a thing, they would want to privatize as much of the output of an fro.
and so which. Maybe necessary in order to, to get the funding to make it happen. So I guess like, how are you thinking about that tension? That was a very long winded. Yeah. [00:40:00] Yeah. Well, there's, there's a lot, a lot there, I think, to loop back to you. So, so I think, right, so, so this idea that we've talked a bit about as sort of default openness, so, so things that can be open for maximum impact should be open.
there are some exceptions to that. So, so if, And it's also has to do partly with how you're scoping the problem. Right? So, so rather than having an SRO that develops drugs, let's say, because drugs really need to be patentable, right. In order to get through clinical trials, we're talking about much more money than the fro funding, you know, to do the initial discovery of a target or something.
Right. So to actually bring that to humans, you know, you need to have the ability to get exclusive IP. for downstream investors and pharma companies that that would get involved in that. so there are some things that need to be patented in order to have to have their impact. but in general, you, you want, I think fro problems to steer themselves to things where indeed.
it can be maximally open and maybe, maybe you, you provide [00:41:00] a system that can be used to, to, or underlie the discovery of a whole new sets of classes of drugs and so on. But you're not so much focused on the drugs themselves. Now, that being said, right. if I invest in an SRO, and I've enabled this thing, right.
It kind of would make sense for the effort, you know, maybe three of the people of, of, of, of 15 in the fro will then go and start a company afterwards that then capitalizes on this and actually develops those drugs or what have you, or it takes it to the next stage. And gosh, it would really make sense if I had funded in fro.
that's, those people would like to take me as a sort of first, first, first refusal to get a good deal on, on investing in this startup, for example. Right. so I think there are indirect network-based, or potentially even legal based, structure, structure based ways to both incentivize the investors and, But it's, it's a weaker, admittedly weaker, incentive financially than, [00:42:00] than, than the full capture of, of, of something.
But then, but then there's, I think this gets back to the previous discussion. So which is sort of the trillions rather than the billions. So if you have something where maybe there are 10 different applications of it, Right in 10 different fields. you know, maybe, maybe we have a better way to measure proteins and based on this better way to measure proteins, we can do things in oncology and we can do things in Alzheimer's and we can do things in a bunch of different directions.
We can do things in diagnostics and pandemic surveillance, and so many fields that one startup, It would be hard even to design, to start, if that could capture all of that value just as it would have been hard to design sort of transistor incorporated. Right. Right. given that, I think there's, there's a lot of reason to.
To do an fro and then explore the space of applications. Use it as a means to explore a full space in which you'll then get [00:43:00] 10 startups. so if I'm the investor, I might like to be involved in all 10 of the new industry, right. And the way to do that would be to create a platform with which I can explore, but then I have a longer time horizon.
Cause I have to first build the thing. Then I have to explore the application space and only then. do I get to invest in a specific verticals, right? Yeah. I think the, the two sort of tricky questions that I, I wonder about what that is one. So you mentioned like, Oh, there's 15 people in an fro, three of them go off to start a startup.
What about those other 12 people? Like, I, I assume that they might be a little bit frustrated if, if that happens, Yeah, because like, like they, they did, they did help generate that value in it. It sort of gets into two questions of like capturing, like sort of kicking back, value generated by research in general, but like, yeah, it could, it could, it could be all 15 people, you know, we saw something [00:44:00] similar with open AI, you know, in a way, for example, converting, you know, into, into a, for profit or at least a big arm of it being, being the for-profit, and keeping all the people.
Right. So you, you, you, you can imagine, just blanket converting. but yeah, I think, I think it's sort of, In the nature of it, that these are supposed to be things that open up such wide spaces that there's, there's sort of enough for everyone, but no, no, no one person necessarily one startup would completely capture.
And I think that's true for clinic Comix too, for example. Right. So if you had really high throughput clinical, connectomics just, just to keep going on this example, that's a great example of perfect. It's a good thing as a good example. It's not. Depending on the details, whether this is exactly the first fro or not.
I think it's totally, totally other issue, but, but. Connectomics there's potentially applications for AI and you know, how, how the neurocircuits work, and sort of fundamental, funding. Mental is a brain architecture and intelligence. although there's a bunch of ranges of the sort of uncertainty of exactly what that's going to be.
So it's hard to sort of [00:45:00] know it until you see the data. There's also potentially applications for something like drug screening, where you could put a bunch of different, Kind of some CRISPR molecules or drug perturbations on, on a, on a brain and then look at what each one does to their, the synopsis or, and look at that in a, in a brain region specific way and sort of have ultra high, but connect to them based drug screening.
Neither of those are things you can start a start up until you have connected. Right. working. but so anyway, so maybe three people would start an AI company and maybe those would be the very risk tolerant ones. and then three would start at, you know, a crisper drug company and, and, and, three would just do, do fundamental neuroscience with it and, take those capabilities and, and, and go, go back into the university system or so on and yeah.
And start using that. Yeah. And the, the sort of the other related to. like creating value with it. there's, there's a little like uncut discomfort that like even I have [00:46:00] with, say like philanthropic or government funding, then going to fund a thing that proceeds to make a couple of people very wealthy.
Which like, and like, there's very much arguments on both sides, right. Where it's like, it'll generate a lot of good for the world. and, and all and, and such. so, so like, I guess what would you say? I guess like, as a, as a, like, if I were a very wealthy philanthropist and I'm like, do it, like, you know, it's like, I'm just giving away money so that these people can.
Yeah, the company is a complicated thing. Right? How much, how many further rich people, you know, did the Rockefeller foundation, you know, investing in the basics of molecular biology or things like that ended up generating? I mean, I think that, I think you, I think in some way the government does want to end up is they want the widely distributed benefit.
And I think everything that should be an SRO should have widely distributed benefits. It shouldn't just [00:47:00] be a kind of, A startup that just, just enhances one, one person. It should be something that really contributes very broadly to economic growth and understanding of the universe and all that. But it's almost inevitable.
I think that, if you create a new industry, you're gonna, you know, you're gonna, you're gonna feel it going to be some more written about rich successful people in that industry. And they're probably going to be some of the people that were involved. Early and thinking about it for the longest and waiting for the right time to really enter it.
And so, yeah, that's a really good point. I guess the, then the question would be like, how do you know, like, like what are, what are sort of a, the sniff test you use to think about whether something would have broadly distributed benefits? That's a great question. Cause it's like connect to them. It seems like fairly clear cut or, or generating sort of like a massive data set that you then open up. Feels very [00:48:00] clear. Cut. it's. We we've talked before about that, like fro is, could like scale up a process or build a proof of concept of, of a technology.
and it, it seems like that it's less clear cut how you can be sure that those are going to, like if they succeed. Yeah. I mean, there are a few different frames on it, but I mean, I think one is, FRS could develop technologies that allow you to really reduce the cost of having some. Downstream set of capabilities.
so, you know, if, just to give you an example, right? If, if we had, much lower costs, gene therapies available, right? So, so sometimes when drug prices are high, you know, this is basically it's recouping these very large R and D costs and then there's competition and, and, and profit and everything involved.
you know, there was the marching squarely situation and, you know, there's a bunch of, sort of. What was that? there was, remember the details, but there [00:49:00] was some instance within which, a financially controlling entity to sort of arbitrarily bumped drug prices way high, right. A particular drug. and then w was, you know, was regarded as an evil person then, and maybe that's right.
but anyway, there are some places I think, within the biomedical system where you can genuinely reduce costs for everyone. Right. and it's not simply that I, you know, I make this drug and I captured a bunch of value on this drug, but you know, it's really, it should be available to everyone and I'm just copying there.
There's genuine possibility to reduce costs. So if I could reduce the cost of, of the actual manufacturing of. The viruses that you use for gene therapy, that's a, that's a process innovation. that would be, you could order as a magnitude drop the cost of gene therapy. If you could figure out what's going on, in the aging process and what are the real levers on a single, you know, biological interventions that would prevent multiple age related diseases that [00:50:00] would massively drop the cost.
Right? So those, those are things where, Maybe even in some ways it would be threatening, to some of, some of the pharma companies, you know, that, that work on specific age related diseases, right? Because you're going to have something that, that replaced, but this is, this is what, you know, things that are broad productivity improvements.
And I think economists and people very broadly agree that, that the science and technology innovations, For the most part. although sometimes they can be used to in a way that sort of, only benefits, a very small number of people that generally speaking there's a lot you can do, with technology that will be extremely broadly shared in terms of benefit, right?
Yeah. Yeah. I mean, I, I do actually, like I agree with that. I'm, I'm just, I'm trying to represent as much skepticism as, as possible. Definitely. I know you agree with that.
And actually, another thing that I have no idea about which I'm really interested in is as you're going and sort of creating this, [00:51:00] this moonshot catalog. how do you tell the difference between people who have these really big ideas who are like hardcore legit? but like maybe a little bit crazy.
And then people who are just crackpots. Yeah, well, I don't claim to be able to do it in every field. and, and I think there's a reason why I've, I'm not trying to do a quantum gravity, fro you know, both, both, because I don't know that that's, you know, I think that's maybe better matched for just individual.
Totally. Open-ended Sunday, you know, fun, brilliant people for 30 or 40 year long period to just do whatever they want. Right. Yeah. For quantum gravity, rather than directed, you know, research, but, But also because there's a class of problem that I think requires a sort of Einsteinian type breakthrough in fr fro is, are not, not perfect for that in terms of finding people.
I mean, I, I find that, there's a lot of pent up need for, this is that's my preliminary feeling. and you can see there's a [00:52:00] question of prioritizing, which are the most important, but there's a huge number of. Process innovations or system building innovations that are needed across many, many fields.
And you don't need to necessarily have things that even sound that crazy. There are some that just kind of just make sense, you know, are, are very simple. You know, we here, here in our lab, we have this measurement technology, but we, you know, we can only have the throughput of one cell, you know, every, every few weeks.
And if we could build the system, we could get a throughput of, you know, A hundred thousand cells, you know, every month or something. Right. there are some, there's some sort of ones that are pretty obvious, or where there's an obvious inefficiency. In kind of, how things are structured. Like every, every company and lab that's that's modeling fusion reactors, and then also within the fusion reactor, each individual component of it, like the neutrons in the wall versus the Plaza and the core, those are basically modeled with different.
Codes many of which are many [00:53:00] decades old. So there's sort of an obvious opportunity to sort of make like a CAD software for fusion, for example, you know, that the, the, it doesn't, it's not actually crazy. It's actually just really basic stuff. In some cases, I think they're ones where we'll need more roadmapping and more bringing people together to really workshop the idea, to really have people that are more expert than me say, critique each other and see what's.
Really going on in the fields. and I also rely on a lot of outside experts. if I have someone comes with an idea, you know, for, for energy, you know, and I'm talking to people that are like former RPE program managers or things like that, that, that know more of the questions. so I think we can, we can, we can do a certain amount of, of due diligence on ideas and.
and then there are some that are, that are really far out. you know, we both have an interest in atomically precise manufacturing, and that that's when, where we don't know the path I think, forward. and so that's maybe a pre fro that's something where you [00:54:00] need a roadmapping approach, but it's maybe not quite ready to, to just immediately do an fro.
Yeah, no, that's, you sort of hit on a really interesting point, which is that. when we think of moonshots, it's generally like this big, exciting thing, but perhaps some of the most valuable is will actually sound incredibly boring, but the things that they'll unlock will be. Extremely exciting. yeah, I think that's true.
And, and you have to distinguish there's there's boring. Right? So, so I think there's, there's some decoupling of exactly how much innovation is required and exactly how important something is. And also just how much brute force is required. So I think in general, our system might under weight, the importance of brute force.
And somewhat overweight the importance of sort of creative, individual breakthrough thinking. at the same time, there are problems where I think we are bottlenecked by thinking I'm like really how to do something, not just to [00:55:00] connect them of brain, but how do you actually do activity map of entire brain?
You actually need to get a bunch of physicists together and stuff to really figure out what's, you know, there's a level of thinking that is not very non-obvious similarly for like truly next gen fabrication. You really, really, really need to do the technology roadmapping approach. And that's a little different than the fro.
And in some cases there may be a, as we discussed, I think in the past, there was sort of a, a continuum potentially between DARPA type programs or programs that would start within the existing systems and try to catalyze the emergence of ideas and discoveries. And then fro is, which are a bit, a bit more cut and dry.
And in some cases, even you could think of it as boring. but just very important. how do we prevent Faros from becoming a political football? because you see this all the time where, you know, a Senator will say, well, like I'll sponsor this bill, as long as we mandate that.
50% of the work has to happen in my particular state or [00:56:00] district. and, and I imagine that that would be counterproductive towards the goals of . so do you, do you have any sense of like how to, how to get around that
probably much easier in philanthropic setting than governments? Although I think I'm overall, I'm, I'm sort of optimistic that, if. If the goals are made very clear, the goal is disruptive, you know, multiplicative improvements in scientific fields. that's the primary goal. It needs to be managed well.
so it's not either about the individual peoples, if you want academic politics and also that it doesn't, doesn't become about sort of, you know, districts, congressional districts, or all sorts of other things. I think there's a certain amount of complexity, but the other, the other thing is. I think there's really amazing things to be done in all sorts of places and by all sorts of people that are not necessarily identified as, as the biggest egos or the largest cities also, although certainly there are hubs that [00:57:00] matter.
yeah. Cool. I think so. I think those are all like the actual questions I have. Is there anything you want to talk about that we have not touched on? Yeah, that's a good question. I mean, how does this fit into two things that you're thinking about, in terms of your overall analysis of the research system, then, do you think this, what is this leave unsolved as well?
if, even if we can get some big philanthropic and government, donors. Yeah. So, so there are sort of two things that I. see it not covering. And so the, the first that you you've sort of touched on is that there are, some problems that still like don't fit into academia, but are not quite at the point where they're ready to be at fro.
And so, they need, the, the like mindset of the fro without. Having this sort of, cut and dryness [00:58:00] that you need to sort of plunk down, like have the confidence to plunk down $50 million. so, so we need sort of a, a, what I would see as a sustainable, way of. Getting to the point of fro type projects.
And as you know, I'm spending a lot of time with that. and then sort of a, the other thing that I've realized is that when, when people, we sort of have these discussions that are like research is broken, I think what we're actually talking about is, is sort of two really separate phenomenon. So, what we've been talking about, like Efros, Are really sort of sitting in like the Valley of death where it's like helping bridge that.
but I think that at the same time, there there's like what I would call like the, the Einstein wouldn't get in any funding problem, which is, as you alluded to there, there are some of these things, like some of the [00:59:00] problems with research that we talk about are just about, The sort of conformity and specialization of really idea based exploratory, like completely uncertain research.
And that's also really important, but I, I think it's what we don't do is, is, is sort of like separate those two things out and say like, these are both fall under the category of research, but are in fact. Extremely different processes. They require very different solutions. Yeah. Actually let me, let me, since you mentioned that, and since we are here together on the podcast, I agree with that and I, I have some things to say about that as well.
So, so I think that the fro is indeed only address, or are designed to address this issue of sort of system building. problems that have a sort of catalytic nature and are a particular kind of pre-commercial stage. Right? So in some ways, [01:00:00] even though I'm so excited about borrows and how much they can unlock, because I think that this is one of two or three categories that has been, you know, under emphasized by current systems or has systems currently have struggled with it.
there are these others. So, so I think that. The, the supporting the next Einstein and people that may have also have just be cognitively socially in any other number of ways, just different and weird and not good at writing grants. You know, not good at competing. Maybe not even good at graduating undergrad.
Yeah. You know, I'm running a lab who are, are brilliant and because the system now. Has proliferate in terms of the number of scientists. it's very competitive and, and there is a, there's a lot of need to sort of filter people based on credentials. So there's this sort of credential there's people that don't fit with perfectly with credentials or with a sort of monoculture of who is able to get NSF grants and go through the university system and [01:01:00] get the PhD and all those different Alexey goosey has this nice blog post is oriented toward biomedical, but saying basically that in order to get through the system, you need to do 10 or 15 things simultaneously.
Well, and also be lucky. And maybe we want to be looking for some people that are only able to do three of those things about, but are orders of magnitude better than others, then there's people even who have done well with those things, but still don't have the funding or sort of sustained ability, to, to pursue their own individual ideas over decades.
even if they do get tenure or something, because the grant system is based on peer review and is, is sort of filtering out really new ideas, for whatever reason, There's kind of the broader issue that Michael Nielsen has talked about, which is sort of the idea that too much funding is centralized in a single organizational model.
So particularly the NIH, the NIH grant is kind of hegemonic as, as, as a structure and as a peer review mechanism. then I think we need more [01:02:00] DARPA stuff. We probably need more darker agencies for other problems. Even though I've, I've sort of said that I think Rose can solve some problems that DARPA DARPA will struggle with.
Likewise, DARPA walls solve problems that fro may struggle with. particularly if there's a very widely distributed expertise across the world that you need to bring together in a, some transient, interesting way, for a little bit more discovery oriented, perhaps in Faros and less deliverable oriented or team oriented.
And then there's even bigger things we need, you know, like we need to be able to create, you know, a bell labs for energy, you know, or sort of something even bigger than fro. so yeah, I think the thing that you're, you're getting at that I is, is sort of simple, but under done is actually analyzing like what the activity is and what.
How to best support it. Yep. Which is instead of just saying [01:03:00] like, ah, there's some research let's give some money to the research and then magical things will happen actually saying like, okay, like, like how does this work? Like what, and then what can we do for these, these specific situation? Yes. I think as you've identified.
Like there's both on the one hand, there's the tendency to micromanage research and say, research has to do this, this with this equipment and this timescale it's entirely, this is sort of subject to milestone. And on the other hand is research is this magical thing. We have no idea. but just. Let other scientists, peer review each other, and just sort of give as much money to it as we can.
and then we see what happens. Right. And I think neither of those, is a, is a good design philosophy, right? Yeah. Yeah. And I think it involves people like thinking it's it's uncomfortable, but like, like thinking and learning about. How, how did you think then understanding how it could, how it could be different?
[01:04:00] How it's not a it's it's a system. Kevin has felt set, said it said it well. And so in some ways it's been designed, but really our scientific systems are something that has evolved into large degree. No one has designed it. It's not. Something that's designed to be optimal is it's a, it's a emergent property of many different people's incentives.
And, if we actually try to apply more design thinking, I think, I think that can be good as long as we're not over overconfident in saying that there's one model for everyone. Yeah. I think that the trick to, sort of fixing. Emergent systems is to like, basically like do little experiments, poking at them.
And that's, that's very much what I see getting fro is going okay. It's like, you're not saying, Oh, we should like dismantle the NSF and have it all be . Okay. Let's do a couple of these. See what happens. That's right. It's I think it's inherently a small perturbation and it it's. And I [01:05:00] think DARPA, by the way is a similar thing.
It's sort of dark. You wouldn't need DARPA. If everything else was already sort of efficient, right. Given that things are not perfectly efficient, Darko has all these, all these sort of this niche that it fills. I think similarly Faros, they can only exist. if you also have a huge university system and you also have companies that that doesn't make sense, otherwise it's, it's a perturbation, but as we, I think it's a perturbation in which you unlock a pretty big pressure stream sort of behind it when you open it up.
So. Excellent. Well, I think that's, that's actually a great place to close. I guess the last question would be, Like, if people are interested in, in Faros, especially like funding or running one, what is the best way for them to reach you? Well, they can, they can talk to me or they can talk to you.
my email has, is prominently listed on my website. Twitter is great. and that, yeah, I really interested in, people that have a kind of specificity [01:06:00] of, of, of what they want of, you know, here here's, here's what I would do, very specifically, but I'm also interested in talking to people that, See problems with the current systems and want to do something and want to learn about, other highly specific fro ideas that others might have, and how to enable those.
50 episodes
Alla avsnitt
×Welcome to Player FM!
Player FM is scanning the web for high-quality podcasts for you to enjoy right now. It's the best podcast app and works on Android, iPhone, and the web. Signup to sync subscriptions across devices.