“Let’s go do that”

A conversation with Astera resident Erika Alden DeBenedictis

What separates a collection of atoms from something alive? That’s the question that drives Erika Alden DeBenedictis, from her undergrad years studying operating systems that transform hunks of metal into machines that think, to her current work as an Astera resident working to terraform Mars.

Astera’s residency program bets big on people and projects that don’t fit within existing institutions. As a resident, Erika is also bringing to life new kinds of organizations to drive change in science. During her residency, she founded Pioneer Labs, a nonprofit with a ~~moonshot~~ marsshot mission to transform microbes to be able to live on Mars.

Christina Agapakis sat down with Erika to discuss what drives her work, the role that agency, emotion, and collaboration play in science, and new models for innovation in bioengineering. The following is an edited transcript of our conversation.

Christina: Why do you want a garden on Mars? What drives your work?

Erika: I do science because it’s a way to learn about myself and the universe. I’ve been drawn to biology specifically because I am fascinated by the junction between something being a bunch of atoms and something being alive. Where is that first moment when something starts being alive? This is a question that to me still holds some real mystery.

I started off in astronomy and similar kinds of “origins” questions, at a much grander scale—how was the universe formed? How was our solar system formed? How did Earth get here? Where did the elements that caused this to be a friendly place for life to evolve come from?

So what could be more delightful than the challenge of figuring out how to create life that could thrive in a place that is famously hostile to life. What is the difference between a planet that’s dead and a planet with the same atoms and a different orientation that is alive?

What do you mean when you say this is about wanting to be right and learning about yourself? What does it mean to learn about yourself in the context of these big questions about the origin of life?

In answering the question, “Can I change the world?” you learn about yourself (your own abilities and limits) and the universe (how it works and what is possible to change).

When I started grad school, I had just finished undergrad as a computer scientist and I had loved operating systems. My favorite part was the bootloader—the computer program that turns a hunk of metal into a live thing that you can interact with and program. And I was fascinated by how that was a human created system that could go from dead to something responsive to you. And I wanted to see if I could do the same thing in biology.

What’s the operating system that boots life up? Can I play with it? So I worked in origin of life research and on genetic code expansion, because the genetic code is the operating system for life. Today, every living thing on Earth uses a genetic code based on three base codons—why three bases? It doesn’t have to be three bases. It could be two bases. It could be four bases.

What I mean when I say I do science as a way to learn about myself is: what as a human is my ability to manipulate living systems? What is my ability to learn about the world? We’re surrounded by life that uses three base codons, how does it work? How flexible is it?

So I went off and I tried to make a four base codon bacteria. What I learned was that it’s very hard and also that it’s not impossible. You can, in individual cases, trick living systems into using four base codons as the exception. It’s possible to edit the operating system of life. In principle you could do it, but it would require a magnitude of engineering effort that’s probably not justified by its application.

That was an example of pushing the limits—as humans, what can we do? Do we actually have the ability to change the world around us? Our ability to change and enact, to have agency in the world is directly related to how well we understand it.

In answering the question, “Can I change the world?” you learn about yourself (your own abilities and limits) and the universe (how it works and what is possible to change).

How do you stay focused on that ambition and possibility of what science can do? As grad students we’re much more likely to complain about cloning than think about the remarkable power we have and the fact that we’ve been handed the tools to bend the fabric of life to our will.

Telling someone else’s joke is always terrible, but there’s that comedian who was making fun of people complaining when the wifi on the plane is bad. “You are participating in a miracle of engineering, you are flying in the air! And it even has wifi!”

What are you complaining about, right? But I think that’s grad school in a nutshell. You’re stuck in a little tin can and you have to clone stuff. It sucks.

I think deep down, probably many scientists are driven by emotion, but it’s not done to talk about it. It’s very Spock—you’re supposed to be analytical, you’re not supposed to talk about the emotional experience of doing science and what fundamentally motivates people.

I wrote a blog post a little while ago of my list of reasons why I love science, and specifically the names for the emotions I feel that delight me as a scientist. Feelings like “finding a diamond in the rough” or “I know a secret.”

The big one for me is the feeling agency—that’s the feeling of “holy shit! if I can get a virus to depend on a tRNA that decodes a non-canonical codon and an orthogonal ribosome, then who am I and what else can I do?” I think it’s a very hard level to get to in life science because just doing life science is so hard. But when you do get to that level of agency where you can do any experiment you want it’s such a thrill because only you can just ask and answer questions and no one’s stopping you except your own ability to clone things.

Some people might argue that Green Mars is not really a practical application of bioengineering, beyond that transition between a dead planet and a live one, what are other reasons for terraforming Mars?

I like Green Mars as a thesis because it is first of all a really well defined problem statement. It is a verifiable vision of what could be accomplished in the long term. When I look at the sky with my naked eye, Mars is red. I want it to be green. Changing the color of that dot is a thing everyone could understand.

In terms of practicality, there’s a lot more work to be done, but it’s possible that greening Mars is eminently doable and shockingly fast and cheap. Recent work has lowered the possible cost a lot. When we last did some estimates, we were getting numbers that place greening Mars in the department of $100B’s and ~30-50 years. It’s possible that Mars could be green in my lifetime for roughly the cost of the US’s interstate highway system. Of course, the biggest problem with these estimates right now is that not enough people are thinking about it. We need more research, more thorough cost estimates, and more discussion about whether or not we want to actually do it if it’s possible.

Meanwhile, researching the possibility of a green Mars involves an infinite number of steps all of which are in the right direction. How do we make human presence net-positive for the surrounding environment, rather than net-negative? Space missions are today constrained by our lack of good ISRU technologies. That’s space science jargon. In situ resource utilization, ISRU. Or in biotech speak, “upcycling low-cost feedstocks”. We just need better, more reliable technology for converting raw materials and waste streams into useful stuff. And that’s a common need for Earth and space alike.

Biology also needs to figure out how to be more robust, less fussy and finicky. Something you would actually trust to be a mission critical technology in a space mission. If I can offer a critique of biotech today it’s that we’re not good at that. We’ve sort of managed to get there with therapeutics manufacturing. That’s the area of life science that has achieved that level of certainty. And the rest of the field really hasn’t even tried. To get toward green Mars, we fix that. That’s a cultural issue that corresponds to an infinite number of actual technical things that need to happen that are all in the right direction and are all worth doing.

Synthetic biology as a discipline really focused on that idea of building an engineering culture and transforming life science into a true engineering discipline. Why do you think biotech isn’t good at that yet?

Biologists are working with systems that are genuinely very hard. Your bacteria probably grow a bit differently on Monday versus Tuesday. And as a biologist you’re like, oh, well, that’s normal.

Now, from a physics perspective, something caused that, but there are so many things that go into altering how living systems work we have largely given up on ever fully tracking it down. As a result, life science just evolved as a field that can’t hope to have things always be fully quantitative and reproducible, because we just live in a universe that’s full of so many unknowns. It’s not practical.

At least it wasn’t practical. More recently we are actually starting to get to predictive models of living systems, things like Alphafold2. We have better automation now so we could actually run the exact same experiment twice. I think there’s lots of reasons to believe that we could change the culture.

It’s also why it’s so hard to automate experiments and why there’s so few things where we have enough data for predictive models beyond protein structure, right?

Yes, with proteins you’re just a tiny step up that ladder from atoms up to that point where the things start being alive. We have to keep climbing that ladder.

So far we chewed away at it just a little bit. It used to be that protein structure was this crazy alive thing. We couldn’t predict it. Now we’ve gone up that ladder, just one more rung, which is so exciting.

Then we just push the limit further. Now the thing that’s complicated is proteins inside cells rather than just by themselves in a tube.

How do you go up that ladder? What do the next rungs look like?

This is what Align works on, which was the other non profit I started. I think the next biological phenomena we will be able to predict with AI is proteins in the context of a living cell.

Alphafold and protein structure prediction is basically a chemistry problem. It’s the chemistry of biomolecules, so there’s a little bit of biology flavor. What is truly and undeniably biology is when you look at those proteins in the context of a cell and how they function together. All of the complexity is there. Predicting the properties of proteins in context, like how well does it express? How functional is it? Those are the next rungs of the ladder. To get there we need more big datasets in life science. Alphafold2 was trained using the Protein Data Bank (PDB), which cost $10B and took 50 years to establish. We need the next PDB, and we need to do it faster and cheaper than last time.

One level up from that is the whole cell model. Can you predict where a microbe will and won’t grow? That’s something we’re working on at Pioneer, because we think about what it would take for cells to grow in the context of Mars. What nutrients are present on Mars and how do we make a microbe that grows there? We’re generating data for that to build better models and build better predictive models to help us do that engineering.

You’ve now started two non-profit organizations to work on these kinds of problems. Is there an analogy here, another moment where there’s a boundary between something dead and something alive? When you go from a collection of people and ideas and resources into real coordinated action and results?

We have to first recognize that, fundamentally, science is incredibly difficult and if we want to solve the hardest problems that we possibly can, those problems are what need to be prioritized. Then we need to design the institutional structure, the incentive structure for all the researchers and the funding source to fit the technical problem that’s being addressed, not the other way around.

Today, most people in biotech think in units of for-profit startup or academic lab, those are basically the only options on the table. But if we look to other industries, there’s plenty of other options, including ones that have maybe never been used in science before but totally could.

One of my favorite examples of this working for biotech is Addgene. Addgene is a nonprofit that managed to fix an incredibly stupid problem: scientists really wanted to share all their custom assembled pieces of DNA that they were making, but they couldn’t do it because it’s a pain in the ass to prep the DNA and FedEx it to somebody. And even worse, you have to interface with your university’s legal office for tech transfer documents, which could take months. No one wants to do that.

Addgene figured out that they could streamline sharing on both ends. You send Addgene your materials and then they’ll do a bunch of QC on it for free (that you’d never do by the way). They store it, prep it, ship it, and handle the legal documentation. And then other people can just go to Addgene and order the samples they need. Today, Addfgene does around $10 million of revenue every year. They identified a problem in the ecosystem, they solved it, and they invented a way to finance it sustainably that just works for everybody.

I think there’s an infinite number of those opportunities, but no one thinks of starting “an Addgene”, even though you could. There’s a lot of solvable problems that prevent us from doing good science. We should find a way to fix it, and then do it again for the next problem.

*The Pioneer Labs team at the Falcon 9 launch*

That seems like another cultural issue? Until pretty recently there was a very pervasive cultural belief in science that if you don’t become a professor you are a failure. It seems like there’s a bit of a flourishing now of different models of what it looks like to be successful and what it looks like to do meaningful work in science.

Some people really find their place in academia and of course they think it’s a wonderful place. But they can be less open minded or encouraging of trainees who want to explore other paths.

It’s definitely changing though. It’s not hard to see how to encourage people to explore other career options. When I was at MIT, an enormous amount of effort was put into making sure that PhD students at MIT thought startups were cool, knew who the VCs were, and knew how to start a company. It’s not hard to imagine doing that for other models. I think people need reasons to think it’s cool.

It’s hard to convince people to take a risk with their careers when there’s no one before them to look at. I think there’s a first cohort of founders right now in flight, experimenting with new models for science. At some point we’ll get to look to them to see what happens next.

Part of it is also the currency of success and the metrics of career advancement, it’s why there’s always such a focus on publishing. The orgs you’re building also are experimenting with what success looks like and with how to communicate results and build a broader community around you, how do you see that changing?

I do wonder if there could be other currencies. For example, in space science, doing a flight test is the ultimate marker of success, not a publication. Or beyond that kind of technical validation, if we had a better ability to share the methods and data that we generate, then you could have publishing be a little more like GitHub, where things are more trackable and you can actually put your finger on whether people are using technologies that other people create. That’s a really strong signal that’s much stronger than citations.

Today there’s a relatively small fraction of the types of data we generate or the types of materials we generate in life science that have venues where they can be shared and where sharing can be tracked. It’s things like plasmids with Addgene, protein structure files with PDB, and genomes with NCBI.

I think creating more of these venues where people can share things is net good. It moves us toward a world where you can have other things aside from just citations that you could point to that showed “I made this and people love it and rely on it.”

Do you think part of what stops people from sharing is how hard it is to do the experiments? That the data is so hard won that there has to be value from keeping that data for yourself?

I actually think it’s the opposite of that. I don’t even want to see someone else’s data. It’s going to be gross and weird in ways I can’t even imagine. It’s not that it’s weird and niche and therefore we want to hold it close, it’s that it’s weird and niche and therefore won’t be useful in someone else’s hands or someone else’s model. I don’t know what they did. Maybe they breathed on it—I don’t want that.

That’s the struggle. There are only a few types of data that—as much as we hate it—we do outsource. No one synthesizes their own oligos anymore. The same with sequencing. When something is standardized enough that it’s worth outsourcing, people do it.

There’s a lot of solvable problems that prevent us from doing good science. We should find a way to fix it, and then do it again for the next problem.

This is another part of the analogy to the history of the computer industry that drove synthetic biology in the early days—standards and open source but also service providers and supply chains. Are we still just early?

In the seventies there was a computer chip manufacturing standard called MOSIS. Prior to this standard occurring, there were a handful of different facilities that could manufacture chips that all had their own unique software for specifying chip designs. As a result, if you were trying to get a chip made, you had to place an enormous order because they’d have to do all this custom work for each one, they didn’t do small batch anything.

That also meant you were locked into the facility you were working with, because you couldn’t just turn around and get it manufactured somewhere else because you’d have to redo all your R&D to even write down what you wanted. As a result, prices were astronomically high.

Then they came up with a standard, which was the right level of compromise—there were some things that you could no longer specify as a result of this standard, you lost some customization ability, but it had enough specificity that you could do mostly everything you would ever want, and it was compatible with all the facilities, and suddenly the cost dropped because the facilities were competing with each other on price for manufacturing.

And because there was a universal software standard, you could batch different people’s orders together, which opened up custom chips to academics who suddenly were able to actually place orders.

I think there will be a MOSIS moment in experimental biology, where we finally come up with the right compromises for the level of specificity for defining the process of experiments. If that’s good enough for most experiments, then as a result everyone starts writing their experiment specifications in that language and it becomes more universal. That’s hard because there’s so many different types of biology and we don’t know which things we specify matter.

That diversity and variability in biology is part of its magic but also a huge challenge for finding these generalizable processes. How do you think that impacts our ability to invest across these areas and the economics of innovation in life science?

Consider the cost of the really big physics projects—every space project is $10 billion plus. Things like space telescopes and particle accelerators aren’t funded by industry because there aren’t many for-profits that would benefit from the outputs—knowing if there’s a Higgs boson is not directly related enough to for-profit activities for it to make sense for industry to fund this work meaningfully.

I actually believe we could do better in life science. Because these fundamental uncertainties about how biology works prevents for-profits in life science from making money. I believe we could do way better than physics in terms of creating new types of financing structures that allow for-profit money to get all the way down to very basic science.

Most of that effort and money today is optimized for small molecule therapeutics, but even in that context, those companies still struggle with knowing what is it that matters about an experimental protocol? How do we transfer protocols and data between labs? How do we determine which methods actually are useful to other people and which ones are just crap that got published in Nature or Science? All of that actually does help industry.

Even the work Pioneer is doing, where the goal is microbes that can live on Mars, what we’re doing is making high-throughput methods for getting data on microbial genotype to phenotype mapping towards building whole cell foundation models, which would be useful for for-profits here on Earth.

I can make a whole listicle of like Erika’s 10 weird financing ideas that she wants to try.

Can we? That sounds great. We should do that.

You’ve worked on building life from scratch and now kickstarting life on a new planet. You’re also building organizations and their cultures from scratch. Can you speak about the culture of technology and the community that you are seeking to build? There are very prominent messages about Mars, about AI models, about the future of biology and technology today coming from a lot of tech leaders, but your perspective feels different.

Most people can’t even imagine what a beneficial relationship between humanity and nature would look like. We don’t have to be exploitative. What I like to imagine in the context of Mars is, what would it look like for human presence somewhere to have positive externalities on the environment?

How can we as humans have a moral framework for reasoning about doing science that doesn’t make us incapacitated by fear that we are stepping on something. I hate the idea that humanity is constantly apologizing for our presence. I don’t think that’s the right way to do it.

I envy the Star Trek universe for their ability to “boldly go.” They don’t seem to apologize for their presence. I wish that I lived in a world where I had moral clarity that it’s okay for me to exist and be alive. Humans take up space and resources and that’s okay.

What I see in a vision for a Green Mars is a direction and a way to do that. It’s an invitation—here is our vision for what we want humanity’s presence to look like. Let’s go do that.