Poking Holes in the Mirror
Aaron Tohuvavohu's Mission to Make Space Telescopes Abundant
This article was drafted by Claude based on a long-form interview with Aaron (that we will release soon) and heavily edited by Ben, who takes full responsibility for the piece’s quality. Putting this at the top to encourage a norm to treat AI like a coauthor and that a human should be ultimately responsible for quality.
The most precious resource in space science isn’t money, rockets, or talent but telescope-seconds. The James-Webb Space telescope is oversubscribed by a factor of twelve (that is, for every hour of operational time, scientists have submitted serious proposals for twelve hours). The Hubble is oversubscribed by a factor of six, after 36 years of operation. The current paradigm of launching one exquisite instrument every decade means that thousands of proposals never get written at all, because astronomers have long since learned not to imagine questions their tools can’t answer.
Aaron Tohuvavohu wants to remove that constraint. As director of Cosmic Frontier Labs, a fifteen-person nonprofit research startup, he is building a fleet of high-performance space telescopes designed to be produced, launched, and iterated on at a pace astronomy has never seen. The organization hopes to usher in a continuous, evolving constellation of telescopes instead of one flagship per decade.
“We can’t build more time,” he says. “But we can build more telescopes.”
If it works, the effect on astronomy would be similar to the effect on biology when the number of electron microscopes went from two or three per country to ten per elite university: a quiet phase transition in the questions a field is allowed to ask.
The Shape of Scarcity
Telescope design decisions constrain discovery. Every space telescope sits in a high-dimensional parameter space — collecting area, angular resolution, spectral bandpass, temporal cadence, survey speed, field of view, among many other parameters. “Flagship missions” like James Webb or Hubble, maximizes a few of these axes.
But any telescope that costs billions of dollars necessarily minimizes its ability to just stare at something cool. Consider a promising exoplanet—an Earth analog at the right orbital distance, plausibly habitable. If you wanted to watch its climate vary across its year, you would need to observe it continuously for the better part of that year. Nothing in orbit will ever do this. Hubble and Webb have a thousand other obligations.
There are so many other experiments that the obligations of exquisite telescopes rule out. For example reverberation mapping: tracking the changing colors in the accretion disk around a supermassive black hole, every hour for months, across UV through infrared, to map its physical structure. Or catching a supernova in its first hours, which requires both rapid retargeting and the institutional willingness to abandon whatever was previously scheduled. Or survey science at surface brightnesses so low you need two full days on a single target to map gas inflows in the cosmic web.
“These aren’t technically impossible with what we have on orbit,” Aaron says. “They’re operationally and programmatically impossible. We just choose not to operate them that way—because we can’t afford to.”
Beyond the new science promised by unlocking high quality telescope time for every scientist who wants to look, abundant telescopes may also unlock next generation instruments themselves. If we could make space telescopes at scale, it would enable truly enormous “synthetic” telescopes whose size is independent of how big a thing we can launch. The intuition is that all telescopes depend on a mirror that collects light from an aperture and delivers it, in phase, to a focus point. Staying in phase means that every ray has to arrive having traveled the same optical path, to better than roughly a tenth of a wavelength. For visible light, that’s about fifty nanometers. It turns out you can punch large holes in the mirror and still form an image, as long as what remains stays aligned to that fifty nanometer tolerance. In the limit, the pieces of the mirror that remain don’t need to be connected. You can put one fragment here, another fragment ten kilometers away, and as long as you maintain a known distance between all the pieces, you could do insane things like resolve features on the surfaces of other stars and directly image planets around them.
Of course, maintaining a fifty nanometer tolerance between many separate telescopes is hard. The current state of the art in space formation flying is ESA’s Proba-3, which last year held two spacecraft a hundred meters apart to roughly one millimeter. One millimeter is a few orders of magnitude more than fifty nanometers, but there are potential technologies to achieve it and without fleets of space telescopes, nobody is going to push the limits on them.
“Building many space telescopes is not the hardest part of that problem,” he says. “But it’s a necessary precondition. And that’s where we’re starting.”
The Long Road to the Lab
Aaron was not always a hardware guy. He started in theoretical physics, working on chalkboard models of how black holes form and radiate gravitational waves. He did his PhD at the University of Toronto in the then-nascent field of multi-messenger astrophysics, combining observations of gravitational waves, photons, and occasional neutrinos to probe environments no terrestrial laboratory can reproduce. Along the way, he led Toronto’s Space Telescope Development Group and was principal investigator on an ultraviolet CubeSat camera.
What changed through that increasingly hardware-focused trajectory was an understanding of the actual bottleneck in astronomy.
“When I started, I wanted to use astrophysical systems as laboratories to test fundamental physics. That’s still true. But what I’ve learned is how the questions you’re allowed to ask are shaped by the tools available—and by what you can imagine the tools doing. Many of my colleagues, who are extraordinarily talented, won’t let themselves think about what science would look like with better tools. Because it’s depressing. Because they can’t have them.”
A decade of watching precious missions get designed, descoped, queued, and rationed convinced him that the way the field builds and operates its most important instruments is “slow and boring and structurally scarce.”
What Astronomers Stopped Asking
There is a historical observation tucked inside Aaron’s vision. A hundred and fifty years ago, the people who made the biggest astronomical discoveries were often the same people who designed the telescope, ground the glass, and hauled it up the mountain. That changed for good reasons: specialization, project scale, technical sophistication beyond any one person’s reach. Observing and instrument-buildering became separate professions, observatories became user facilities, and more astronomy got done.
But the downside is that now most observational astronomers no longer deeply understand how their instruments work. And when you don’t understand your tool, two things happen: First, you do worse science at the edge of knowledge, where pulling real signal out of marginal photons requires knowing exactly how your detector lies to you. Second, you lose the ability to push back when engineers or program managers say something is impossible.
“A lot of the time, the answer [to whether something is possible] is: the magic space wizard said it was hard,” Aaron says. “And I’m serious. You go to your friend with an amazing idea—an X-ray formation-flying interferometer, say—and you ask how hard it would be, and they say, well, I talked to an engineer at NASA, and they said this and that. And when you actually start asking why, you discover most astronomers never asked. Because it’s socially unacceptable to not defer to engineering expertise, even when that expertise is only adjacent to the problem.”
The thing he thinks astronomy is most wrong about, then, isn’t a specific scientific question, but this epistemic posture. He wants astronomers to stop accepting that something can’t be done because “it’s hard.” Instead, astronomers (and everybody else!) should start poking at the assumptions – which parts of the cost are technical, which are programmatic, and which are simply the accumulated sediment of how it’s always been done? The way to earn the right to push back, he argues, is to become a real expert in the hardware yourself.
From Whitepaper to Spacecraft
The founding story of Cosmic Frontier Labs is almost embarrassingly social.
Aaron joined the first cohort of the Brains research accelerator after encouragement from a friend to develop his ideas around fixing telescope scarcity. Brains acted as a forcing function for turning his conviction into a concrete plan. The fellowship gave him the space to sharpen the idea, the network to find the first believers, and the legitimacy to be taken seriously in rooms that wouldn’t otherwise have opened.
As part of the program, Aaron wrote a short whitepaper (note that, like letters, the difficulty of writing a whitepaper is inversely proportional to length) which diffused through the network, apparently with enough of the right words and the right framing at the right moment, until it reached wealthy sponsors willing to fund a first try.
Cosmic Frontier Labs formally started on January 1, 2025. By March, there were three people. Fifteen months in, there are now fifteen. About a third of the team are controls engineers, because diffraction-limited imaging in orbit is first a pointing problem and second an optomechanics problem and third everything else. Hardware is on the manufacturing floor, environmental testing has begun, and a launch is booked for the first flight system 529 days from the interview (there is a visible counter in the office).
There is a single metric that applies to every engineering decision: dollars per high-quality observing second. The numerator is the amortized total cost of the system. The denominator is a complicated integrated model of performance over time. Cosmic Frontier Labs aims to drive that number lower than $1 per observing second – 1/5 the current cost of the Keck Observatory, one of the best land-based telescopes.
What He Would Tell His Younger Self
Asked what advice he would give the version of himself from five years ago, Aaron gives two answers:
First: actually become an expert. Not a vibes-level skeptic of the establishment, but someone who has built the hard thing with their own hands, who knows what the real constraints are and where the soft assumptions are hiding. If the problem is a hardware problem, start building it, even with scrap parts, and let momentum compound.
Second: write publicly about the crazy idea. Light a beacon. When you start saying out loud what should be built, many of the people in your field will dismiss you as naive or deluded or young. But the other nerds—the believers, the fellow travelers you don’t yet know exist—will find the beacon and come to it. He didn’t do this enough, and thinks things would have gone faster if he had.
Both pieces of advice describe the same person. Someone willing to be publicly wrong about something important, and willing to spend a decade getting good enough to eventually be right. Someone who read one too many schedules that said fifteen years and a billion dollars, and decided to find out whether that was actually a fact about the Universe or just an unexamined convention.
The first flight is less than 500 days away. The counter is ticking down.