Kepler Found Thousands of Planets and Not One Earth Twin

David Kipping is an astronomer at Columbia University, director of the Cool Worlds Lab, and the host of a YouTube channel that walks through exoplanets, exomoons, and biosignatures more clearly than most papers. Cool worlds, in his usage, are planets outside the Solar System that are cold enough for moons, rings, and perhaps life. Lex Fridman brought him on to talk about those worlds, and about the quieter question that drives a minority of astrophysicists: why are we here, and are we alone?

NASA's Kepler mission is still the dominant source of known exoplanets. It stared for about 4.35 years and found thousands of them. What it did not find, Kipping said, is a single confirmed planet around a Sun-like star, on an Earth-like orbit, that could reasonably share Earth's composition. That zero is the starting fact of the conversation. The logic of why it is so hard to find a cool, terrestrial globe is also, he argues, the logic of the entire search for alien life.

The first half of exoplanet history was almost a monopoly of hot worlds. Kipping is blunt that this is an artifact of the tools, not of scientific taste. The two methods that first worked — Doppler spectroscopy, which watches a star wobble, and the transit method, which watches it dim — both have a strong bias toward planets that sit close to their stars. Close-in planets tug harder, transit more often, and pile up signal before a telescope's time runs out.

Hot planets are not boring. Their atmospheric chemistry is alien in the ordinary sense. One world close to Kipping's heart is TrES-2b, whose atmosphere is so dark it reflects less light than coal. That kind of object is photometrically bizarre and scientifically rich. It also resembles nothing like home. In his split of the field, one kind of astrophysicist asks how the machinery of the universe works — the Big Bang, expansion, galaxy formation. The other kind, the whisper in the ear, asks whether anyone else is out there. Hot planets will not get you to that answer.

What you want, he said, are more terrestrial globes: planets like Earth, or even moons around gas giants that could be habitable. Those are the ones that lurk in the shadows of detectability. Every property that makes a world cool enough to matter also makes it a worse target for the surveys we actually fly.

The workhorse is the transit: an eclipse in miniature, as a planet crosses its star and blocks a sliver of light. Not every planet does this. The orbit has to be lined up with our line of sight. About one percent of hot Jupiters transit. For Earth-like planets the geometric chance falls to about half a percent, maybe a quarter of a percent. Temperature and alignment are not directly linked; both are set by distance from the star. Farther out means cooler, and farther out also means a thinner slice of lucky viewing angles.

Frequency is the second trap. A hot Jupiter orbits in one to three days, so a week at the telescope can catch two or three full cycles. An Earth analog needs three or four years on the same star. Signal strength is the third. A true Earth crossing a Sun-like star dims the light by 84 parts per million. Kipping's comparison is a firefly in front of a giant floodlight at a stadium. Fridman offered a party image: you cannot see anyone's face, only who briefly occludes whom. Kipping pushed it further. Your vision is so blurry that the world is a smear, and the only clue is that a light got fainter for a moment.

Resolution cannot save you. The nearest stars are four light-years away; most of Kepler's targets were thousands of light-years away. A one-meter telescope cannot separate star from planet on the sky. Everything collapses into a single pixel that may contain dozens of planets, a star, even several stars. All you can do is watch that pixel dim. Humans have become detectives of a very faint crime scene, inferring whole worlds from a blink.

Kepler was supposed to last longer. It died after roughly 4.35 years. For an Earth-like planet, that is just enough time to catch four transits, which was treated as the minimum for a detection. More repeats would have stacked signal to noise — tick, tick, tick — until the pattern rose out of the noise. Four was the edge. The spacecraft stopped right where astronomers expected Earth analogs to start showing up.

So it saw zero. Kipping calls that a great shame and, in the next breath, work for the next generation. There is a technical story and a luck story: the mission's lifetime was simply a little too short. Cool worlds, from the beginning, have been a walk against the detection bias. The more a planet looks like Earth, the more the method's weaknesses stack. The zero is not a verdict that such planets do not exist. It is a map of where the cliff was.

The other path is direct imaging: pull the planet's light apart from the star's and actually take a picture. That method has its own favorites. You want the planet far from the star so the angular separation is large, the star nearby, and the planet itself hot, because hotter means brighter. Stack those preferences and you mostly catch planets that are still forming, still glowing with leftover heat. Earth-like worlds have already cooled. Their visible light is pitiful next to a newborn planet.

Kipping calls a photograph of a pale blue dot the dream, even if it is only one pixel. For now, with transits and with most telescopes, the entire Solar System is that one pixel. You do not even get the pixel of the planet. You get a brief fade in a blended point of light. He does not hide how hard this is. The fact that people can infer a world at thousands of light-years from that fade is, in his telling, something to be in awe of — and still not the picture anyone actually wants.

Smaller stars help. The same Earth-size planet blocks a larger fraction of a smaller star's light, so red dwarfs have become the most accessible hunting ground. TRAPPIST-1 is the exhibit. The star is about one-eighth the mass of the Sun and hosts seven planets smaller than Earth, detectable even from the ground. TRAPPIST-1e sits at a distance where liquid water could be stable. Kipping gives the numbers: roughly ninety percent of Earth's radius, eighty percent of the mass, and currently one of the top targets for possible life.

Then the caveats arrive. Late M dwarfs have a long adolescence. A Sun-like star settles onto the main sequence in something like a hundred million years. These small red stars can take a billion years or more to calm down, and during that stretch they pour out X-rays and ultraviolet radiation that can strip an atmosphere and desiccate a planet. Water delivered later by comets might still be lost in the prolonged storm. The near-term implication is a little disappointing: if we detect biosignatures soon, they will almost certainly be on a red-dwarf world, not on a twin of Earth circling a yellow star.

Finding something would still be progress. It just may not look like the postcard anyone drew in advance. The first inhabited world we can actually reach with data may be a place whose star spent an eon trying to boil it dry.

Life is hard to define. Kipping's workaround is to skip the definition and look at what life *does*, then hunt the traces. The classic answer was oxygen, a byproduct of photosynthesis. Earth did not always have it. After the Great Oxidation Event, oxygen saturated the oceans and then the air. Remove every plant today and oxygen would not linger. It is reactive. In about a million years it would oxidize rocks and minerals and disappear. A thick oxygen atmosphere on another world was therefore long treated as a photosynthesis flag.

That logic has loosened. Most of Earth's history had life and no oxygen, so oxygen as a sole test throws out most of the living record. Non-biological oxygen is also on the table. Ultraviolet light in the upper atmosphere can split water vapor; hydrogen escapes because it is light — the same reason Earth holds almost no hydrogen or helium — and leftover oxygen oxidizes the surface, leaving a residual signal. False positives have become a central worry in biosignature work.

The search list has grown. Nitrous oxide, laughing gas, is a microbial product. Methane by itself does not prove much; in combination with other gases it can. Phosphine is on the list too. The point is not a single magic molecule. It is a pattern of chemistry that is hard to explain without metabolism, and a growing humility about how many abiotic paths can mimic one.

Phosphine keeps coming up because of a claim, about two years before this conversation, of a phosphine-like spectral absorption in Venus's atmosphere. Kipping unpacks it in three layers. First, is the line even in the data? Second, if it is, is it phosphine or something like sulfur dioxide, which is not a biosignature? Third, if it really is phosphine, does that require life? "The judge and jury still out on that," he said. Scientific fights often sit in that state for years.

Venus's surface is a terrible place to imagine anything thriving. The dense atmosphere changes the story with altitude. High enough, temperature and pressure become fairly similar to Earth's surface. Microbes, in the speculation, could stir in those clouds and produce phosphine. That possibility has reinvigorated the case for going back. NASA has VERITAS and DAVINCI aimed at the 2030s; there is a European mission, and Kipping mentioned a possible Chinese one. Apart from the Soviet Venera era, Venus has been neglected next to Mars.

There is also an operational advantage. If life is there, it is airborne. A balloon or a drone could fly through and scoop samples. Mars and the icy moons require digging. For a field that has spent decades staring at a dusty surface, a cloud deck you can literally drift through is a different kind of experiment.

Mars has shown methane. The debate mirrors Venus: biology, which happens on Earth, or geology we do not fully understand, such as a subsurface reservoir leaking with the seasons. Europa is a harder problem. The ice thickness is still unclear, and melting through it to a buried ocean is an engineering nightmare. Worse than the engineering is contamination. Looking for life with a telescope is not the same as landing. Visit, and you may introduce the life yourself. Completely clearing every microbe and spore from a lander is close to impossible. Planets already exchange material naturally; spacecraft accelerate the traffic.

Europa's subsurface ocean has probably been sealed for a very long time. Drill it and you may be the first to send bacterial spores into that environment, where they could compete, grow, or be mistaken for the signal you came to find. Kipping calls this an ethical problem and admits there is no good resolution yet. There is a reverse risk too. Whatever survives those conditions is probably good at surviving in general, and might hitch a ride home. Sample return is not even required anymore. A small laboratory can run the experiments in place and radio the data back. If you did bring back "European sushi," he joked, New York billionaires would pay well for it.

Fridman, thinking as an engineer, wanted to see the scoopers: Venus, Europa, Mars. Curiosity has already been leaving sample tubes on the Martian surface that look, in photos, like lightsaber hilts, waiting for a later return mission. Kipping finds that engineering impressive. What stuns him more is landing. Light-travel time means no human can fly the descent. You trust the code. During the famous "seven minutes of hell," the vehicle must deploy a parachute, drop a heat shield, fire jets, switch to a sky crane, lower the rover, and fly the crane away so it does not hit the vehicle. Any step can end the mission. Ground teams wait for a signal that is already history. Consistently building machines that do this autonomously is, he said, one of NASA's most impressive engineering acts. Physics is unsentimental: takeoff is easier than landing.

Kipping's interest in Starship is specific: cost per kilogram. Even if the real price is ten times what Elon Musk originally promised, the drop would still be a revolution. Cheap mass to orbit means you can fly large things, including a telescope that does not have to fold. JWST's primary mirror is 6.5 meters. No fairing at the time could hold it, so the mirror became origami, with on the order of 130 failure points in the unfolding. The astronomical community held its breath through deployment. Designing and repeatedly testing that mechanism ate a large share of the project's cost.

If mass is cheap, you stop shaving honeycomb out of mirrors to save grams — the trick that made Hubble and Webb so expensive. You could take existing four- and five-meter ground-based blanks, bolt them to a chassis, and fly a space telescope. Kipping mentioned interest from the Breakthrough Foundation in that kind of approach. Follow the logic and you do not have one Webb. You have dozens. Today cosmologists, star-formation researchers, exoplanet scientists, deep-field observers, and people measuring the expansion rate all fight over the same observatory. Give each community a dedicated copy, staring at a handful of the brightest exoplanet host stars, and the science changes kind.

Kipping ran the numbers on TRAPPIST-1e. Point Webb at it for oxygen, nitrous oxide, methane, and you need on the order of two hundred transits to scrape a 2.5-sigma hint that almost nobody in astronomy would take seriously. The planet transits every six days. Two hundred events is about four years of catching nearly every pass, with little room to insert other programs. In principle, Webb has the sensitivity to nibble at a biosignature on an Earth-size planet around a non-Sun-like star. In practice, you cannot give the entire observatory to one target. Starship's promise is to turn that from a monopoly into a survey: many telescopes, a net instead of a single precious machine.

Who gets the hours is decided by a time allocation committee, or TAC. Kipping has served on them. Hubble was typically oversubscribed ten to one, sometimes twenty. Webb cycle 1 was about six or seven to one. Cycle 2, with a deadline at the end of that month and his team locked on proposals, he expected to approach twenty to one. The hard part is not only the ratio. Time-critical transits collide in non-obvious ways: different periods, durations, pointings, instruments. A deep field is easy. Point at a patch whenever the telescope is free and accumulate a month of photons over the mission. Cool worlds are the opposite. An alien watching Earth transit the Sun would get one twelve-hour chance per year. Miss it and wait. Hot planets offer a hundred opportunities, so a committee can shop among them. Kipping's team is asking for the events with one bite of the cherry.

Fridman, from a computer-science reflex, said you schedule the rarest jobs first. Webb is the first machine that can do cool worlds, and those observations are scarce, so the algorithm would put them at the front. Kipping laughed that he should apply to the committee. Then he noted the catch: TACs are not code. They use human judgment about excitement and chance of return. Within exoplanets, he expects most Webb time to go to atmospheric retrieval — detecting molecules, not biosignatures, which he thinks is pushing the telescope too far. A carbon-dioxide-rich atmosphere on TRAPPIST-1e would not prove or disprove life. It would say Venus-like or Mars-like, and it would be the first real characterization of pressure and temperature on that class of world. That science is almost guaranteed to learn something, because planets usually have atmospheres. Looking for moons you are not sure exist is a harder sell.

The Cool Worlds pitch is exomoons. The Solar System is crowded with them at similar scale: the Moon, Io, Callisto, Europa, Ganymede, Titan, each roughly thirty percent the size of Earth. That abundance suggests such bodies are ordinary. Kepler cannot find them. Hubble cannot. Webb is the first telescope that can. Kipping called it a new window. Most exoplanet time on Webb will still, he expects, flow to atmospheres, the safer product. His defense of moons is that a detection would already be a deep result. If other stars have an Io and a Europa, then icy bodies, water, and habitable niches are standard galactic equipment. You cannot score that inference by success rate alone.

NASA has chased a number called eta Earth: how often Earth-like planets occur. Kepler's main job was to measure it, and it only flirted with an answer. The preliminary picture is that Earth-like planets around Sun-like stars are not inevitable. The rate looks like one to ten percent. Another number is louder: about half of Sun-like stars have a mini-Neptune, a Neptune, or a Jupiter in the habitable zone. That architecture is common. If the goal is Earth-like *worlds*, Kipping said, moons have to be in the census. Planets are not guaranteed; large moons already exist in our own system in bulk.

The research path from cool worlds to exomoons is one question: is the Solar System's layout the usual outcome, or a freak? There is no shortcut. It needs bigger mirrors, longer stares, smarter scheduling, and people willing to bet telescope time on things that might not be there. Kepler's four-and-a-third years bought a zero. Kipping does not treat that zero as failure. It drew the boundary. The cliff was visible. The next machines have to step over it.