You're an astronaut on the Moon. Long day of moonwalking, regolith sampling, flag-planting. You kick off your boots, pull up Netflix, hit play on Stranger Things. And nothing happens. Two point seven seconds of nothing. Then it buffers. Then it fails.
Because the speed of light is too slow.
Which is a sentence you don't hear every day.
No, it's genuinely the problem. Light in a vacuum travels at two hundred ninety-nine million, seven hundred ninety-two thousand, four hundred fifty-eight meters per second. That's the universe's posted speed limit. Nothing negotiates it. And the Moon is three hundred eighty-four thousand four hundred kilometers away, so even at that blistering speed, a signal takes about one point two eight seconds just to get there. Another one point two eight to come back. Round trip, two point five six seconds. That's your ping time. To your Netflix server. On a good day.
You're telling me my Moon Netflix experience is bottlenecked by the fundamental structure of spacetime.
It gets worse. That two point five six seconds is the bare minimum latency, before any processing, routing, or buffering. And the internet as we built it, the protocols that deliver Netflix to your phone right now, were designed for round-trip times measured in single-digit milliseconds. Throw in two and a half seconds of delay and the whole thing seizes up like an engine without oil.
The question Daniel sent us is really two questions. First: how does space communication actually work, from Apollo to the ISS, given the insane physics involved. And second: could an astronaut on the Moon stream Netflix from Earth, and if not, how far away from that are we. Which is a beautiful way to frame it, because it turns a physics lecture into a test case.
The physics is brutal. You've got a transmitter on a spacecraft, maybe twenty watts on the old Apollo Lunar Module, and you're trying to talk to Earth across nearly four hundred thousand kilometers of vacuum. The signal spreads out in a sphere. By the time it reaches us, it's spread across an area proportional to the distance squared. The inverse-square law means doubling the distance quarters the signal power. At lunar distance, you're trying to pick out a whisper from across a continent.
They did it. In nineteen sixty-nine. With computers less powerful than a modern microwave.
They did it with the Deep Space Network, three massive dish complexes spaced roughly a hundred twenty degrees apart around the Earth. Goldstone in California, Madrid in Spain, Canberra in Australia. So as the Earth rotates, at least one complex always has line of sight to the spacecraft. The dishes are seventy meters across. They can pick up signals that are something like ten to the twentieth power times weaker than what your cell phone handles.
Ten to the twentieth. That's the kind of number that stops meaning anything.
It's the difference between a shout and a gnat clearing its throat.
Yet the Apollo TV broadcast made it through. Grainy, ghostly, ten frames per second, three hundred twenty lines of resolution. About half a megabit per second effective. People watched Neil Armstrong step off the ladder, live.
There was a delay baked into the physics. But yes, it worked. And that was S-band radio, around two point two to two point three gigahertz, with about a ten megahertz channel. The Lunar Module had a twenty-watt transmitter. That's less power than a refrigerator light bulb, beaming across the void.
The core tension here is that space communications sits at this weird intersection of three constraints. Latency, which is the speed-of-light problem and you simply cannot negotiate with it. Bandwidth, which is about what frequencies you're allocated and how you modulate your signal. And throughput, which is the actual bits-per-second you manage to push through after all the physics has taken its cut.
Those three constraints interact in ways that are not obvious. You might think, okay, we'll just crank up the power. Build bigger dishes. But the Shannon-Hartley theorem sets a hard ceiling on how much data you can push through a given channel at a given signal-to-noise ratio. And distance destroys signal-to-noise ratio. Every time you double the distance, you don't just halve the data rate. It's worse than that.
Which brings us back to our astronaut, sitting in her habitat, staring at a buffering wheel. The question isn't just "can we beam Netflix to the Moon." The question is: what would it take, technically, to make that work, and at what point does the physics say no.
The answer turns out to be weirder than most people expect. Because the raw bandwidth might actually be there, or nearly there. The ISS today gets about three hundred megabits per second downlink through Ku-band and the Tracking and Data Relay Satellite System. That's enough for HD video, multiple streams. But the latency, those two and a half seconds, breaks the internet's core protocol in a way that makes streaming nearly impossible.
We'd have the pipe, but the plumbing doesn't work.
The plumbing was designed for a house, and we're trying to run it across a solar system.
Let's dig into this. Latency, bandwidth, throughput. And we'll use the Moon-Netflix question as our test case throughout. Because if you can understand why Netflix breaks at the Moon, you understand the entire problem of interplanetary communication.
Let's start with the one you can't negotiate with.
The speed of light in vacuum is two hundred ninety-nine million, seven hundred ninety-two thousand, four hundred fifty-eight meters per second. That's not a technical limitation we can engineer around. It's baked into the structure of the universe. Earth-Moon distance of roughly three hundred eighty-four thousand four hundred kilometers means a one-way trip for any signal is about one point two eight seconds. Round trip, two point five six. That's your absolute minimum ping time, the floor below which nothing can ever go, no matter how clever your hardware gets.
Just to put that in perspective, the typical ping time between New York and London on a good fiber connection is about seventy milliseconds. So we're talking roughly thirty-five times slower. And that's just the Moon. Mars at closest approach, fifty-four point six million kilometers, is three minutes one way. At its farthest, four hundred one million kilometers, you're looking at twenty-two minutes. A conversation becomes impossible. You send a question, go make lunch, come back, and the answer is arriving.
Which is why the Apollo missions didn't use any kind of packet-switched protocol. There was no TCP, no acknowledgments in the modern sense. It was a continuous stream of telemetry and voice, modulated onto an S-band carrier at around two point two to two point three gigahertz. The Lunar Module had a twenty-watt transmitter. That's less than the light bulb in your desk lamp, beaming across nearly four hundred thousand kilometers of vacuum. The signal, by the time it reaches Earth, is spread across an enormous sphere. The inverse-square law means the power per unit area drops with the square of the distance. At lunar distance, the signal is unbelievably faint.
The Deep Space Network was built specifically to catch that whisper. Three complexes, spaced about a hundred twenty degrees apart around the globe, so that as Earth rotates, at least one is always facing the spacecraft. Goldstone in the Mojave, Madrid in Spain, Canberra in Australia. The big dishes, the seventy-meter ones, have a collecting area of about thirty-eight hundred square meters. They can pick up signals that are something like ten to the twentieth power times weaker than what your phone handles.
They have to track the spacecraft continuously. The Moon is not stationary. The Earth is rotating. The spacecraft is orbiting. So these enormous dishes, weighing thousands of tons, are constantly slewing to maintain lock on a signal source that's moving through the sky. The pointing accuracy required is on the order of thousandths of a degree. If you're off by a hair, you miss the signal entirely.
Which makes the whole thing sound like trying to thread a needle while riding a bicycle. Across a football field. In the dark.
That's not far off. And this brings us to the second constraint, bandwidth. The Apollo missions used S-band, around two point two to two point three gigahertz, with a channel about ten megahertz wide. That frequency band was chosen partly because it's relatively unaffected by the ionosphere, partly because the technology was mature, and partly because it was what the international frequency allocation bodies had set aside for space operations. The International Telecommunication Union coordinates all of this. Every frequency band is contested real estate. You can't just decide to broadcast on whatever frequency you want. You need an allocation, and space operations have specific bands reserved.
Bandwidth is partly physics and partly bureaucracy.
That's a fair summary. The physics says higher frequencies can carry more data. If you double the frequency, you can roughly double the data rate, all else being equal. That's why modern systems have moved up. The ISS uses Ku-band, twelve to eighteen gigahertz, for its high-rate data link, getting three hundred megabits per second downlink through TDRSS, a constellation of geostationary relay satellites. But higher frequencies come with tradeoffs. They're more directional, requiring more precise antenna pointing. And they're more susceptible to atmospheric attenuation. Rain fade is a real problem for Ku-band. A heavy storm between the ground station and the satellite can knock out the link.
You climb the frequency ladder for more speed, and you trade away robustness.
Then there's modulation. How you encode the data onto the carrier wave. Apollo used phase modulation, state of the art at the time. Modern systems use much more sophisticated schemes—quadrature amplitude modulation, orthogonal frequency-division multiplexing, the same techniques that power your Wi-Fi and 4G. But the fundamental Shannon-Hartley limit still applies. The maximum data rate you can push through a channel is the bandwidth times the logarithm of one plus the signal-to-noise ratio. And that signal-to-noise ratio, that's where distance is the silent killer.
The three constraints are interlocked. Latency is fixed by physics. Bandwidth is allocated by treaty and limited by frequency choice. And throughput, the actual bits you manage to push, is governed by this theorem that ties it all together.
The theorem is ruthless. Every time you double the distance, the signal power drops by a factor of four. Your signal-to-noise ratio drops. And the logarithm in Shannon-Hartley means that as your SNR falls, your data rate falls with it, and eventually you hit a cliff where you're getting almost nothing through. That's why Voyager One, at twenty-four billion kilometers, transmits at a hundred and sixty bits per second. That's slower than a nineteen-eighties dial-up modem by a factor of three hundred. And that's using the full power of the DSN's seventy-meter dishes, cryogenically cooled receivers, the most sensitive radio equipment ever built.
The Moon-Netflix question isn't just about whether we have enough raw bandwidth. It's about whether the entire stack, from physics through protocol, can sustain a streaming session. And that's where we need to look at what happens when you try to run Earth's internet protocols across a two-and-a-half-second delay.
Let's talk throughput. The actual bits per second. Apollo's Lunar Module, that twenty-watt S-band transmitter, managed about half a megabit per second effective for the TV broadcast. Ten frames per second, three hundred twenty lines of resolution. That's a postage stamp of video, and it was a triumph. Today, the ISS gets three hundred megabits per second downlink through Ku-band and TDRSS. That's a factor of six hundred improvement in fifty-some years.
The ISS uplink is only twenty-five megabits per second. Hugely asymmetric, which makes sense. Astronauts are consuming far more data than they're generating. Commands go up, high-def video comes down. Three hundred megabits is enough for multiple HD streams. It is not enough for raw 4K, which Netflix pegs at twenty-five megabits per stream. You could do one or two 4K streams if the link were dedicated, but the ISS shares that pipe with experiments, telemetry, voice, everything.
The raw bandwidth at lunar distance, with current tech, is in the ballpark. But now push farther. New Horizons at Pluto, seven and a half billion kilometers out, managed one to two kilobits per second. Voyager One, at twenty-four billion kilometers, is down to a hundred and sixty bits per second. That's eight characters per second if you're sending ASCII text.
This is where the Shannon-Hartley theorem stops being an abstraction and becomes the brutal scorekeeper. The maximum data rate is the channel bandwidth times the logarithm base two of one plus the signal-to-noise ratio. That logarithm is the killer. When your SNR is high, you get nice linear gains from adding bandwidth. But as distance crushes your SNR, the logarithm flattens out. You double the distance, the signal power quarters, the SNR quarters, and the log of a number approaching zero is itself approaching zero. You fall off a cliff.
Why not just build bigger dishes? Crank up the transmit power?
We basically have. The DSN's seventy-meter dishes are already among the largest steerable radio antennas on the planet. You could theoretically build a hundred-meter dish, but the cost scales horribly. The structure has to maintain its parabolic shape to within a fraction of a wavelength while slewing to track a spacecraft. For S-band, that's about a centimeter of surface accuracy across a seventy-meter span. The engineering is already at the edge of what's practical. And on the spacecraft side, you can't just add power. Every watt costs mass, and mass is the currency of spaceflight. Apollo's twenty-watt transmitter was a considered tradeoff against fuel, life support, scientific instruments.
Even if you could double the dish size, it only buys you a factor of four in signal. One extra step on the distance ladder and you've lost it again.
The inverse-square law means you're fighting an exponential battle with linear weapons. Every time you double the distance, you need to quadruple the collecting area just to stay even. By the time you're at Mars, the numbers get absurd. To get ISS-level throughput at Mars's closest approach, you'd need a dish on Earth roughly the size of a small city. Or a transmitter on the spacecraft that would require its own nuclear power plant.
The relationship between distance and throughput is basically: you lose, and then you lose faster.
That brings us to the protocol problem. Because even if you solve the raw throughput question, even if you somehow get three hundred megabits to the Moon, the internet's core plumbing still breaks. TCP, the Transmission Control Protocol, was designed in the nineteen seventies for a network where round-trip times were measured in milliseconds. It has a mechanism called slow-start. When a connection opens, TCP doesn't just blast data at full speed. It sends a few packets, waits for acknowledgment, then ramps up.
It's cautious. Like a driver tapping the brakes to check for ice.
On Earth, that caution is smart. The ramp-up takes maybe a few hundred milliseconds. But on a lunar link with a two-and-a-half-second round trip, slow-start becomes glacial. Every ACK takes two and a half seconds to arrive. So TCP spends forever inching up its transmission window, never reaching full speed before the user gives up and throws their tablet at the wall.
NASA actually tested this.
They did, in twenty thirteen. They simulated a lunar-distance link and ran standard TCP over it. Throughput collapsed to about one percent of the available bandwidth. You'd have a three-hundred-megabit pipe and you'd be getting three megabits through it, with constant stuttering. The protocol simply wasn't built for celestial distances.
The pipe is there, the water pressure is fine, but the valve keeps thinking the pipe is broken and shuts itself off.
The fix exists. It's called delay-tolerant networking, DTN. Instead of requiring instant acknowledgment for every packet, DTN uses a store-and-forward model. Data gets bundled, transmitted, and stored at intermediate nodes. Acknowledgments are optional. If a link drops, the bundle just waits until the connection returns. It's the postal service instead of a phone call.
This isn't theoretical. The ISS has been running DTN experiments since two thousand nine.
It's proven technology, just not widely deployed. And this is where the Artemis program gets interesting. NASA's targeting a lunar return by twenty twenty-seven, with the Lunar Gateway station in orbit around the Moon. Gateway is going to carry an optical communications terminal, using lasers instead of radio. The Laser Communications Relay Demonstration, LCRD, launched in December twenty twenty-one, already proved this works. It achieved one point two gigabits per second from geostationary orbit to the ground. That's ten to a hundred times what you get from equivalent radio systems.
Like we're living in the future.
We are, but the future has fine print. Optical links require microradian pointing accuracy. You're shooting a beam of light at a target that's moving through space, and if your aim is off by a thousandth of a degree, you miss. And clouds block lasers. If your ground station is under cloud cover, the link is dead. So you need multiple ground stations, or you relay through satellites above the weather.
Laser is a huge leap forward, but it's not a magic wand. You trade radio's relative forgiveness for laser's raw speed, and you accept that weather becomes your enemy.
For the Moon-Netflix question, optical plus DTN changes the calculation entirely. Netflix recommends twenty-five megabits for 4K streaming. At lunar distance, with Ku-band radio through a TDRSS-like relay, you could theoretically get three hundred megabits down. That's more than enough raw bandwidth for 4K. But the protocol problem kills you. Netflix uses adaptive bitrate streaming over TCP. The player requests chunks of video, measures how fast they arrive, and adjusts quality. With a two-and-a-half-second round trip and TCP slow-start, the player would see terrible throughput, degrade to the lowest quality, maybe three hundred kilobits per second, and still buffer constantly.
You'd be watching Stranger Things in potato resolution, with a spinning wheel every ten seconds?
If you're lucky. The real solution is to pre-load content locally. Put a media server on the Moon. Sync it during off hours using DTN bundles. The astronaut streams from the local cache, no latency, no protocol collapse. That's not streaming from Earth in the Netflix sense, but it delivers the same experience.
Which feels like cheating, but it's the honest answer. You don't stream from Earth.
If you insist on true streaming, optical links and a custom UDP-based protocol with enormous buffers could make it viable by around twenty thirty. You'd need to abandon TCP entirely, design something that expects two-and-a-half-second gaps and just keeps shoveling data. It wouldn't be interactive. You couldn't scrub through the timeline without a painful delay. But you could hit play and watch.
How far out could you push that? Where's the actual cliff where even optical plus custom protocols can't sustain a Netflix stream?
The Shannon-Hartley limit gives us a rough answer. Assume a ten gigahertz carrier with a five-hundred-megahertz channel, a seventy-meter dish on Earth, and a one-meter dish on the spacecraft. You could theoretically sustain twenty-five megabits per second out to about two to three million kilometers. That's five to eight times the Moon's distance. Beyond that, the data rate drops below streaming viability.
The Moon is comfortably inside the streaming bubble. Mars is not.
Mars at closest approach is fifty-four million kilometers. At that distance, with the same setup, you'd get maybe a hundred kilobits per second. That's enough for text, email, still images if you're patient. A Martian astronaut is not streaming anything from Earth. They're running a fully local media library, synced during the weeks when Earth and Mars are closest, and even that sync is a trickle.
A hundred kilobits per second. That's dial-up. The future is amazing and also deeply disappointing.
What do you actually do with all of this if you're building a mission? The first rule is: do not assume Earth internet protocols work. They don't. TCP collapses, as we've seen. You need delay-tolerant networking, large local caches, and an architecture that treats connectivity as intermittent by design. The ISS has been running DTN experiments since two thousand nine. This is not speculative. It's deployed.
The ISS experience teaches something subtle. It's not just about the protocol. It's about designing the entire mission around asynchronous communication patterns. You don't ask "can I stream this." You ask "what data needs to be where, and by when." You pre-position. You accept that real-time is a luxury that stops at about one light-second.
Which is a mindset shift. Terrestrial internet taught us to expect instant everything. Space forces you back to something closer to the age of sail. You send a request, you wait, something comes back. The waiting is part of the architecture, not a bug.
That's the real takeaway from the Moon-Netflix question. It's a fun thought experiment, but what it actually reveals is that latency, not bandwidth, is the fundamental barrier to a truly interplanetary internet. We've been conditioned by decades of broadband marketing to think speed means bits per second. But speed also means seconds per bit. And at planetary distances, seconds per bit is what dominates.
Every deep-space mission from here on out has to treat communication as a first-class engineering constraint, not an afterthought. You don't design the rover and then ask comms to figure out how to get the data home. You design the rover around the comms budget. Because the comms budget is physics, and physics doesn't negotiate.
For anyone listening who wants to get hands-on with this, there's a simple exercise. Pull up a speed-of-light calculator and compute the round-trip time to any celestial body you're curious about. Then ask: at that latency, what would break? What protocol would seize up? What experience would degrade? And the next time you see a headline about laser communication to Mars, ask the question that almost never appears in the press release: what's the bitrate at that distance? Because that number tells you the real state of the art.
It's the difference between "we established a link" and "we streamed a movie." The first is a press release. The second is a civilization.
That leaves us with one big open question. As we push toward Mars and beyond, are we going to need a fundamentally new communication paradigm? Something that sidesteps the speed of light entirely? Because right now, everything we've talked about—radio, laser, DTN—it's all just coping with the speed of light. None of it beats the speed of light.
People bring up quantum entanglement. Two particles, entangled, measure one and the other instantly reflects the state, no matter the distance. Einstein called it spooky action at a distance. And every few years someone writes a headline suggesting we'll use it for faster-than-light communication.
Every few years physicists have to gently explain that no, you cannot. Entanglement lets you correlate measurements, but you can't use it to send information. If I measure my particle, I get a random result. I know your particle now has the correlated state, but I can't control what that state is. I can't encode a message in it. It's like we both open identical envelopes containing random letters. We know they match, but we can't choose what the letter says.
Quantum entanglement as a communication channel is the universe's most elaborate practical joke. Instant correlation, zero information transfer.
Neutrino beams have been proposed too. Neutrinos pass through almost anything, so you could theoretically beam a signal straight through the Earth, straight through the Sun, no line-of-sight required. But the problem is they pass through almost anything. Your detector has to be enormous and buried in a mountain or under Antarctic ice to catch even a handful of interactions. Fermilab demonstrated a neutrino-based message in twenty twelve. They encoded the word "neutrino" and sent it through two hundred forty meters of rock. The data rate was zero point one bits per second. With a particle accelerator and a multi-ton detector.
We've ruled out quantum entanglement, and neutrino beams are a bit per ten seconds through a mountain. Not exactly a Netflix competitor.
Which means, for the foreseeable future, we are stuck with electromagnetic waves and the speed of light. But that's not a failure. It's a design constraint. The next decade is going to be about making the most of what we've got. Optical links will become standard. The Artemis Lunar Gateway will carry a laser terminal. NASA's working on an optical relay network that could eventually serve as the backbone for a cislunar internet. And DTN is maturing from experiment to operational infrastructure.
We're not beating the speed of light. We're building a solar-system-scale network on a planetary-scale protocol stack, and learning to live with the delays.
That's the cosmic perspective, really. The speed of light is not just a speed limit for travel. It's a speed limit for culture, for conversation, for the shared experience of being human across distances. A Martian colony will never be part of Earth's real-time internet. It will have its own internet, its own media servers, its own cultural moment, synchronized in bursts when the planets align. We're not building one internet. We're building a network of internets, separated by light-minutes, loosely coupled.
Which is strangely beautiful. The solar system as a federated network. Each planet a node with its own cache, its own local culture, occasionally syncing with the rest.
The Moon is the first real test of that architecture. Close enough that the latency is annoying but not conversation-breaking. Far enough that Earth's internet won't just extend there seamlessly. It's the boundary between the terrestrial and the interplanetary. The place where we learn to build the stack that will take us everywhere else.
Daniel, the answer to your question is: no, an astronaut on the Moon cannot stream Netflix from Earth today. Not in any watchable way. But give it a few years, some lasers, some new protocols, and a willingness to pre-load the season finale. And the real answer is that asking the question teaches you everything important about the problem.
Now: Hilbert's daily fun fact.
Hilbert: During the late Victorian period, an enterprising telegraph operator stationed in the Kuril Islands discovered that the local arctic foxes could be trained to carry small spools of wire across frozen straits between islands, establishing temporary field telegraph connections where undersea cable had been severed by ice flows. He maintained a team of twelve foxes, each named after a different European telegraph code abbreviation, and claimed they could lay a hundred meters of line in under four minutes.
...right.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you enjoyed this, go tell someone the speed of light is why they can't binge-watch Stranger Things on the Moon. And you can find us at my weird prompts dot com.
We'll be back with another one soon.
