Daniel sent us this one, and it's got layers. He's asking about using the stars as a practical navigational fallback — his father worked in the oil industry in Africa, taught him some constellations as a kid, but the details have faded. The core question is: if GPS goes down, whether from jamming or infrastructure failure, can you actually pull your position or orientation from the night sky? And the second layer — is there an app that can do it for you, just point your phone at the stars and get a reading? It's a question about ancient technique meeting modern fragility.
The timing on this is sharper than most people realize. GPS jamming isn't theoretical anymore — Baltic airspace has been dealing with persistent interference for months now, commercial flights losing navigation mid-route. The Russians have been running electronic warfare exercises that spill over into civilian bands. Finland and Sweden have both filed formal complaints with ICAO. This isn't a hypothetical prep for some distant scenario — it's happening over Europe right now.
The fallback isn't just for when you're lost in the woods. It's for when you're in a city and the entire positioning layer flickers out.
And here's the thing about the North Star that most people get wrong — it's not the brightest star in the sky. Not even close. Polaris is about the forty-eighth brightest star. The reason it works is its position, not its brightness. It sits almost exactly above the Earth's rotational north pole, within about zero point seven degrees. So while every other star traces arcs across the sky as the Earth rotates, Polaris stays put. It's the pivot point.
The pivot point. So the entire sky rotates around a single fixed nail, and if you can find that nail, you know which way north is.
More than that — you know your latitude. The angle of Polaris above the horizon equals your latitude in the northern hemisphere. If Polaris is forty-one degrees above the horizon, you're at forty-one degrees north. That's it. No instruments needed beyond your own hand. Hold your fist out at arm's length — that's roughly ten degrees. Stack fists from the horizon up to Polaris, count the fists, multiply by ten, and you've got your latitude within a few degrees.
Wait, so this is the original global positioning system, and it's just... angle measurement with your thumb.
With your thumb. And sailors were doing this with instruments like the astrolabe and the cross-staff going back to at least the fifteenth century. The Portuguese navigators used the North Star to sail down the coast of Africa. But here's where it gets interesting — once you cross the equator, Polaris drops below the horizon and disappears. So southern hemisphere navigation had to use entirely different stars.
Which is of course where the Southern Cross comes in. But that one doesn't sit at the south celestial pole the way Polaris does for north. It's offset.
Right, the Southern Cross points toward the south celestial pole, but it's not sitting right on it. You have to extend the long axis of the cross about four and a half times its length and you'll hit a point in empty sky that marks south. There's no bright southern pole star. The closest is Sigma Octantis, which is so dim it's basically useless for practical navigation. So southern hemisphere sailors had to get more creative. They used multiple constellations together — the Southern Cross plus the two pointer stars, Alpha and Beta Centauri, to triangulate.
The North Star is almost absurdly convenient by comparison. A single bright-enough star, fixed in place, directly giving you latitude and direction. It's like the universe put a navigation beacon at the north pole and forgot to put one at the south.
The real mind-bender is that it won't last. The Earth's axis precesses — it wobbles like a spinning top over a cycle of about twenty-six thousand years. Right now Polaris happens to be extremely close to the north celestial pole, but around three thousand BC, the pole star was Thuban in the constellation Draco. The ancient Egyptians may have used it to align the pyramids. And in about twelve thousand years, Vega in Lyra will be the north star. So we're living in a particularly convenient window for northern hemisphere navigation.
A twenty-six-thousand-year wobble, and we happened to get the good parking spot. That's almost suspicious.
Now, the second part of the prompt — can you use a smartphone camera to get your position from the stars — the answer is yes, and the technology is surprisingly mature. The core technique is called star-field identification. A camera captures an image of the night sky, an algorithm identifies the pattern of stars in the frame, matches it against a catalog, and from that it can compute the camera's orientation and, with enough processing, its position.
It's doing what a sextant does, but instead of measuring one star's angle above the horizon, it's pattern-matching the entire visible sky against a database.
And this isn't just a smartphone gimmick — it's how spacecraft navigate. Satellites and interplanetary probes use star trackers as their primary orientation system. They have catalogs of thousands of stars with known positions, and they match what they see against the catalog to determine exactly which way they're pointing, often to arcsecond precision.
Arcsecond precision from pattern-matching dots of light. That's wild. But a spacecraft doesn't have to deal with clouds, light pollution, or the phone moving around in someone's hand.
Right, and those are the real constraints. But there are apps that do this on the ground. SkyView, Star Walk, Stellarium — these are primarily augmented reality astronomy apps, but the underlying engine is exactly what you'd need. You point your phone at the sky, the app identifies the star pattern, and it overlays constellation labels and object names on the screen. That's orientation solved — the phone knows precisely which direction it's looking, and it can give you your compass heading from that.
Can it give you your location? Orientation is one thing. Knowing you're facing north-northwest is useful, but it doesn't tell you where on Earth you are.
That's the harder problem. To get a full position fix from the stars — latitude and longitude — you need more than just star identification. You need to measure the altitude of specific stars above the horizon and combine that with precise time. That's how celestial navigation at sea works. You take a sighting of a known star with a sextant, measure its angle above the horizon, note the exact time, and then you consult a set of tables — the Nautical Almanac — that tell you where that star should be at that moment. From the difference between the measured angle and the calculated angle, you derive your position.
The timekeeping is as critical as the star sighting. If your clock is off by a few seconds, your position is off by miles.
More than miles. The Earth rotates about fifteen degrees per hour, which at the equator is roughly a thousand miles. A clock error of just four seconds can put you a nautical mile off. This is why the invention of the marine chronometer in the eighteenth century was such a massive breakthrough — it finally gave ships a clock that stayed accurate through months at sea, through temperature changes, through the pitching and rolling. John Harrison spent decades developing it. Before that, navigators could get latitude easily from the North Star or the noon sun, but longitude was basically guesswork.
The longitude problem. That was the great scientific challenge of the era, right? Nations were offering prizes for solving it.
The British Longitude Act of 1714 offered twenty thousand pounds — enormous money at the time — for a method that could determine longitude within half a degree. Galileo tried to use the moons of Jupiter as a celestial clock. Others tried to use the moon's motion against the background stars. Harrison's clocks won. And the principle hasn't changed. GPS works on the same basic idea — precise time signals from known orbital positions, and your receiver triangulates from the time differences. It's celestial navigation with artificial stars and atomic clocks.
We've gone from Harrison's mechanical clocks to atomic clocks on satellites, and now we're circling back to using actual stars again because we're jamming the satellites.
That's the irony. The most advanced militaries on Earth are investing in celestial navigation training again. The US Naval Academy brought back celestial navigation instruction in 2015 after having dropped it from the curriculum in the late nineties. The Air Force has been retraining navigators on sextant use. The reasoning is exactly what the prompt is getting at — in a conflict where the GPS constellation is degraded or destroyed, you need a backup that can't be jammed, can't be hacked, and doesn't require infrastructure.
You can't jam a star.
You cannot jam a star. And that's the fundamental resilience of celestial navigation. It's completely passive — you're just receiving photons that have been traveling for years or centuries. There's no signal to interfere with, no transmitter to target. The only thing an adversary can do is try to blind your sensors, but even that is hard to do across the entire sky simultaneously.
Let's get practical. If someone is listening to this and wants to actually be able to do this — not just theoretically, but as a real skill — where do they start?
Start with Polaris and your fist. Go outside on a clear night, find the Big Dipper. The two stars at the outer edge of the bowl — Merak and Dubhe — point directly at Polaris. They're called the pointer stars. Follow that line about five times the distance between them, and you'll hit the North Star. Once you've found it, you now know true north. Not magnetic north, which drifts and requires correction — true geographic north. That alone is valuable if you're navigating with a map.
For latitude, you're measuring the angle from horizon to Polaris with your fist.
With your fist. Hold your arm fully extended, make a fist. The width of your fist from bottom to top is about ten degrees. Stack fists from the horizon up to Polaris. If it's four fists, you're at about forty degrees north. That puts you roughly at the latitude of Beijing or Philadelphia or Madrid. If it's two fists, you're around twenty degrees north — roughly Mexico City or Mumbai. It's not precise, but it's enough to tell you which climate zone you're in and whether you need to head north or south to reach your destination.
That's the zero-equipment version. What about the app version?
There are a couple of approaches. The simplest is any good star-gazing app — SkyView, Star Walk Two, Stellarium Mobile. You point your phone at the sky, and it identifies what you're looking at. From that, you immediately know your orientation. But most of these apps are also using your phone's GPS and compass to help identify the stars — they're solving the problem in the other direction, from known position to star identification. For the use case where GPS is unavailable, you'd need an app that works the other way around — from star identification to position.
Which is a harder problem, because the phone has to do pure astrometric solving without any prior location hint.
There's a technique called blind astrometric solving — the software sees an unknown star field and has to match it against a catalog without any hint of where on Earth the camera is or which way it's pointing. This is computationally intensive, but it's a solved problem. There's an open-source tool called Astrometry.net that does exactly this. You upload an image of the night sky, and it identifies the star field and returns the sky coordinates of the image center, the pixel scale, and the orientation. It works with surprisingly bad input — phone cameras, partial cloud cover, even some light pollution.
Does this run on a phone, or does it need a server?
The full Astrometry.net solver traditionally runs on a server because the index files are large — they contain millions of star patterns pre-computed for fast matching. But there have been efforts to make it run locally on mobile devices. The core algorithm is called a geometric hashing approach — it looks at sets of four stars, computes a hash from their relative positions, and matches that hash against a pre-computed index. The index tells it which part of the sky that pattern corresponds to. Once it has a match, it can compute the full astrometric solution.
It's not matching individual stars — it's matching star patterns, the geometry between them. Which means it works even if some stars are obscured or the image is noisy.
And this is the same principle that spacecraft star trackers use. They have a catalog of guide stars and they match patterns. The difference is that spacecraft trackers are purpose-built hardware with known optical properties, operating in perfect darkness with no atmosphere. A phone camera has to deal with atmospheric distortion, light pollution, lens distortion, and the fact that the user is holding it by hand.
What's the current state of mobile astrometric solving? Is there something you can actually install today?
There are a few options, though none are as polished as the mainstream astronomy apps. There's an app called Lost in Space that does blind solving on the phone. There are research projects that have demonstrated real-time solving on smartphone processors. The computational challenge has gotten easier as phone processors have gotten faster — a modern phone can do in seconds what took a desktop minutes a decade ago. But I'll be honest, the user experience is still rough. You need a reasonably dark sky, a steady hand or a tripod, and some patience.
We're at the "it works but it's not seamless" stage. Which means if someone wants this as a genuine backup, they should probably practice with it before they actually need it.
That's the key point. Any navigation skill — whether it's using a map and compass, celestial navigation, or even an app-based backup — degrades under stress if you haven't practiced. The time to figure out that your star-solver app needs a tripod to work reliably is not when you're actually lost and your phone battery is at twelve percent.
The prepper's paradox — the gear works, but only if you've already failed with it in safe conditions.
There's another wrinkle that I think gets overlooked in these discussions. Celestial navigation only works when you can see the sky. If you're in a forest with heavy canopy, or in a city with tall buildings, or it's overcast — which it is, statistically, about sixty to seventy percent of the time in many populated regions — you're out of luck. The stars are a fallback, but they're not a universal fallback.
The real resilience strategy is layered. GPS is layer one. When GPS is jammed or down, you might fall back to inertial navigation — which is what missiles and some high-end military vehicles use, dead reckoning from a known starting point using accelerometers and gyroscopes. When that drifts too far, you might use celestial as a correction. And when the sky is obscured, you're down to map and compass and terrain association.
That hierarchy is exactly how military navigation is structured. An aircraft might use GPS as its primary, with an inertial navigation system as backup, and celestial as a periodic correction to reset the drift on the inertial system. The SR-seventy-one Blackbird had an astro-inertial navigation system that would automatically track stars through a quartz window in the fuselage to correct its position while flying at Mach three. That was in the nineteen sixties.
The SR-seventy-one was using star trackers at Mach three in the sixties, and I'm here trying to get my phone to identify the Big Dipper through light pollution. Technology is not a straight line.
It really isn't. And the Blackbird's system was called the NAS-fourteen — the Nortronics Astro-Inertial Navigation System. It could track up to sixty-one stars in broad daylight, because at eighty thousand feet the sky is dark enough during the day for bright stars to be visible. The system would lock onto a star, track it, and use the measured angle to update the inertial platform. Position accuracy was reportedly within a few hundred meters after hours of flight. That's better than a lot of consumer GPS receivers from the early two thousands.
What you're telling me is that the backup system for a Cold War spy plane is more capable than my phone's fallback navigation. And nobody's really commercialized that for civilian use.
The market hasn't demanded it. For most people, most of the time, GPS works and it's free. The scenarios where GPS is unavailable and you urgently need navigation and you can see the stars and you don't have any other backup — that's a narrow overlap. But it's getting wider. The jamming incidents in the Baltic, the Black Sea, the Middle East — these aren't rare anomalies anymore. Ships in the Mediterranean have reported GPS disruptions. Israel has dealt with GPS spoofing that makes receivers show incorrect positions, sometimes showing planes on the ground as being at the Beirut airport.
Spoofing is even scarier than jamming. Jamming you know about — your receiver says "no signal" and you switch to backup. Spoofing gives you a wrong position that looks correct. You might not know you're being spoofed until you're off course.
That's where celestial becomes not just a backup but a truth source. If your GPS says you're at a certain position but your star sighting says you're thirty miles north, you know the GPS is lying. A star can't be spoofed. The photons hitting your sensor came from that star — there's no man-in-the-middle attack on starlight.
Unless someone puts a drone with a bright LED in the sky and flies it in a pattern that mimics a constellation. Which, I admit, is getting into Bond villain territory.
That would be an incredibly sophisticated attack, and I think if your adversary is flying constellation-mimicking drone swarms to fool your star tracker, you have bigger problems.
So let's talk about the Southern Hemisphere, because the prompt mentioned Africa — the Congo specifically. If you're south of the equator, Polaris is below the horizon. What's the fallback there?
The Southern Cross is the primary reference, but as I said, it's not as straightforward. You find the Southern Cross — Crux — which is a compact, bright constellation that actually looks like a cross. You extend the long axis about four and a half times, and that gives you the south celestial pole. From that point, drop straight down to the horizon and you have south. For latitude, you're measuring the altitude of the south celestial pole above the horizon, same principle as Polaris in the north. But there's no bright star at that point, so you're estimating.
There are tricks for finding south without the cross being visible?
One of the most reliable is using Orion, which is visible from most of the inhabited world. Orion's belt is three bright stars in a line. The sword hangs below it. Orion rises roughly in the east and sets roughly in the west everywhere on Earth because it sits near the celestial equator. If you can find Orion, you can get a rough east-west line from its rising and setting. The belt stars also point toward Sirius, the brightest star in the sky, and Aldebaran, a bright reddish star in Taurus. But none of these give you a fixed pole the way Polaris does.
The northern hemisphere really did get the better deal, navigationally speaking. One bright star, fixed in place, visible most of the year from most inhabited latitudes.
That's probably part of why northern hemisphere civilizations developed certain navigational capacities earlier. The Polynesians, of course, were the great exception — they navigated vast stretches of the Pacific using star paths, wave patterns, bird behavior, and a deep knowledge of multiple rising and setting stars. They didn't have a pole star to rely on, so they memorized the rising and setting positions of dozens of stars and used them as directional markers. It's arguably a more sophisticated system than just measuring Polaris.
They were navigating by the whole sky, not just one star. Which circles back to the app question — the star-field identification approach is essentially the Polynesian method automated. Pattern-match the whole visible field, not just one reference point.
And that's why it works anywhere on Earth, any time of night, any season — as long as you can see enough stars to form a recognizable pattern. The software doesn't care whether it's looking at Polaris or the Southern Cross or some random patch of the Milky Way. It just needs enough stars to get a unique match against the catalog.
What's the minimum? How many stars do you need to be visible for a blind solve to work?
net can often solve with as few as four or five visible stars if they're bright enough and the pattern is distinctive. But in practice, you want more like ten to fifteen for a reliable solve, especially with a phone camera where the star positions might be slightly distorted by lens effects. The more stars in the frame, the faster and more confident the match.
That's where light pollution becomes the real enemy. Not clouds, not jamming — just the fact that most people live under skies where they can see maybe a dozen stars total.
The Bortle scale measures sky darkness from one to nine. A Bortle one sky — truly dark, no light pollution — you can see thousands of stars, the Milky Way casts shadows, it's spectacular. Most suburban skies are Bortle five or six — you can see maybe a few hundred stars. Urban skies are Bortle eight or nine — you're down to the brightest planets and a handful of first-magnitude stars. At that level, automated solving gets very difficult because there aren't enough stars in the frame to form unique patterns.
The practical utility of a star-solver app is inversely proportional to how close you are to a city. Which is exactly where you're most likely to be when GPS goes down in a non-conflict scenario — some kind of infrastructure failure in an urban area.
That's the cruel irony. The places where you're most likely to need a backup are the places where the sky is least readable. In a true wilderness scenario, you probably have dark skies and can navigate by stars easily — but you're also less dependent on GPS because you're navigating by terrain and map. In the city, GPS fails and you look up and see... the glow of a Walmart parking lot.
The Walmart parking lot, nature's sextant.
The practical takeaway is that celestial backup is a layer, not a panacea. It works brilliantly in some contexts and not at all in others. The key is knowing which is which before you need it.
Let's talk about the AI angle, because the prompt mentioned it and I think there's something interesting here. We've been talking about pattern-matching against a star catalog, which is a classic algorithmic approach. Could you train a neural network to do this better — to solve from noisier data, from fewer stars, from partial sky obscured by clouds or buildings?
There's been research on using convolutional neural networks for star identification, and they show promise in handling degraded images better than the geometric hashing approach. A neural network can learn to recognize star patterns even when individual stars are missing or displaced — it's learning the statistical regularities of the entire sky, not just matching discrete patterns.
It's the difference between fingerprint matching and face recognition. Fingerprint matching looks for specific points that correspond. Face recognition learns the overall statistical distribution of faces and can recognize you even with half your face covered.
The neural network approach has another advantage — it can potentially incorporate other cues. If the camera also captures a bit of horizon, or some terrain silhouette, the network could learn to use those as additional constraints on position and orientation. It's multimodal — stars plus horizon plus maybe even the magnetic field reading from the phone's compass, all fused into one position estimate.
The training data is basically infinite. You can simulate the night sky from any position on Earth, any time, any weather conditions, and generate millions of labeled training images.
You can train on synthetic data that covers every possible degradation — light pollution gradients, partial cloud cover, atmospheric refraction, lens distortion, sensor noise. And then fine-tune on real images. This is a near-ideal problem for deep learning.
Where is this actually being deployed? Is anyone shipping a neural star solver in a product?
Not in a consumer product that I'm aware of. The military is certainly working on it — DARPA has had programs on alternative navigation, and there are defense contractors developing AI-enhanced celestial navigation for platforms that might operate in GPS-denied environments. But on the civilian side, the existing algorithmic solvers work well enough that there hasn't been a strong push to replace them with neural approaches.
Yet being the operative word. As the chips in phones get dedicated neural processing units, running a star-solver network locally becomes trivial. You could have it running continuously in the background, fusing with GPS and inertial data, and when GPS drops out, it seamlessly takes over.
That's the vision for what's called assured position, navigation, and timing — assured PNT. The idea is that you don't have a single navigation source that you fall back to when the primary fails. You have multiple sources running simultaneously — GPS, inertial, celestial, magnetic, terrain-referenced, signals-of-opportunity like cell towers and Wi-Fi — and they're all fused into one continuous position estimate. If any one source degrades or drops out, the system compensates automatically. You never even notice.
The future of navigation isn't picking the right backup. It's having so many overlapping systems that no single failure matters.
Celestial is uniquely valuable in that mix because it's the only one that's truly unjammable, globally available, and doesn't require any infrastructure. It's the ultimate ground truth. Everything else can be spoofed or denied, but the stars just keep shining in exactly the positions we've cataloged for centuries.
There's something philosophically satisfying about that. We've built this incredible technological edifice — satellites and atomic clocks and global networks — and the ultimate fallback is the same thing sailors used five hundred years ago.
The same thing birds use. And sea turtles. And dung beetles, which actually navigate using the Milky Way. A dung beetle can orient itself by the galactic plane.
The dung beetle has a backup navigation system that doesn't require GPS.
It doesn't even need a clear view of individual stars. Just the broad glow of the Milky Way is enough for it to maintain a straight line while rolling its dung ball. That's a level of robustness we haven't achieved with our phones yet.
The dung beetle is ahead of Silicon Valley on assured PNT. I'm going to sit with that.
The takeaway for anyone listening is this: learn to find Polaris. It takes five minutes and it works forever. If you want a technological backup, download a star-gazing app and practice using it to determine your orientation without relying on the phone's GPS. Understand its limitations — light pollution, cloud cover, canopy — but also understand its unique strength, which is that it cannot be turned off, jammed, spoofed, or discontinued. The stars are the one navigation system with a hundred percent uptime guarantee for the next few billion years.
If you want to go deeper, get a planisphere — a rotating star chart that shows you what's visible on any given night. They cost about ten dollars and they never run out of battery. Between a planisphere and your fist, you've got a navigation system that works anywhere north of the equator, any clear night, with zero electronics.
The planisphere is one of the most underrated tools in existence. It's just two disks of cardboard with a rivet in the middle, and it solves the entire celestial sphere for any time and date. You rotate the disk to match the date with the time, and the window shows you exactly what stars are visible and where they are in the sky. It's a mechanical star computer.
It was invented when?
The modern planisphere dates to the early twentieth century, but the principle goes back to the astrolabe, which is over two thousand years old. The astrolabe was essentially a planisphere with sighting vanes for measuring star altitudes. It was the smartphone of the ancient world — it told time, it gave directions, it calculated star positions, it could even do trigonometry.
The astrolabe was the smartphone of the ancient world. I love that. And now we're trying to make our smartphones into astrolabes.
And the circle is about twenty-six thousand years, if we're talking about precession.
Okay, so to wrap this into something actionable: if someone wants to actually be able to navigate by stars, the progression is find Polaris with the pointer stars, learn to estimate latitude with your fist, get a planisphere to understand what's visible when, and if you want a tech backup, find an app that does star-field identification and practice with it in conditions where you don't need it. So that when you do, it's second nature.
One more thing I'd add: learn to take a star sighting with your phone camera. Even if you don't have a specialized app, a photograph of the night sky contains enough information for an astrometric solve later. If you're genuinely lost and you have a phone with a camera but no signal and no specialized app, take the best photograph of the sky you can — steady the phone on something, use night mode if available — and that image can be solved when you have connectivity or access to software. It's a data capture you can act on later.
That's a practical tip. Capture the data now, process it when you can. The stars aren't going anywhere — the same stars will be in essentially the same positions tomorrow night. The photograph is a timestamped record of your celestial reference frame.
The stars move about one degree per day due to Earth's orbit around the sun — that's why the night sky shifts with the seasons — but one degree is easy to correct for if you know the date. The stars themselves are effectively fixed on human timescales.
The final answer to the prompt is: yes, you can navigate by stars as a practical fallback. Yes, there are apps that do star identification. The gap is in blind position solving on a phone without prior location, which works but isn't seamless yet. And the most important part is practicing before you need it, because the stars won't fail you, but your ability to read them might.
That's the thing about celestial navigation. It's a skill, not just knowledge. You can understand the theory perfectly and still be unable to find Polaris in a real sky with real light pollution and real trees blocking half your view. The theory is simple. The practice takes repetition.
Like anything worth knowing.
Like anything worth knowing.
Now: Hilbert's daily fun fact.
Hilbert: The Turkmenistan government declared the traditional Turkmen chotki abacus extinct in the nineteen nineties, only for an ethnographer to discover a family in the Karakum Desert still producing them in two thousand seventeen, using a seventeen-bead variant that had no known historical record.
Seventeen-bead variant. Of course there is.
This has been My Weird Prompts. Thanks to our producer, Hilbert Flumingtop. If you enjoyed this, leave us a review wherever you listen — it helps. We're back next week.
