The Core That Changed Its Mind

OlaPicture this: you're falling, not through air, through rock, and somewhere around three thousand kilometers down, the rock just stops being rock.

AmaraOoh, I love when you go full ghost story on me this early.

OlaIt's liquid iron and nickel down there, Amara, churning like a pot that never stops boiling, and that churn is the reason your compass points north.

AmaraI'm Amara.

OlaAnd I'm Ola, and this is the show where we drag you into the planet's basement and don't apologize for it.

AmaraOkay, so the field just comes from stirring metal. That's it?

OlaThat's the engine, yeah. And every so often, the whole thing loses its nerve. The field fades, the poles wander for thousands of years, then settle down on the opposite side of the planet.

AmaraWait, thousands of years? Not like a bad Tuesday.

OlaThousands, and that's the version we've trusted for decades. Slow, orderly, a flip you could set a very patient clock to.

AmaraI sense a but coming.

OlaThere's always a but because that's the tidy version, the one science writers cover. Poles wander, they settle, everybody moves on with their lives. What that version doesn't tell you is whether the record we're trusting to count these flips is actually complete.

AmaraWait, complete like are you telling me there are missing chapters in this book?

OlaThat's the question the whole episode hangs on. If the core can behave for millions of years and then throw something nobody expected, the rocks and sediments we read this history from might be leaving entire events out.

AmaraWhole reversals just erased? Possibly erased or possibly never recorded in the first place. And here's the part that makes it personal. The shield keeping cosmic radiation off your skin right now runs on this exact same unpredictable machine. Okay, that's actually unsettling in a fun way. I'm ready. Let's go down there.

OlaAfter you. Okay, so if the field usually takes its sweet time settling down after a flip, why do you keep making that face like something's wrong with the story?

AmaraOh, because I don't trust an engine that runs quietly for so long and then just doesn't.

OlaOla this trustful of a slow bureaucratic process? Shocking.

AmaraOh, fair, but think about it. We've built this whole picture of the core as this dependable thousand-year metronome underground.

OlaRight. The poles wander, they settle, everybody moves on with their lives.

AmaraAnd that picture comes from real work. There's a piece out of Utah written up on Physorg by a science writer named Maffly laying out exactly that sequence.

OlaThe fade, the wobble, the settle.

AmaraExactly, which is fine as far as it goes.

OlaBut.

AmaraBut?

OlaA clock that's reliable for millions of years and then throws one wild outlier isn't really a clock anymore.

AmaraOoh, okay. I like where this is going.

OlaIt's a system with a memory. It remembers stress, it remembers weird flow, and every so often it just refuses to behave.

AmaraSo we're not debunking the textbook picture. We're saying it's the average, not the rule.

OlaThat's the move. Average behavior, individual chaos.

AmaraOoh, wait. Is this you setting me up for a number?

OlaI might be.

AmaraOla.

OlaWe said thousands of years to swap poles. That's the going rate. Everyone agrees.

AmaraAnd?

OlaAnd a twenty twenty-six study found one transition that didn't take thousands of years.

AmaraHow long?

OlaTry Seven times that.

AmaraSeven ti-- hold on, hold on. So if the Normal move is a few Thousand years-

OlaThis one just Kept going.

AmaraShut up. Seriously?

OlaSeriously.

AmaraOkay, That's Not a flip. That's a full Reversal.

OlaA very long, very messy Reversal, yeah.

AmaraSo the Core didn't just change its mind, it Took forever to actually commit, which is the Whole problem with calling any of this predictable. I'm into it. I'm fully into it now.

OlaGood, Because that number is the Whole reason we're doing this episode.

AmaraWait, Wait, before you tell me the number again, walk me through why it even matters that it's long instead of just weird.

OlaBecause duration tells you something about the mechanism. A fast flip and a slow flip might be completely different physical events happening down there.

AmaraDifferent events, same label.

OlaRight, and if we've been lumping them together this whole time-

AmaraThen the record's lying to us a little

OlaOr at least oversimplifying.

AmaraOkay, I need the actual number. No more teasing.

OlaYou're going to want to sit down for this one.

AmaraI'm sitting.

OlaAll right. So get this. A researcher named Yamamoto and his team went digging through deep sea sediment cores from the Eocene, and they found two transitions that just would not finish. Wait, how long are we talking?

AmaraOne took eighteen thousand years, the other seventy thousand years.

OlaSeventy thousand years. That's not a pole wandering. That's a pole taking a sabbatical.

AmaraRight? And remember, the younger reversals we know about wrap up in under ten thousand years. This one ran seven times past that.

OlaSo the core just forgot what it was doing halfway through?

AmaraKind of, yeah, and here's the part that got me. The same paper says their models predict some transitions could stretch to a hundred and thirty thousand years.

OlaA hundred and thirty thousand.

AmaraYeah.

OlaA hundred and thirty thousand years is longer than modern humans have existed as a species.

AmaraOh, I hadn't even done that math.

OlaSo what happens in the middle of one of these failed transitions? Does the field just sit there weak and directionless for tens of thousands of years?

AmaraThis is where it gets good. The authors describe something they call multiple failed reversals. The field starts to flip, loses momentum, snaps partway back, tries again.

Speaker 3Multiple failed reversals. So it's not one clean flip, it's the core trying, choking, trying again.

AmaraExactly. Picture flipping a light switch, and it just buzzes in the middle, stuck between on and off.

Speaker 3That's unsettling actually because we've been talking about the geo dynamo like it's a machine with a schedule.

AmaraAnd this paper's saying the schedule includes false starts.

Speaker 3Okay, but I wanna push on this a little. If the field is weak and stuttering for seventy thousand years, is that a slow news week for the planet, or is that actually dangerous?

AmaraThat's the argument scientists are having right now because a weaker field means more cosmic radiation getting through.

Speaker 3For seventy thousand years?

AmaraFor seventy thousand years potentially, yeah. Not a decade, not a century, an entire epoch's worth of extra exposure.

Speaker 3Which changes how I think about the whole reversal record. If some of these things take that long and stutter that much-

AmaraHow would we even know we found all of them?

Speaker 3That's exactly where my head went. If a reversal can fail, restart, fail again, some of that could look like noise in the rock instead of a real event.

AmaraOr it could just not show up at all if the rock from that stretch got weathered away or never recorded properly.

Speaker 3So the rock record isn't just slow to reveal these things, it might be leaving whole ones out.

AmaraWhich means the number of reversals we think happened could be wrong. Not slightly wrong, maybe missing entire events wrong.

Speaker 3Then the question isn't just how long a reversal takes anymore-

AmaraIt's how many we've never even clocked.

Speaker 3Okay, so the record has gaps. Question is, are they actual holes or did nobody go looking in the right spots?

AmaraSomebody went looking. A 2026 paper, lead author named Yoshi Mora, ran the whole reversal record through something called adaptive kernel density estimation.

Speaker 3Which means what in human words?

AmaraIt smooths out the timeline of reversals and looks for stretches where the rate drops off a cliff. If reversals usually cluster and then suddenly go quiet for way too long, that quiet stretch is suspicious.

Speaker 3And they found some?

AmaraFour of them, all after the Cretaceous Normal Superchron, that long dead calm we mentioned earlier, spaced something like twelve to fifteen and a half million years apart.

Speaker 3So four separate windows in Earth's past where the field might have flipped and nobody wrote it down.

AmaraThat's the claim. PhysOrg covered it under a headline that basically says Earth's past may be incomplete, and it's built on exactly that model. The National Institute of Polar Research Group ran the statistics.

Speaker 3Okay, but a statistical trough is still just a gap on a graph. It's not a rock. Show me a rock.

AmaraOh, funny you say that.

Speaker 3Oh, don't do the pause. I'm doing the pause. Ethiopia, the Simien Mountains, there's a flood basalt sequence there called Lima-Limo Lima-Limo, about two kilometers of stacked lava.

AmaraTwo kilometers of rock, just lava layer after layer?

Speaker 3Exactly. And a researcher named Kidane went through it and found four reversals nobody had ever recorded before sitting right there in the magnetic signature of the rock.

AmaraWait, four? Same number as the statistical study?

Speaker 3Same number, which sounds huge.

AmaraIt is huge.

Speaker 3It's suggestive. It is not the same four.

AmaraWhat do you mean not the same four?

Speaker 3Nobody's shown that the Ethiopian reversals sit inside Yoshimura's predicted windows. Different methods, different datasets, different rocks. They just both point at the idea that we're missing events.

AmaraSo it's less case closed and more two witnesses independently say the record is lying.

Speaker 3Right, and they haven't compared notes.

AmaraOkay, but as a journalist, tell me, does that matter? If two totally separate approaches both land on there's more reversals than we've counted, isn't that basically the same finding twice?

Speaker 3It's not nothing, but I keep my notebook out for a reason. A coincidence in the count isn't a match in the timeline. If it turned out the Lima-Limo reversals actually fell in one of those twelve to fifteen million year troughs, that's a real correlation. Right now, it's two studies shouting the same general suspicion from different rooms.

AmaraFine, but four unrecorded flips baked into two kilometers of Ethiopian lava is not a small thing to shrug off.

Speaker 3I'm not shrugging it off. I'm saying we don't yet know if those are catastrophic missed events or just short, quiet flips that didn't leave much of a mark.

AmaraSo if I'm keeping score, we've got one statistical model pointing at four windows and one mountain in Ethiopia holding four actual reversals, and you're telling me I'm not allowed to just add them together?

Speaker 3You're not, no. Adding them assumes they're the same four, and nobody's tested that. For all we know, Lima-Limo's four sit entirely outside Yoshimura's troughs, and the real total is eight missing reversals, not four.

AmaraOr the totals overlap and the true number's smaller than either study alone.

Speaker 3Right. That's the annoying part of good science. It hands you more questions before it hands you a clean answer.

AmaraEither way, the picture from the last twenty minutes is reversals can take way longer than we thought, and there might be whole reversals we've never even logged.

Speaker 3Which means the past record isn't the tidy stripe pattern people picture. It's got missing pages.

AmaraAnd normally I'd say, "Fine, that's ancient history. Who cares?" Except-

Speaker 3Except?

AmaraExcept the core isn't exactly sitting still right now either. There's new satellite data on what it's doing under the Pacific this year.

Speaker 3Wait, current data, not fossil data?

AmaraCurrent, like happening while we record this episode.

Speaker 3Okay, now you have my attention.

AmaraGood, because you're going to want to sit down for this one.

Speaker 4Okay, so get this. While we've been down in the past, the core just did something in the present that nobody predicted. Wait, present as in now?

AmaraPresent as in 2026 satellite data. Earth's molten core beneath the Pacific just flipped which way it's flowing.

Speaker 4Flipped how? Like a full reversal pole to pole?

AmaraNo, no. This is the flow inside the outer core itself, not the whole magnetic field. But scientists are saying the surge already looks like it's weakening.

Speaker 4So it spiked, and it's already fading?

AmaraThat's the read right now.

Speaker 4Okay, but unexpectedly reversed is doing a lot of work in that sentence. What was it doing before?

AmaraThis is where the ESA Swarm satellites come in. For decades, everyone assumed the flow under there was dominated by stable westward motion, like a conveyor belt going one direction.

Speaker 4Sure, the textbook picture.

AmaraExcept starting in 2010, it just switched. Strong eastward flow, same spot beneath the Pacific.

Speaker 42010? That's not ancient history. That's during my lifetime.

AmaraRight, and ESA's own mission scientist, Elisabetta Iorfida, said it herself. Regional changes can emerge rapidly, and she means within a single decade.

Speaker 4A decade for something we thought moved on geological scales?

AmaraA whole current under an ocean basin turning around inside 10 years.

Speaker 4What made it happen?

AmaraNobody actually knows, but there's a theory, and it's a good one. The lead author on the study, a researcher named Madsen, thinks the timing isn't a coincidence.

Speaker 4Coincidence with what?

AmaraHe's proposed the 2010 flip is, quote, contemporary with a change in behavior in the inner core as inferred from geodesy and seismology.

Speaker 4So the inner core did something around the same time, and he's saying maybe the outer core felt it.

AmaraThat's the hypothesis. Not proven, a hypothesis.

Speaker 4I want to sit with that for a second because if the inner core can nudge the outer core into a regional flip inside 10 years, that's a different core than the one we opened this episode with.

AmaraThe slow, patient one.

Speaker 4Yeah, that one doesn't exist anymore, or it never did, and we just hadn't looked closely enough.

AmaraWhich one do you believe?

Speaker 4Honestly, I think we've been reading a coarse instrument and calling the picture stable. Zoom in, and it's twitchy.

AmaraTwitchy is generous. This is the mechanism that makes our whole magnetic shield. If a chunk of it can lurch sideways in a decade-

Speaker 4Then the assumption everyone builds on that the field itself is dependable day to day needs a second look.

AmaraSo we've got a hidden past apparently missing whole events.

Speaker 4And now a present that's moving faster than the models expected.

AmaraSo put those two together, and there's really one question left.

Speaker 4Whether any of that means the shield over our heads right now is something we can actually count on. So put both threads on the table for me, Amara. One reversal that stretched to 70,000 years and a model plus a mountain in Ethiopia both hinting at reversals nobody wrote down.

AmaraRight, and separately, the core just changed its mind under the Pacific for no reason anyone can explain.

Speaker 4Two different centuries of evidence, same conclusion.

AmaraWhich is?

Speaker 4The field doesn't just flip and settle. It negotiates.

AmaraIt negotiates. Okay, I'll take that.

Speaker 4Yamamoto's group didn't just find the 70,000-year transition. They ran the physics forward and got models predicting transitions up to 130,000 years.

Amara130,000?

Speaker 4That's not a typo. That's the geodynamo saying it can take longer than the entire span of modern humans just to make up its mind.

AmaraAnd meanwhile, Yoshimura's statistics say there could be four whole missing windows sitting in gaps we assumed were boring.

Speaker 4Right. Calm stretches that are calm because we can't see into them, not because nothing happened.

AmaraSo the honest version isn't the record is wrong. It's the record might be selective.

Speaker 4I'd put it stronger than selective. If transitions can run 15 times longer than the textbook number, and some of them apparently fail and restart before completing, the count of reversals in any given era could be off in either direction.

AmaraWait, either direction? I thought this was all pointing toward more reversals hiding in there.

Speaker 4It could also mean fewer clean ones and more attempts. A failed reversal doesn't flip the poles. It just costs time and maybe weakens the field for a while.

AmaraWhich is the part that actually matters for us, not the history nerd part.

Speaker 4Sure, but I'm not going to pretend we know that a weak stretch back then did anything to life on the surface. Nobody's shown that link.

AmaraFair. So no dead dinosaurs from a stuttering field.

Speaker 4None that anyone's demonstrated. What we can say is the machine has a track record of getting stuck, and it's stuck adjacent right now with that eastward surge under the Pacific.

AmaraA surge nobody's explained.

Speaker 4A surge nobody's explained, tied by one researcher to something happening at the inner core, and even that's a guess he's still testing.

AmaraSo if I ask you straight, is the shield reliable? What do you actually say?

Speaker 4I say it's reliable the way a decades-long average is reliable. Any single decade, any single century, it can do something nobody predicted.

AmaraThat's not exactly comforting.

OlaIt's not supposed to be comforting. It's supposed to be accurate. We genuinely don't know why the flow flipped, and we genuinely don't know what the missing reversals looked like when they happened.

AmaraTwo open questions, no tidy bow.

OlaNone, which is a strange place to land an episode about the planet's magnet.

AmaraStrange, but it's where the data actually put us. Okay, let's wrap this thing up.

OlaSure.

AmaraSo where does that leave us? Honestly, nowhere neat. Three papers, three separate signs the core doesn't run on a schedule, and not one of them tells us when the next flip starts or how long it'll take once it does.

OlaThat's the honest answer, though.

AmaraIt is. The magnetic shield up top isn't broken, but the machine generating it downstairs has a memory that skips and stutters in ways we're only just measuring. That's the whole picture, uncertainty we can finally point at with numbers instead of just assuming it's stable.

OlaYou know, we started this episode by heading down into the core together.

AmaraWe did.

OlaAnd now we're climbing back out with less certainty than when we went in.

AmaraThat's basically fieldwork.

OlaIt's like pickleball, actually. You think you understand the bounce, and then the court surface changes under you.

AmaraDeadpan. Only the court is 3,000 kilometers down and made of molten iron.

OlaSure. Minor detail.

AmaraIf this cracked something open for you, subscribe, leave us a review. It genuinely helps other people find the show.

OlaAnd if you've got an epic you want us to dig into or you think we got something wrong here, tell us.

AmaraEmail hello@heymato.com. Favorite era, a theory you think we butchered, whatever.

OlaWe read every one.

AmaraWe do, even the ones that just say we're wrong about the pickleball analogies.

OlaThose are correct, for the record.

AmaraWe'll see you next time, hopefully with a core that's decided to behave.