Brian Reed: Hey. Good to be here. I have been sitting with this paper since it dropped this morning and I — honestly, I'm still kind of turning it over.
Brian Reed: Science Bulletin, today — September 30th. It's about an X-ray flash called EP250704a. Or GRB 250704B, depending on who's cataloguing it. Three satellites caught it: Einstein Probe, SVOM, and Insight-HXMT. And the flash lasted about ten minutes.
Brian Reed: Ten minutes. For context — and I had to look this up to make sure I wasn't misreading it — the standard short gamma-ray burst from a neutron star merger runs under two seconds.
Eliza Ward: Wait, hold on. The detection date on this — that's July 4, 2025. That's over a year ago.
Brian Reed: Right, so — yeah, that's the thing I keep snagging on. The event happened fifteen months ago. The paper lands today. And the headline reads like breaking news, but there's a year and change of analysis buried inside that gap.
Eliza Ward: And the conclusion they're landing on after all that time is that this may represent — let me get this straight — the formation of a magnetar from two colliding neutron stars.
Brian Reed: Right — but the part that doesn't fit is that 'magnetar formation' is the conclusion. What actually lands in the data is a ten-minute X-ray flash. Everything after that is inference.
Eliza Ward: Which is worth slowing down on, because people are missing this. Neutron star mergers aren't mysterious. We've seen the signature. It's under two seconds. Short gamma-ray burst, done. EP250704a ran for ten minutes. That's not a little long. That's 300 times longer.
Brian Reed: Three hundred times.
Eliza Ward: Think of it like — you're expecting a camera flash. One burst. Instead you get a floodlight that someone left on for ten full minutes. Same source category, completely different observational profile. And if you were sorting old data looking for camera flashes, you would have filed the floodlight under 'unknown' and moved on.
Brian Reed: That's — yeah, that's the part that actually unnerves me a little. How many of those are sitting in archival data right now?
Eliza Ward: That's a real open question and the paper doesn't close it. They're not — I mean, they're not claiming they've gone back and reindexed anything. The implication is just sitting there.
Brian Reed: So the headline is 'we saw a magnetar born' — but the actual discovery might be that we've been systematically misfiling a whole class of merger events because they ran too long to match the filter.
Eliza Ward: That's the sharper version of what's new here, yeah. The magnetar is the conclusion. The broken filter is the actual finding.
Brian Reed: And we don't know — wait, has anyone put a number on how many extended X-ray flashes might qualify? Or is that still speculative?
Eliza Ward: Nobody's put a number on it — and that's actually where I want to push, because the circulating take right now is worse than 'we don't know.' The take is 'astronomers confirm magnetar birth.' That's the headline. And that's just — that's not what the Science Bulletin paper says.
Brian Reed: What does it actually say?
Eliza Ward: 'Strong evidence.' That's the phrase. Not confirmed. Not directly observed. They analyzed the afterglow and remnant data — that's the VLT and the VLA doing follow-up after the Einstein Probe alert — and they drew an inference. A well-supported one, but an inference.
Brian Reed: So the part that's missing is — wait, do they say anywhere that they actually imaged the magnetar remnant?
Eliza Ward: No. Explicitly no. They don't claim an unambiguous direct image. That's the hedge sitting right there in the abstract. A researcher reading the actual paper this morning sees it immediately. The headline writers just... dropped it.
Brian Reed: Okay, so — I mean, does the gravitational-wave picture help here at all? Because if LIGO had caught a merger signal at the same moment, that's independent confirmation of the neutron star collision at least.
Eliza Ward: That's the gap. LIGO-Virgo-KAGRA's GWTC-4.0 catalog only runs through January 2024. EP250704a happened July 4, 2025. There's no gravitational-wave counterpart reported — not because they looked and found nothing, but because the coverage window doesn't reach it.
Brian Reed: So does that make the electromagnetic case weaker? Like, without the GW confirmation, are we actually on thinner ice than the headline suggests?
Eliza Ward: Not weaker — incomplete. Those are different things. The VLT and VLA data are real. The afterglow analysis is real. It's just that electromagnetic evidence alone doesn't cross from 'strong inference' to 'confirmed.' And — actually, there's a second 2026 magnetar report, a supernova called SN 2024afav, completely different formation pathway, and what those two together mean for how confidently you can assign any single event to one mechanism is the part we need to get into.
Brian Reed: Wait — two different formation pathways, same year? Because SN 2024afav is a supernova. That's not two neutron stars colliding. That's a massive star's core collapsing.
Eliza Ward: Completely different mechanism. And — this is the part that matters — the observational fingerprint is different too. EP250704a's signal is an extended X-ray afterglow. SN 2024afav was detected via what they're calling a relativistic chirp. Two events, two pathways, two distinct signatures.
Brian Reed: So when someone tries to build a unified picture — 'magnetars form through at least two channels' — is that one finding, or is it just... two separate papers sitting next to each other?
Eliza Ward: The latter. Each study stands on its own evidence. The two-pathway picture is — I mean, it's emerging, it's interpretive, it's not a single confirmed finding that both teams signed off on together. Someone is going to write that synthesis paper. It hasn't landed yet.
Brian Reed: That's — yeah, that's actually the part that changes how I'd read the archival question. Because if there are two signatures to look for now — extended X-ray flashes and relativistic chirps — the census problem just got bigger.
Eliza Ward: Right. And that's real science even if it feels like bookkeeping. Imagine a grad student right now pulling Insight-HXMT archival files from 2022, looking for X-ray events that ran longer than thirty seconds and got filed under 'unclassified.' That work — that's what actually expands the known population of these events.
Brian Reed: And before this year, there was almost no direct observational evidence for magnetar birth through either channel. So we're not refining a crowded field — we're basically at two data points.
Eliza Ward: Which is exactly why you can't — wait, actually this is the confidence problem in miniature. Two events is enough to say both pathways exist. It is not enough to say how common either one is, or whether the chirp and the X-ray afterglow exhaust the possible signatures.
Brian Reed: So the concrete thing to watch is the archival reanalysis — whether extended X-ray flashes already sitting in old data get reclassified as merger candidates. That's the number that would actually tell us if the EP250704a picture holds.
Eliza Ward: And that's — I mean, that's where I actually land on this whole thing. The archival reanalysis is the empirical work. But the thing that would actually *settle* EP250704a specifically is a future neutron star merger where LIGO-Virgo-KAGRA catches the gravitational wave signal *and* an X-ray observatory catches a ten-minute flash at the same moment. Both, simultaneously. Because if that happens and you see the extended emission again, the EP250704a interpretation gets a lot stronger. And if it doesn't — if the next merger is two seconds and done — then something about *this* event is still unexplained.
Brian Reed: Which means we're basically waiting for the next merger to either validate the reading or force a rethink. And we have no idea when that is.
Eliza Ward: No timeline. That's — yeah, that's the honest answer. The Science Bulletin paper says 'strong evidence.' Not confirmed. And that restraint isn't a hedge, it's — wait, actually I think that *is* the finding. That's where the science honestly sits. Strong evidence, waiting for a second event to either confirm or complicate it.
Brian Reed: The restraint is the finding. I think that's right. Fifteen months of analysis before they'd say even that much—that stuck with me.
Eliza Ward: That part I respect. Go.
