Did a Small-Brained Cousin Bury Its Dead? Inside the Homo naledi Fight

Imagine you are handed a skeleton — not a museum cast, a real one, still half in the dirt — and asked one question: did somebody put this here on purpose? You cannot ask anyone. You cannot watch it happen. All you have is the position of the bones, the layers of soil around them, and whatever chemistry you can coax out of a tooth. There is no test in any lab that reads “buried” or “not buried” off a screen. There is only an argument, built one inference at a time, that one story fits the dirt better than its rivals.

That argument is now playing out, in public, over a hominin that should not have been capable of making a decision this complicated at all. Homo nalediA hominin species described from more than 1,550 fossils (minimum 15 individuals) recovered from the Dinaledi Chamber of South Africa's Rising Star cave system in November 2013 and March 2014. Berger et al., eLife 2015 (article 09560). had a skull that held between 465 and 560 cubic centimetersEndocranial volume estimates from the original species description: ~560 cc for larger (presumed male) crania, ~465 cc for smaller (presumed female) crania — within the australopith range, roughly a third the volume of a modern human brain. Berger et al., eLife 2015. of brain — squarely in the range of an australopith, nowhere near the roughly 1,300–1,400 cc typical of Homo sapiens. And yet, according to one team of paleoanthropologists, it dug holes for its dead, placed the bodies inside with a discernible posture, and back-filled the pits — a behavior otherwise attested only in Homo sapiens and Neanderthals, and only starting around 100,000 years ago, at sites like Qafzeh and Skhul in Israel. According to a second team of equally credentialed scientists, re-examining the exact same sediment, there is no scientific evidence for anything of the kind.

Both teams published in serious, peer-reviewed venues. Neither has changed its mind. This primer is not going to adjudicate who is right — nobody outside that cave can, yet. What it will do is walk through how an argument like this is actually built and attacked: what counts as evidence for “on purpose” when all you have is bones and dirt, why a peer-review process built to be transparent turned this into a public brawl instead of a quiet correction, and why a brand-new 2026 twist — a claim that every sampled skeleton in the chamber is female — shows the same reasoning trap in miniature.

560 cc560 cc560 ccMAX BRAIN VOLUMEvs. ~1,350 cc in H. sapiens

335 ka335 ka335 kaOLDEST FOSSIL AGEyoungest bound: 236 ka

100,000+ YRS100,000+ YRS100,000+ YRSEARLIER THANthe previous oldest known deliberate burial

The misconception: a stunning number is not the same as a settled question

Here is the wrong mental model almost everyone brings to a story like this: if the odds of something happening by chance are a million to one, then the dramatic explanation must be correct. It feels like arithmetic settling a debate. It is actually where debates like this one usually go to hide.

Run the numbers yourself. In June 2026, the same Rising Star research team published a genuinely new finding, unrelated to the burial question until you think about it for a second: a proteomic analysis of 23 teeth from at least 20 different H. naledi individuals found zero copies of amelogenin-Y, the enamel protein encoded on the Y chromosome that (unlike its X-chromosome counterpart) only biological males carry. Every individual they could test read as female — a finding independently reported by Smithsonian Magazine, which quotes the study team concluding there is “no convincing evidence” supporting “the confident identification of male individuals” in the chamber. Lee Berger, the project’s lead, put the odds in one sentence: “quite literally one in a million”Lee Berger, quoted on the probability of sampling twenty individuals who are all one sex by chance. University of the Witwatersrand press release, June 2026.

runnable · js▸ run

output

Run the code to see its output here.

Run it and the one-sided number lands at roughly 1 in 1,048,576 — Berger’s “one in a million,” almost to the digit. That number is real, and it is doing real work: it says the pattern almost certainly is not an accident of a small, randomly-sampled group. What it cannot do, no matter how small it gets, is tell you which non-accidental explanation is correct. A vanishingly small p-value rejects the boring null hypothesis; it does not crown whichever exciting alternative you’d already been hoping for. Even the paper’s own supporters note the honest alternative: the amelogenin-Y gene itself could have degraded, mutated, or gone missing in this specific, geographically isolated population over hundreds of thousands of years — in which case every individual could still include males whose Y-linked marker simply didn’t survive to be read. Sampling bias, preservation chemistry, and a genuinely sex-biased cave all predict the same zero. The number that sounds like a verdict is really a filter, and it only filters out one hypothesis: pure chance.

Hold onto that shape. It is exactly the shape of the fight over whether H. naledi buried its dead at all — except instead of one statistic, the evidence is a chamber full of bones, and instead of one alternative explanation, there are half a dozen.

Concept 1: what would even count as “on purpose”?

Before the naledi case can make sense, you need the actual toolkit archaeologists use to distinguish a grave from an accident, because “the bones were found together” proves almost nothing on its own. Caves swallow bodies constantly, with zero intent involved: animals fall down shafts and die; floodwater sweeps carcasses into low points and sorts them by size the way a river sorts pebbles; predators drag prey into dens and leave gnawed remains in piles. A chamber full of hominin bones is, by default, evidence of a death trap or a carnivore lair — not a cemetery. The burden sits entirely on the side arguing for intent.

The discipline that does this work is called taphonomythe study of what happens to an organism's remains between death and discovery — decomposition, scavenging, transport, burial, and fossilization — used to reconstruct how a fossil assemblage actually formed., and the specific sub-discipline for reading a body’s position is archaeothanatologythe systematic reconstruction of a body's decomposition sequence from the spatial relationships between its bones — which joints stayed articulated, which disarticulated and in what order, and what that implies about whether sediment supported the body early or late.. Its logic is mechanical, almost forensic: as soft tissue rots, ligaments loosen in a predictable order, and joints fall apart one by one unless something is physically holding the bones in place while it happens — most commonly, sediment that has already been packed around the body. Find a joint still perfectly articulated in an “impossible” position — a hand still curled as if just released, a foot standing on its own — and you have evidence that sediment arrived fast, before decomposition could scatter it. That is the single strongest kind of clue this field has for “somebody filled this hole in soon after the body went in it.”

This is exactly the class of evidence the original 2023–2025 eLife paper builds its case on. Berger and colleagues describe three areas within the Dinaledi Chamber — nicknamed the Puzzle BoxThe original 2013–2014 excavation area: a dense subsurface deposit representing a minimum of 6 individuals, with articulated hand-and-wrist and foot-and-ankle elements. Berger et al., eLife 2025 (article 89106)., Dinaledi Feature 1Discovered in 2018: predominantly a single adult body, with possible juvenile elements above it, showing articulated and matrix-supported elements. Left partially unexcavated to preserve testability for future teams. Berger et al., eLife 2025., and the Hill Antechamber FeatureDiscovered 2017–2018, more than 15 meters from the earlier Dinaledi excavations: remains of a minimum of 3–4 individuals, primarily a late juvenile, with an articulated hand and wrist plus a substantially articulated right foot. Berger et al., eLife 2025. — where they say the bone arrangement crosses that threshold.

1. 01

   Puzzle Box — the original 2013–2014 dig

   A dense subsurface deposit yielding a minimum of 6 individuals. Articulated hand-and-wrist elements sit alongside articulated foot-and-ankle elements — joints that, left exposed on an open cave floor, fall apart within weeks as ligaments decay.

2. 02

   Dinaledi Feature 1 — discovered 2018

   Predominantly a single adult body, with possible juvenile bone above it. The team deliberately left part of this feature unexcavated — a bet that a future, independent team could re-test the claim against sediment nobody has touched yet.

3. 03

   Hill Antechamber Feature — discovered 2017–2018

   Fifteen-plus meters from the earlier dig: a minimum of 3–4 individuals, mostly a late juvenile. The paper describes the body as deposited supine, knees and ankles flexed into what it calls a cross-legged position — and, resting inside the same feature at a 25-degree tilt, a single shaped stone the authors call a possible artifact.

That stone object, resting at a 25-degree tilt inside the Hill Antechamber Feature, drew immediate pushback from reviewers¹1At least one reviewer asked the authors to simply drop the object from the paper rather than defend an ambiguous claim. They refused, writing in their public reply that “omitting this one object from the report would be simply dishonest.” Whatever you make of the object itself, that exchange — a named objection and the authors’ named refusal, both permanently on the record — previews the argument to come. source — a small, early preview of the larger fight. Berger and colleagues argue the same three features rule out every ordinary alternative at once: no tooth marks or chew damage that would indicate carnivore accumulationBodies or bones dragged in and consumed by a predator, typically leaving characteristic bite/gnaw marks — explicitly checked for, and reported absent, in the Dinaledi features. Berger et al., eLife 2025.; sediment that formed under dry conditions with no sign of the coarse, sorted grain you’d expect from a flash flood or mudflowWater-borne deposition typically leaves coarse, size-sorted sediment and abraded bone surfaces from tumbling in current — neither of which the team reports finding in the fine-grained Unit 3 sediment. Berger et al., eLife 2025 and Dirks et al., eLife 2015. that would have washed bodies in; and bone tilted at angles the paper calls impossible to hold “subvertical” without matrix support already packed around it before the soft tissue was gone. Line all of that up, they argue, and the simplest remaining explanation is that something dug a hole, put a body in it, and filled it back in.

None of this is new to the 2023 papers, either — it goes back to the very first description of the site. The original 2015 geological paper on the Dinaledi Chamber already leaned toward intent, concluding that the “preliminary evidence is consistent with deliberate body disposal in a single location,” while cautioning that “a number of other explanations cannot be completely ruled out.” A decade of escalating claims sits on top of that one careful hedge.

The other side of the same sediment

Here is where the fight actually is. Two independent, peer-reviewed critiques — not press releases, not documentary rebuttals, published papers with their own reviewers — looked at the identical sediment and reached the opposite reading.

In November 2023, María Martinón-Torres and colleagues published a direct response in the Journal of Human Evolution, arguing that substantial additional documentation and analysis are needed before natural agents and post-depositional processes can be ruled out. In this primer’s own reading, that gap amounts to the absence of a consistent, feature-by-feature accounting of which joints stayed articulated, which fell apart, and in what sequence, applied evenly across all three named areas rather than showcased for one. Without that documentation, Martinón-Torres and colleagues argue, the paper hasn’t excluded the ordinary alternatives: a natural death trap, water transport of bodies or body parts, or scavenging — each of which can also leave partially articulated bones behind.

Then, in January 2025, Kimberly Foecke, Alain Queffelec, and Robyn Pickering went after the geochemistry directly in PaleoAnthropology, in a paper titled, without much room for interpretation, “No Geoarchaeological Evidence for Deliberate Burial by Homo naledi.” Their claim is not just “the interpretation is premature” — it’s that the underlying statistics don’t hold up: the sediment samples the original team compared as “inside the pit” versus “outside the pit” physically overlapped, violating the basic experimental design the comparison depends on, and the resulting data reads, in their words, as “fitted to a presupposed narrative” rather than the other way around.

filter4 rows

The same evidence, read two ways

Evidence line↕	Berger et al.: reads as burial↕	Critics' natural-process reading↕
Subvertical / tilted bone orientation	Impossible without sediment already packed around the body before ligaments decayed	Sediment slumping into a decomposition void can tilt bones the same way — no digging required
Sharp sediment-layer contrast within a “feature”	A dug-and-refilled pit reads differently from the undisturbed floor around it	Differential compaction and water percolation sort sediment into layers over 250,000 years on their own
Dry, non-fluvial sediment composition	Rules out a flash flood or mudflow sweeping bodies in all at once	Consistent with gravity creep or animals dying one at a time over millennia — dry doesn't mean dug
Geochemistry of “inside” vs “outside” the feature	Sediment chemistry inside the proposed pits measurably differs from sediment outside them	The sampled zones overlapped and the statistical comparison was mis-specified — properly analyzed, the difference isn't there

Go deeper: what a rejected null hypothesis does and doesn't prove here

eLife’s own editorial assessment of the 2023 submission put this in exactly the language of the misconception you already tested above. Summarizing the four reviewers, whom it described as “in strong consensus that the methods, data, and analyses do not support the primary conclusions,” the assessment stated that “the null hypothesis must be that these skeletons accumulated naturally” and that the burden of proof sat entirely with the authors to rule that null out — not merely to show the burial story was possible, but that it was the only story the sediment supported. Rejecting an accumulation-by-accident null is necessary for a burial claim. It has never been sufficient, on its own, for either side of this argument.

Notice what the table above is not doing: it is not splitting one side into “the scientists” and the other into “the skeptics.” Every row has a credentialed, peer-reviewed author on both sides, reading the identical photographs, the identical grain-size data, the identical joint positions. That is what makes this a genuine live dispute rather than a settled finding fighting off cranks — the disagreement survives contact with the same raw material, which is exactly the situation archaeothanatology was built to resolve and, so far, has not.

Concept 2: the geography that makes “how did they even get in” load-bearing

There’s a second, less statistical line of attack, and it’s just as central: getting a body — or a whole hominin, alive, repeatedly, over generations — into this chamber was never trivial. The Dinaledi Chamber sits roughly 30 meters below the surfaceAnd approximately 80 meters, along the known route, from the nearest cave opening — a chamber roughly 15 meters long, with variable widths not exceeding 2.5 meters. Dirks et al., eLife 2017 (article 24231). and about 80 meters, by the original known route, from daylight. Reaching it means clearing three obstacles in sequence: a 15-meter climb up a knife-edged dolomite ridge nicknamed the Dragon's BackA sharp-edged dolomite ridge roughly 15 meters high that excavators had to climb before descending toward the chamber. Wikipedia, “Rising Star Cave.”, a horizontal squeeze called Superman's CrawlA passage so narrow most people can fit only with one arm pinned to the body and the other extended overhead, Superman-flight style. Wikipedia, “Rising Star Cave.” — passable, per the site’s own Wikipedia entry, only with one arm pinned to your side and the other stretched overhead — and finally a near-vertical chute measured at 12 meters long and about 20 centimeters wide on averageDiscovered by recreational cavers Rick Hunter and Steven Tucker on 13 September 2013; the six original excavators were chosen in part for being small and slender enough to fit through it. Wikipedia, “Rising Star Cave.”, roughly the width of a hardcover book turned on its spine.

Nodes

• Surface entrance
• Dragon's Back (15m climb)
• Superman's Crawl
• The Chute (12m, ~20cm wide)
• Dinaledi Chamber (−30m)
• Puzzle Box (6 individuals)
• Feature 1 (1 adult)
• Hill Antechamber Feature (3–4 individuals)
• Robbins et al. 2021: alternate route (−80m of travel)

Connections

• Surface entrance → Dragon's Back (15m climb) (known route)
• Dragon's Back (15m climb) → Superman's Crawl
• Superman's Crawl → The Chute (12m, ~20cm wide)
• The Chute (12m, ~20cm wide) → Dinaledi Chamber (−30m)
• Dinaledi Chamber (−30m) → Puzzle Box (6 individuals)
• Dinaledi Chamber (−30m) → Feature 1 (1 adult)
• Dinaledi Chamber (−30m) → Hill Antechamber Feature (3–4 individuals)
• Surface entrance → Robbins et al. 2021: alternate route (−80m of travel) (proposed shortcut)
• Robbins et al. 2021: alternate route (−80m of travel) → Dinaledi Chamber (−30m) (80m shorter, contested)

Why does a shorter route matter at all to a claim about bones and sediment? Because of a principle older than any of these papers: the explanation that requires fewer unlikely steps wins, all else equal. An accidental fall is a single, ordinary event; a repeated, deliberate funerary practice by a small-brained hominin requires planning, memory of the route, and a shared behavior across individuals and generations. A harder original route forces more of that extra machinery on you, because “wandered in and got stuck” stops being plausible. An easier alternate route removes that requirement — which is exactly why the length of a crawlspace 250,000 years old is not a side detail in this argument. It is load-bearing for the whole case.

Berger’s team argues the harder route is not an obstacle to their claim, because H. naledi’s curved fingers and toes — part of the original 2015 species description — mark it as an exceptional climber, well within reach of a route modern excavators found merely difficult. One 2023 peer reviewer pushed back directly on that framing, pointing to a 2021 study by Robbins and colleagues describing an alternative approach that cuts roughly 80 meters off the traversal — meaning easy, repeated, even accidental access is more parsimonious than the original route implies. The authors’ published reply doesn’t dismiss the alternate route; it argues that even granting it, H. naledi’s climbing anatomy still makes deliberate, repeated entry the better-supported story. Reviewer and authors are on the record, disagreeing, in the same public document. Nobody’s comment vanished into a rejection letter.

Omitting this one object from the report would be simply dishonest.

That single line — about the disputed stone object from the Hill Antechamber Feature — is worth sitting with, because it captures something the traditional peer-review process usually hides: authors arguing, in public, for including evidence a reviewer would rather see cut. Most papers never let you watch that negotiation happen. This one did, on purpose, by design of the journal itself — which is the actual “aha” of this whole story.

Concept 3: the review process was built to make the fight visible, not to end it

This is where the story stops being only about bones and starts being about how science decides what counts as known. In June and July 2023, Berger’s team posted three linked manuscripts — the burial paper, a companion claim about carved marks on the cave wall, and a broader paper on the cognitive implications of both — as reviewed preprints on eLife, a publishing model the journal had adopted specifically to break with the traditional up-or-down gate.

A reviewed preprinteLife's publishing model (adopted 2023): a paper is posted publicly alongside its named or anonymous peer reviews, an editorial “eLife assessment” rating both significance and evidence strength, and the authors' response — with no formal accept/reject decision. Every version stays online permanently. replaces the traditional binary. A paper doesn’t get rejected into silence, and it doesn’t get quietly approved either — it gets an eLife assessmentA short editorial summary pairing a significance word (e.g. landmark, important, valuable) with an evidence-strength word (e.g. compelling, convincing, solid, incomplete, inadequate), attached permanently to every version of a reviewed preprint. and its reviewers’ names or numbers, forever attached to that exact version.

The first assessment, dated 12 July 2023, did not equivocate. Significance: landmarkeLife's own significance rating, v1 (12 July 2023): demonstrating the earliest known intentional funerary practice, in a small-brained hominin, would be a ‘landmark’ finding if substantiated.. Evidence strength: inadequatev1 eLife assessment (12 July 2023): ‘the evidence for these claims is considered inadequate in the current version of the study’ — with all four reviewers in strong consensus that the methods, data, and analyses did not support the primary conclusions..

Demonstration of the earliest known instance of intentional funerary practice — with a relatively small-brained hominin engaging in a highly complex behavior that has otherwise been observed from Homo sapiens and Homo neanderthalensis — would be a landmark finding. However, the evidence for these claims is considered inadequate in the current version of the study.

That verdict — “landmark, if true; inadequate, as shown” — is the whole tension of this story compressed into one sentence, and it stayed publicly attached to the paper while Berger’s team went back to the sediment. Twenty months later, on 28 March 2025, a revised version posted with the same significance rating but a split evidence call: one reviewer had come around, writing that the new material now “convincingly” demonstrated intentional burial; the other still marked it “incomplete,” specifically because the question of how the hominins got into the chamber remained, in their words, unresolved. eLife’s final published assessment — attached to the Version of Record as of 1 September 2025 — records both positions side by side, changing neither:

One of the reviewers concludes that the findings convincingly demonstrate intentional burial practices, while another considers evidence for such an unambiguous conclusion to be incomplete given a lack of definitive knowledge around how the hominins got into the chamber.

Event 1 of 10: 13 Sep 2013, Cavers find the chute

012013opening

13 Sep 2013

Cavers find the chute

Rick Hunter and Steven Tucker locate the entrance to what becomes the Dinaledi Chamber.

022013+49 days

1 Nov 2013

First excavation season

Berger's team recovers the first of what becomes 1,550+ fossils from at least 15 individuals.

032015+2 yr

10 Sep 2015

Homo naledi named

New species described; brain volume 465–560 cc, within the australopith range.

042017+2 yr

9 May 2017

Fossils dated

Four independent techniques place H. naledi at 236,000–335,000 years old.

052023+6 yr

12 Jul 2023

Burial claim posted — 'inadequate'

eLife's v1 assessment: 'landmark' significance, but all four reviewers call the burial evidence 'inadequate' as submitted.

062023+4 mo

10 Nov 2023

First published rebuttal

Martinón-Torres et al., Journal of Human Evolution: 'No scientific evidence' for burial or rock art.

072025+1 yr

22 Jan 2025

Second rebuttal: the sediment itself

Foecke, Queffelec & Pickering, PaleoAnthropology: the burial-pit geochemistry doesn't hold up statistically.

082025+2 mo

28 Mar 2025

Revised manuscript — split verdict

eLife v2: one reviewer now finds the case 'convincing'; the other still calls it 'incomplete.'

092025+5 mo

1 Sep 2025

Version of Record published

The paper is finalized with both reviewer positions permanently attached, no up-or-down vote taken.

102026+10 mo

24 Jun 2026

A new twist: 20 for 20

A Cell paper finds zero male protein markers across 23 teeth from 20+ individuals — reopening who, exactly, this chamber was for.

Notice what did not happen anywhere in this chronology: no editor overruled a reviewer, no reviewer’s objection was quietly dropped, and no side got the last word by default. A traditional journal would have handled the 2023 “inadequate” verdict by rejecting the paper outright — at which point Martinón-Torres and Foecke would have had nothing specific to respond to, because the rejected manuscript would likely never have become public in that form. Instead, everyone’s position — the original 2023 rejection-strength critique, the two 2023–2025 published rebuttals, and the split 2025 reviewer verdict — sits on the permanent record, at the same URL, for anyone to check. eLife’s transparency didn’t produce agreement. It produced an argument you can actually read.

1. Q1A 2025 eLife 'reviewed preprint' carries its reviewers' comments publicly alongside the paper, with no formal accept/reject. Compared to a traditionally rejected manuscript, what does this change?

   • Nothing — unconvincing work still just disappears either way
   • Disagreement becomes a permanent, citable part of the public record instead of vanishing into a rejection letter
   • It means every reviewer must agree before the paper can be posted at all

2. Q2Berger called the odds of 20-for-20 same-sex teeth 'one in a million.' What does that number actually rule out?

   • Every non-random explanation at once, including gene loss and sampling bias
   • Only pure random 50/50 chance — it doesn't distinguish between the remaining non-chance explanations
   • The possibility that any of the 20 individuals were biologically male

3. Q3According to Foecke, Queffelec & Pickering (2025), what specifically undermines the claim that “burial pit” sediment chemically differs from the surrounding floor?

   • The sampled 'inside' and 'outside' zones physically overlapped, violating the comparison's own design
   • No sediment was ever sampled at all
   • The two research teams analyzed sediment from entirely different cave systems

The 2026 twist runs the same play again

Which brings the story back to where this primer started. The all-female protein result is barely a month old as of this writing, and it is already following the identical arc as the burial claim: a dramatic, quantifiable pattern (zero AMELY markers across 20+ individuals), a headline-ready number attached to it (“one in a million”), a proposed dramatic explanation (the earliest known sex-specific mortuary space, predating any comparable practice in Homo sapiens or Neanderthals), and an immediate, entirely reasonable competing explanation that shares the same underlying data: the amelogenin-Y gene degrading or dropping out across an isolated population over 250,000+ years, which would produce the exact same zero without a single deliberate choice involved.

Notice that this new claim doesn’t even resolve the older one — it depends on it. “A sex-specific burial chamber” only means something if there’s a burial chamber to begin with, which is precisely the fact still under open dispute in the previous section. The 2026 protein paper is a second dramatic number sitting on top of a first dramatic claim that two independent, peer-reviewed critiques still say the sediment doesn’t support. That is not a knock against the protein work itself — the peptide chemistry is a genuinely separate, well-controlled result. It is a reminder of exactly the lesson from the RunnableCode sandbox at the top of this primer: a small, real probability can be entirely correct about ruling out coincidence, and still tell you nothing about which of several remaining explanations you should believe.

Go deeper: why neither side of this story is going to be settled by one more paper

Both the burial claim and the all-female claim share a structural weakness that no additional statistic fixes on its own: a single cave system, excavated by a single team, is not independent replication. The 2023 eLife reviewers said this outright, asking that “independent teams” be allowed to examine the same features — precisely because archaeothanatological judgment calls (is this joint “articulated,” is this sediment layer “sharp”) are exactly the kind of thing a second set of trained eyes, with no stake in the original claim, can confirm or overturn in a way a re-analysis of the same photographs cannot. Until that happens, on either question, the honest state of the science is genuinely split — not “leaning” one way, split.

Where this leaves you

Strip away the specific bones and this is a primer about a general skill: reading a claim that rests on ambiguous physical evidence for a behavior nobody witnessed. The move that matters is always the same, whether the “evidence” is a tilted femur or a missing protein marker. First, ask what the null explanation predicts — what would this look like if nothing deliberate happened at all? Second, ask whether the dramatic explanation has actually been tested against that null, with a method that could have found the null true, or whether the dramatic explanation was simply assumed and then illustrated. Third — and this is the step the “one in a million” framing tempts you to skip — ask how many different non-null explanations would produce the exact same pattern you’re looking at, because a small probability only eliminates one of them.

By that standard, the Homo naledi burial claim is not settled science being denied by cranks, and it is not debunked pseudoscience still limping along on a Netflix documentary’s momentum. It is a live, technically serious disagreement between credentialed teams, conducted almost entirely in public, over what a specific archaeothanatological signature does and doesn’t prove about intent. eLife’s model didn’t manufacture that disagreement — it just refused to hide it behind a rejection letter the way a traditional journal would have. The next time a headline tells you extraordinary evidence has “proven” a dramatic story, look for the null hypothesis it’s supposedly ruling out, and ask what else, besides the dramatic story, would produce the exact same number.