COVID-19 Origin: 2025 Updated Analysis - Lab Leak vs. Natural Zoonotic Spillover

March 2025 updated "big picture" analysis of COVID-19 origins... low confidence but still more likely natural zoonosis via wet market

Mar 13, 2025

The origin of SARS-CoV-2, the virus causing COVID-19, remains a subject of intense debate. Two primary hypotheses have emerged: (1) zoonotic spillover (a jump from animals to humans, possibly at a wet market in Wuhan) vs. (2) laboratory leak (an accidental release from research at the Wuhan Institute of Virology or another lab).

I’ve put together a report to examine evidence and arguments for each scenario — drawing on: intelligence assessments, scientific findings, Chinese government (CCP) actions, independent expert analyses, historical precedents, and probabilistic reasoning.

TLDR: My odds still favor natural zoonosis (~60%) via the Huanan wet market over lab leak (~40%) — even after a recently-unearthed German Intelligence report expressed subjective 80-95% confidence in a COVID-19 lab leak. That said, until smoking gun evidence emerges, I retain “low confidence” in any prediction. If a lab leak occurred, I retain “high confidence” it was accidental (unintentional).

I. Intelligence Agency Assessments (U.S. & International)

United States Intelligence: U.S. agencies have not reached a consensus on COVID-19’s origin, and their assessments diverge:

Federal Bureau of Investigation (FBI): The FBI concluded with moderate confidence that a lab accident in China was the most likely cause of the pandemic. FBI Director Christopher Wray stated the Bureau “suspects the Wuhan lab” as the origin, citing the lack of evidence for an animal spillover in Wuhan and the distance of the outbreak from natural bat reservoirs. This moderate-confidence assessment is based largely on circumstantial evidence and the absence of a proven animal host.
Department of Energy (DOE): In early 2023, the DOE shifted its stance and assessed that a laboratory leak was likely, but with low confidence. This conclusion (revealed in a classified report to the White House and Congress) was first reported by The Wall Street Journal and reflects new intelligence and expert analysis that caused DOE analysts to lean toward the lab-leak hypothesis. The low confidence indicates significant uncertainty and a need for more information.
Central Intelligence Agency (CIA): After long stating it could not determine the origin, the CIA in January 2025 issued a new assessment that a “research-related incident” is more likely than a natural origin. However, the CIA assigned only low confidence to this judgment and explicitly noted that both scenarios remain plausible. This suggests the CIA sees slightly stronger indications of a lab origin but still lacks definitive evidence. (Notably, this shift came after CIA Director William Burns urged a clear determination due to the pandemic’s significance.
Other U.S. Intelligence Elements: According to declassified summaries, several other intelligence components and the National Intelligence Council lean toward a natural spillover (zoonosis) with low confidence. An October 2021 Office of the Director of National Intelligence (ODNI) report stated the Intelligence Community was divided on origins and might never reach a definitive answer, with analysts split between the lab-leak and zoonotic theories. All agencies agree the virus was not developed as a bioweapon. In sum, U.S. intelligence remains split, with the FBI, DOE, and now CIA favoring a lab origin to varying degrees of confidence, while others favor natural emergence – but none have high confidence in their conclusions.

International Intelligence: Allied and foreign intelligence services have conducted their own inquiries, sometimes arriving at strong (if secret) conclusions:

Germany (BND): In March 2025, media reports revealed that Germany’s foreign intelligence service (Bundesnachrichtendienst, BND) concluded as early as 2020 that an 80–95% probability exists that COVID-19 originated from a laboratory accident in Wuhan. The BND had gathered indications of risky gain-of-function experiments at the Wuhan Institute of Virology and multiple safety protocol violations. This high-confidence internal assessment was reportedly shared with the U.S. CIA in late 2024. However, previous German administrations kept these findings confidential, fearing political fallout. Only with new leadership did details emerge that the BND strongly suspected a lab origin from the start, based on both public data and a covert operation (“Saaremaa”) targeting Chinese labs. The 80–95% lab-leak probability cited by BND is notably higher than most other agencies’ estimates, indicating that German intel found the lab-leak evidence compelling. (It’s worth noting that prominent German virologist Christian Drosten commented that Chinese scientists “have the technical capabilities to provide evidence for the natural origin” if it were true, underscoring frustration that such evidence hasn’t been forthcoming.
Other Countries: Official public conclusions from other intelligence agencies are scarce. The UK, for example, has not released an official assessment, though a former MI6 chief (Sir Richard Dearlove) has publicly voiced his belief in a lab leak. Likewise, intelligence bodies in Australia, Japan, and others have generally supported thorough investigations but have not declared firm conclusions. Overall, no foreign intelligence service is known to definitively endorse the zoonotic market origin in strong terms; rather, many quietly consider a lab scenario plausible. The consensus among the “Five Eyes” countries in mid-2020 reportedly was that a natural origin was more likely but that a lab leak could not be ruled out. Recent shifts (like the CIA and BND findings) show increasing willingness to entertain the lab-leak theory, albeit often with caveats of low confidence.

Overall: Intelligence assessments are divided. U.S. agencies differ, with some favoring a lab accident (FBI, DOE, CIA) and others undecided or leaning natural, all with low to moderate confidence. Germany’s BND stands out by assigning a very high probability to a lab origin. Importantly, no agency claims to have smoking-gun proof for either hypothesis – the findings rely on piecing together indirect evidence, and several agencies acknowledge that both lab and natural spillover remain plausible.

REFERENCES

II. Scientific & Logical Evidence (COVID Origin)

From a scientific perspective, researchers have examined the genetic makeup of SARS-CoV-2, the pattern of early cases, and evolutionary history of coronaviruses to infer the likely origin. Here we evaluate key evidence and the current expert thinking:

Genomic Features & Engineering Debate: Early in the pandemic, some hypothesized that SARS-CoV-2’s genome bore signs of lab manipulation – particularly the presence of a unique furin cleavage site in the spike protein not seen in close relatives (like bat coronavirus RaTG13). This feature (an “RRAR” amino acid insertion) enhances the virus’s ability to infect human cells, leading to speculation that it might have been artificially inserted. One oft-cited claim was the occurrence of two consecutive “CGG” codons (encoding arginine) at the furin site, an uncommon pattern in coronaviruses, which some argued was a telltale sign of genetic engineering. However, comprehensive analyses by virologists have debunked the notion that the genome shows clear engineering fingerprints. Scientists note that similar furin cleavage sites exist in other coronaviruses (e.g., MERS-CoV has one), and “rare” codon pairs can occur naturally. In fact, if researchers had intentionally inserted a cleavage site, they likely would have used a more typical sequence (such as “RRKR”) expected to work well; SARS-CoV-2’s particular sequence was not an obvious choice and even virologists were initially puzzled by it. The way the furin site is encoded – via a complex frameshift insertion – looks more like a quirky product of natural evolution than a synthetic construct. Furthermore, as the virus evolved in humans, it actually mutated that site toward a more “normal” form, suggesting the original was suboptimal. No telltale lab vector sequences or restriction enzyme cut sites were found in the genome to indicate cloning. The current scientific consensus is that there is no definitive genetic signature of bioengineering in SARS-CoV-2, although this doesn’t exclude a lab accident with a naturally evolved virus. In summary, genetic evidence does not conclusively favor either scenario: the genome is consistent with natural evolution, and nothing in it definitively proves artificial creation.

Closest Viral Relatives: SARS-CoV-2’s nearest known relatives are bat coronaviruses discovered in nature – e.g., RaTG13 (96.2% similarity) and BANAL-52 from Laos (~96.8%). These are too divergent to be a direct precursor, implying a gap in sampling. Proponents of zoonotic origin argue that undiscovered viruses in wildlife could easily bridge this gap. Indeed, virologists like Edward Holmes and colleagues have pointed out that our knowledge of coronaviruses in bats and intermediate hosts is incomplete, and more sampling could find a progenitor. Lab-leak proponents counter that the Wuhan lab was actively collecting and experimenting on a large catalog of bat viruses (including unpublished ones), raising the possibility that SARS-CoV-2 could have been among those or derived from them. Notably, there is no public record that any lab had a SARS-CoV-2 strain (or an extremely close ancestor) prior to the outbreak. Wuhan Institute of Virology scientists have stated they did not have SARS-CoV-2 in their possession, and no evidence has surfaced to contradict that. The lack of a known “virus X” that was modified into SARS-CoV-2 leaves the lab-leak genetic path speculative. Meanwhile, finding a very close cousin in an animal could strongly support a natural origin – but to date, no such animal source has been pinpointed. This absence of a definitive intermediate host after three years remains a crucial gap: as a U.S. Senate report noted, despite extensive testing of animals, “there is no evidence that any animal was infected with SARS-CoV-2 prior to the first human cases”. This missing link is problematic for the natural spillover theory, though not fatal – it took years to find the animal source for some diseases, and in a few cases it remains unknown.

Epidemiological Evidence – Early Cases in Wuhan: The initial outbreak geography and timing provide important clues. Wuhan was the first epicenter, with the earliest known cases in December 2019. Critically, many of the early cases were clustered around the Huanan Seafood Wholesale Market in Wuhan. Several of the first patients either worked at the market or had visited it, and environmental samples from the market (swabs of floors, cages, etc.) tested positive for the virus. Two landmark studies published in Science (July 2022) provided strong evidence that the market was the epicenter of the outbreak, supporting a spillover event there. One analysis of case geography showed that the density of early COVID cases was highest in the direct vicinity of the Huanan market, not uniformly spread through Wuhan – this kind of clustering is what one would expect if the market was the source of the virus transmission to humans. Another study analyzing the viral genomes from early cases found they fell into two distinct lineages (A and B), which likely correspond to two separate spillover events at the market around late November to early December 2019. In other words, scientists suspect that two jumps from animals to humans occurred independently at Huanan market, seeding the two early viral strains – a very specific pattern that is hard to explain except by multiple infected animals or animal batches. As commentator Scott Alexander noted, “it would be a bizarre coincidence if a lab leak pandemic was first detected at a wet market – and an even more bizarre coincidence if a lab leak somehow split into two lineages that both happened to emerge at that market”. This observation encapsulates why many experts see the market origin as more plausible: a lab leak would have to magically center itself on the one seafood market in Wuhan, whereas a natural spillover can easily account for the market cluster (infected animals being sold there). The strong epidemiological linkage to the market is one of the most compelling pieces of evidence for zoonotic origin.

Animal Host and Raccoon Dog Evidence: A major bolster to the zoonotic theory came in 2023 with new analysis of China’s own data. Chinese CDC researchers had swabbed the Huanan market in early 2020; an international team re-examined genetic sequences from those swabs and discovered that some COVID-positive samples also contained DNA from raccoon dogs – an animal known to be susceptible to coronaviruses and sold live in the market. This finding doesn’t prove that raccoon dogs were infected or that they transmitted the virus to humans, but it does show that such animals (or their meat) were present exactly in the locations where virus was found. As The Atlantic reported, “genetic samples from the market show that raccoon dogs being illegally sold there could have been carrying and shedding the virus in 2019. It’s some of the strongest support yet that the pandemic began when SARS-CoV-2 hopped from animals into humans, rather than in an accident among scientists” . Many virologists saw this as a crucial piece of the puzzle that had been missing – essentially a smoking gun that an infected animal was at the market. The data are still circumstantial (no actual animal carcass was tested before the market was cleared), but this evidence aligns neatly with a natural spillover narrative. It also underscores that Chinese authorities had data pointing to a market animal source but did not share it openly at the time.

Expert Consensus and Scientific Opinion: Within the scientific community, the pendulum of opinion has swung as new data emerged, but many experts continue to favor a zoonotic origin. Early on, a group of virologists authored The Proximal Origin of SARS-CoV-2 (Nature Medicine, March 2020), which argued the genome showed no signs of engineering and strongly leaned toward natural emergence. Over time, additional evidence (market clustering, raccoon dog data, etc.) has reinforced the natural hypothesis for many. As of 2023, prominent researchers like Dr. Angela Rasmussen, Dr. Edward Holmes, and others have publicly stated that a natural spillover at the market is the most likely scenario given available evidence. A comprehensive 2021 review in Cell by Holmes et al. concluded that all known facts were consistent with a zoonotic origin and that a lab leak, while not impossible, lacked supporting evidence. On the other hand, a minority of scientists and investigators remain concerned about a lab origin – notably Dr. Alina Chan and Lord Matt Ridley, authors of the book Viral, and molecular biologist Richard Ebright have pointed out the coincidences and called for deeper investigation into the WIV. These experts highlight that no host animal has been identified, and that WIV was conducting high-risk research on similar viruses, so prudence requires keeping the lab hypothesis on the table. The mainstream scientific consensus, insofar as one exists, leans toward natural spillover – but it is a softer consensus than for past outbreaks, due to the unusual circumstances. Many scientists would agree with a recent Chinese study’s conclusion: “we believe the coronavirus was originally transmitted to humans by animals rather than by a laboratory leak. However, more investigation is needed to determine the source”. In sum, most virologists find a natural origin more plausible based on current data (especially the epidemiological pattern and analogies to past pandemics), but virtually all concede that without China’s full cooperation and more evidence, one cannot be 100% certain.

Evidence favoring a zoonotic spillover:

Multiple early cases directly linked to the Huanan wildlife market, with a geographic cluster centered on that location.
Two early viral lineages (A and B) suggesting separate introductions into humans, consistent with multiple infected animals at the market.
Detection of SARS-CoV-2 genetic material mingled with DNA of wild animals (e.g. raccoon dogs) in the market’s environmental samples.
Precedent of SARS-CoV-1 (2002–03) emerging from animal markets (civets) and numerous other coronaviruses in nature; most pandemics historically have zoonotic origins.
No clear genetic evidence of bioengineering; virus features can be explained by known evolutionary mechanisms.
WIV scientists insist they had no such virus; closest known relatives are bat viruses found in the wild, pointing to a natural reservoir.

Evidence favoring a lab leak:

The outbreak’s ground zero was Wuhan, a large city that happens to host China’s leading coronavirus research lab. It’s an uncanny coincidence for a natural jump to occur in the one city researching similar viruses – this raises Bayesian prior odds for a lab scenario.
Wuhan Institute of Virology (WIV) researchers were known to be collecting and experimenting on novel bat coronaviruses, including gain-of-function work to test infectivity. Intelligence reports (U.S. and German) indicated WIV had biosafety issues and possibly risky experiments that could have gone awry.
Reports (not officially confirmed by China) that WIV staff fell ill with COVID-like symptoms in November 2019 have circulated, suggesting a possible lab infection chain (U.S. intel has said they cannot confirm this conclusively).
China’s reluctance to share raw data or allow full investigations of WIV is seen as suspicious. If it were a natural spillover, full transparency would presumably only help confirm it, whereas a lab origin would prompt a cover-up – China’s obstruction is at least consistent with a cover-up scenario.
Despite intensive searching, no intermediate animal host has been identified, and no animal outbreaks were reported prior to the Wuhan human outbreak. By contrast, in the SARS-1 outbreak, market animals (palm civets) with nearly identical virus were found within months. The absence of evidence in animals (while not evidence of absence) makes some suspect that maybe there was none – because the source was a lab, not wildlife.
Some statistical analyses have questioned whether the market truly was the origin or just an amplification location. For example, one recent re-analysis (Stoyan & Chiu, 2023) argued that the market cluster could be a result of superspreading there and not necessarily the first jump – though this view is in the minority.

Ultimately, the scientific case is circumstantial for both sides. The data from genomics and field studies tend to align more with natural spillover, and many experts find that explanation simpler and in line with past experience.

However, critical gaps (no proven animal infection prior to humans, and no transparency into the lab’s records) keep the lab-leak theory plausible.

The situation is akin to a puzzle missing key pieces: what we have fits a natural origin quite well, but we cannot prove the lab hypothesis false without those missing pieces – especially given the unique coincidence of a major coronavirus lab in the outbreak city.

REFERENCES

III. Actions of the Chinese Government (CCP)

The behavior of Chinese authorities and the Chinese Communist Party (CCP) in the aftermath of the outbreak is a major point of analysis.

QUESTION: Do China’s actions align more with a government trying to cover up a laboratory accident, or with a government reacting (perhaps over-cautiously) to a natural zoonotic outbreak?

Initial Suppression of Information: In the earliest weeks (December 2019 – January 2020), Chinese authorities downplayed the outbreak. Doctors in Wuhan who sounded the alarm (such as the late Dr. Li Wenliang) were reprimanded for “spreading rumors.” Laboratories that independently decoded the virus’s genome were ordered to destroy samples and not release findings without authorization. In mid-January 2020, China’s National Health Commission reportedly issued a notice barring labs from publishing any data on the novel virus and ordering existing samples either be moved to designated facilities or destroyed – an action Beijing later confirmed was done for “biosafety reasons”. While Chinese officials denied it was to hide evidence, the destruction of early samples and gagging of information naturally fueled suspicions of a cover-up. If a lab leak had occurred, early samples might have revealed telltale signs or linked back to a lab strain – their destruction is consistent with an intent to prevent tracing. (Even in a natural spillover scenario, the reflex to control information and avoid blame still applies – so this alone isn’t proof of a lab origin, but it set a tone of opacity.)

Silencing of Scientists & Journalists: Chinese researchers and citizens who tried to share information about the virus’s origin faced pressure. In February 2020, two Chinese academics posted (and quickly retracted) a preprint suggesting the virus might have leaked from a Wuhan lab – their university publicly disavowed it, and the paper vanished, implying official disapproval. Chinese authorities also clamped down on discussion of the outbreak’s origin, censoring social media speculation about the labs. Meanwhile, citizen journalists who reported from Wuhan (filming conditions in hospitals and markets) were detained. This pattern of intimidation and tight message control is more suggestive of a government in damage-control mode (as would be expected if an embarrassing lab accident occurred) than a government confident in a natural-origin narrative. Had the virus unquestionably come from a wild animal, Chinese scientists would arguably be celebrated for tracing it and eliminating the source; instead, some scientists (even very senior ones) have been muzzled regarding origins. Notably, in 2020 the director of China’s CDC, Dr. Gao Fu, hinted that the Huanan market might have been a victim rather than the source (suggesting the virus could have been brought in) – a statement that downplayed the market origin. This could indicate an attempt to deflect blame from domestic sources (whether market or lab) toward murkier scenarios.

Data Obfuscation & Withholding: International teams have repeatedly requested raw data on early cases and viruses from China – with little success. The World Health Organization (WHO) sent a joint mission to Wuhan in early 2021, but the team was not given access to all records, such as raw patient data from early cases, and was not allowed to audit the Wuhan labs. Key databases at the Wuhan Institute of Virology, including one containing details of thousands of bat coronavirus samples, were taken offline in late 2019 and have not been shared since. China has also delayed or refused to share certain findings: for example, the raccoon dog genetic data discussed earlier was uploaded to a public genetic database by Chinese researchers only in March 2023 (over three years after it was collected), and then it was quickly removed – it became available to the global scientific community only because some researchers downloaded it before it vanished. This kind of obfuscation has been noted by experts: The Atlantic described a “swirl of data obfuscation by Chinese authorities” that impeded efforts to understand how the pandemic began. Such behavior is consistent with a government that does not want independent parties finding something incriminating. If a lab accident occurred, one would expect exactly this pattern: tight control of all potentially revealing data (lab records, sample databases, original virus sequences). On the other hand, even in a natural spillover, China might fear blame for poor food safety or early missteps and thus still hide information – but arguably there would be less to hide (e.g. hiding the existence of illegal wildlife trade is simpler than hiding a lab accident).

Refusal of Further Investigation: Perhaps the clearest indication of China’s stance is its response to international calls for transparency. In July 2021, the WHO proposed a Phase 2 origin study, which specifically included investigating the lab leak hypothesis and auditing Wuhan labs. Beijing flatly rejected this plan. A top Chinese health official said they would “not accept such an origins-tracing plan” and dismissed the inclusion of a lab protocol breach hypothesis as “disregarding common sense and defying science”. China’s officials insisted that any suggestion of a lab leak was politically motivated and not scientifically grounded, and they have since refused to participate in further WHO-origin probes that involve scrutinizing labs. The head of WHO, Dr. Tedros Adhanom Ghebreyesus, in 2021 complained that China’s lack of sharing raw data from the early outbreak was hindering the investigation. China defended itself by citing patient privacy and biosecurity concerns, but to outside observers this comes across as stonewalling. The refusal to allow independent lab audits or full data transparency is widely interpreted as an attempt to control the narrative. If the pandemic truly came from a wild animal, one might expect China to welcome lab audits to clear its flagship lab of suspicion – instead, the denial of access raises eyebrows.

Promotion of Alternative Theories: The Chinese government has also actively promoted other origin theories to muddy the waters. State media and officials have at times suggested COVID-19 might have started outside China – for instance, pointing to frozen food packaging or a U.S. military laboratory (Fort Detrick) as potential sources. In March 2020, Beijing officials floated the idea that the virus could have been brought to Wuhan during the World Military Games in October 2019 by foreign athletes. These claims are unsubstantiated and seen by experts as distractions. The proliferation of such theories indicates a reluctance by China to accept any responsibility. While not directly proving a lab leak, this propaganda approach fits a scenario where the truth is inconvenient – a classic “throw sand in their eyes” tactic consistent with a cover-up.

Pattern of Cover-up vs. Cooperation: Comparing China’s response to this outbreak with past zoonotic outbreaks is telling. During the SARS-1 outbreak (2003), Chinese officials also initially covered up the outbreak – that was a zoonotic event, yet the cover-up was about the outbreak’s existence, not its origin. Once international pressure mounted, China did eventually help identify civet cats as the source and shut down wildlife markets (albeit after some delay). In the COVID-19 case, China did shut down the Huanan market and banned farming of certain wild animals, which suggests they at least considered an animal source plausible. However, their complete lack of interest in finding an intermediate host publicly is notable. Instead of a transparent search for the animal reservoir (as happened with SARS-1, where Chinese scientists were on the hunt for animal hosts), the focus quickly shifted to controlling the narrative. For example, in early 2020, the Chinese CDC tested tens of thousands of animal samples across China – and reported none were positive for SARS-CoV-2, then largely went quiet on the animal origin hunt. If the government truly believed the market spillover theory, one might expect a much more aggressive and open hunt for the host species to exonerate the labs. The relative silence on this front (apart from denying that the Wuhan market findings were conclusive) could indicate that authorities either know it wasn’t a natural spillover or simply prioritize deflecting blame over scientific answers.

Overall: Chinese government actions have largely been obstructive and secretive, which many interpret as aligning with a lab-leak cover-up. Key examples include ordering the destruction of early virus samples, punishing whistleblowers, refusing to share data or allow lab inspections, and rejecting WHO’s follow-up investigations. Such behavior is consistent with a desire to hide something like a lab accident. However, it’s important to note that authoritarian regimes may reflexively hide information even for a natural event, out of fear of political repercussions. Thus, while China’s lack of transparency strongly suggests they are concealing the origin, it doesn’t conclusively tell us which origin it is – only that the true origin is politically inconvenient for the CCP. Whether that inconvenience is covering up risky research or simply covering up mismanagement of wildlife trade remains the crux of the matter.

REFERENCES

IV. Expert & Independent Analyses (COVID)

Beyond official agencies and academic scientists, independent experts, forecasters, and “rationalist” communities have weighed in on COVID-19’s origin, often applying first-principles reasoning or probabilistic analysis to the problem. These perspectives are valuable for synthesizing evidence from multiple domains.

Superforecasters (Good Judgment Project): A group of seasoned forecasters (the “Superforecasters”) undertook a detailed assessment of COVID-19 origins in late 2023. Initially, they leaned towards the prevailing scientific view that a natural zoonosis was more likely (~73%) versus a lab leak (~25%). After reviewing new studies and evidence, they updated their probabilities slightly but remained in favor of a spillover. By February 2024, the Superforecasters’ aggregated judgment was roughly 74% chance of a zoonotic origin and 26% chance of a research-related lab leak. In other words, they judge the natural spillover about three times more likely than a lab accident. Notably, about 10 out of 54 forecasters dissented and actually viewed a lab accident as the most likely cause, illustrating that a substantial minority still found the lab hypothesis credible. The forecasters cited the lack of a confirmed animal host as a weakness in the zoonotic case, but they also noted the strong epidemiological market data and the historical base rate of zoonotic outbreaks favor the natural hypothesis. This careful, crowd-based analysis essentially aligns with the idea that while a lab leak is possible, the weight of evidence tilts toward a spillover, though with significant uncertainty.

Rationalist and Online Community Debates: On forums such as LessWrong and in the writings of bloggers like Scott Alexander (Astral Codex Ten), the COVID origin debate has been intensely discussed. These communities often use Bayesian reasoning and scrutinize details outside the constraints of official narratives. Two illuminating examples from an Astral Codex Ten-organized debate highlight how different independent analysts can see the same evidence and come to opposite conclusions:

In 2022, Rootclaim, an online platform that uses Bayesian analysis, boldly calculated a very high probability in favor of the lab-leak. After inputting numerous pieces of evidence, Rootclaim’s model concluded a ~90–94% probability that COVID-19 came from a lab. This was an extreme position, driven by factors like “prior probability” (arguing that a pandemic starting in Wuhan with its virology institutes is an extraordinary coincidence) and perceived low likelihood of the specific Wuhan market scenario. Rootclaim even offered large bets to defend their analysis.
Taking up that challenge, an independent researcher (Peter “tgof137”) bet $100,000 against Rootclaim’s pro-lab stance. After exhaustive research, he became over 90% convinced the lab-leak hypothesis was false, meaning he personally estimated >90% probability of a natural origin. He argued that an unbiased review of the data should strongly favor zoonosis – citing evidence like the two lineages and market clustering – and was willing to stake a significant sum on that conclusion. In his estimation, “a smart and unbiased person would vote for zoonosis with, say, 80% odds after seeing all the evidence”. The debate between these parties, judged by experts, ultimately found the zoonosis side’s arguments more convincing, at least to the judges, vindicating Peter’s stance.

Scott Alexander summarized many points from this debate on Astral Codex Ten, noting, for instance, how the unusual genetic features (like the furin site) and epidemiology can be plausibly explained by natural events. He highlighted the improbability of the “two strains at a market” coincidence under a lab-leak scenario and pointed out the hurdles one has to assume for a lab leak (e.g., WIV researchers somehow had a virus very similar to one (BANAL-52) that wasn’t discovered until later, inserted an unexpected cleavage site, cultured the virus in a way they’d never done before, and it leaked exactly when and where an outbreak would implicate a market). These observations were used to argue the lab-leak scenario requires a chain of unlikely events, whereas the spillover scenario, while also requiring coincidences, is more straightforward given the context of a live-animal market. At the same time, the rationalist community remains vigilant about not dismissing the lab-leak possibility; they emphasize how uncertainty and potential confirmation biases (in both directions) cloud the picture.

Perspectives of Notable Individuals: Some high-profile individuals have publicly stated probabilities or beliefs: for example, former CDC Director Robert Redfield told Congress in 2023 that he believes a lab origin is the most likely explanation (his personal view, not an official CDC position). Nobel laureate Dr. David Baltimore initially called certain genome features a “smoking gun” for engineering, then later walked back that language. Alina Chan (Broad Institute) has said she is ~50/50 but leans that a lab scenario hasn’t been given fair consideration. Meanwhile, many leading virologists like Dr. Ian Lipkin (who co-authored a paper favoring natural origin) have stuck with the zoonotic theory. Superforecasting tournaments and polls among experts have also been conducted – for instance, a 2021 informal poll of experts by Nature found ~18% thought lab-leak was possible or likely, while the rest leaned natural. These numbers have likely shifted somewhat as more intelligence info came out.

Contributions from Bayesian Analysis: Independently, a number of analyses have attempted a Bayesian calculation of the origin probability, explicitly factoring in prior probabilities and new evidence. One such attempt (Wuhan scientist Daoyu Zhang’s preprint, and another by statistician Jonathan Pekar et al.) tried to quantify how the case distribution in Wuhan points to the market. On the other side, data analyst Francois Balloux mused about how to incorporate the “prior” of an unprecedented lab pandemic. In essence, a Bayesian approach says: Prior to knowing anything, a novel outbreak is, say, 95% likely zoonotic (given history) and 5% lab accident. Then update that prior with evidence: The location in Wuhan (with a BSL-4 virology lab) tilts odds toward lab, the market clustering tilts heavily toward spillover, the lack of an animal found tilts back toward lab a bit, the raccoon dog DNA tilts back to zoonotic, the CCP cover-up behavior tilts toward lab, etc. The end result of such qualitative updating depends on how one weighs each factor. Many independent analysts end up somewhere in the middle – neither 95% nor 5%. For example, science writer (and self-described rationalist) Zeynep Tufekci wrote in 2022 that both scenarios are credible but she leans perhaps 60:40 in favor of zoonotic. Others, like intelligence expert skeptic Jeffrey Lewis, have said they lean slightly toward lab due to the coincidences but with low confidence.

Overall: Independent analyses show a spectrum of opinions: some highly analytically minded people conclude the evidence still favors a market spillover (often assigning ~70–90% probability to it), while others argue the conjunction of a novel outbreak in Wuhan and Chinese secrecy makes a lab leak more likely (sometimes giving >90% to lab). Aggregations like the Superforecasters tend to land in between these extremes, e.g. ~25% chance of lab leak. These exercises underscore the uncertainty – reasonable people with access to the same facts can come to different probabilistic conclusions. However, there is a common thread: virtually no independent expert claims 100% certainty for either side. All acknowledge gaps in evidence. Thus, these rational analyses help frame the debate as one of probabilities and weight of evidence, rather than black-and-white proof.

REFERENCES

V. Historical & Comparative Context

Looking at the broader history of pandemics and known lab incidents can inform how we assess COVID-19’s origin.

Past Pandemics and Outbreaks: Almost all major infectious disease outbreaks in modern history have been of natural origin. The 2002–2003 SARS outbreak was traced to horseshoe bats and intermediate civet cats in live markets; the MERS outbreak (2012) came from camels (ultimately from bats); HIV/AIDS crossed over from primates; Ebola outbreaks have been linked to bats or other wild animals. The 1918 flu pandemic was avian in origin, and the 2009 H1N1 “swine flu” pandemic was a reassortment in pigs. Zoonotic spillovers are the rule, not the exception. Indeed, whenever a new disease emerges, scientists’ working assumption (prior) is that it came from nature unless proven otherwise. Past outbreaks have attracted conspiracy theories too – e.g., claims that HIV was a lab creation or that Ebola escaped a lab – but extensive research has confirmed natural origins for each. It’s worth noting that early uncertainty is common: for SARS-1, it took a few months to find civets and bats harboring the virus; for Nipah virus in Malaysia (1998), it took time to realize fruit bats infecting farm pigs were to blame. In this context, a natural origin for SARS-CoV-2 would follow the typical pattern (especially given Southeast Asia’s history as a hotspot for emerging viruses).

However, COVID-19’s case has unusual aspects compared to typical zoonoses. One, the initial outbreak was in a metropolitan area in winter – whereas many zoonoses start in rural areas or involve animal handlers in closer contact with wildlife. Two, despite three years, no animal source has been pinned down, whereas for SARS-1, evidence of the virus in market animals was obtained relatively quickly. And third, unlike any past pandemic, here there is a virology lab studying the very kind of viruses that caused the pandemic essentially at ground zero. Those differences keep the lab hypothesis alive in a way that had to be dismissed outright for, say, Ebola (where no lab in West Africa was storing Ebola before 2014).

Laboratory Incidents: Lab accidents with pathogens are not hypothetical – they happen with unsettling regularity, though usually on a small scale. Notably, the SARS-1 virus escaped from laboratories multiple times after the 2003 outbreak. In 2004, lab workers in Beijing and in Singapore were accidentally infected with SARS due to safety lapses. Fortunately, those incidents were contained and did not spark new outbreaks, but they highlight that even well-known viruses can leak if protocols fail. In 1978, a researcher in the UK died from a smallpox lab accident, prompting major biosecurity reforms. In 2007, a leak from a high-security lab pipe in the UK caused an outbreak of foot-and-mouth disease in cattle. These cases prove that labs are not infallible – even in countries with strong safety cultures. In the case of China, there’s documentation of weaker safety practices in some labs. Chinese officials themselves acknowledged “complex and grave challenges in biosecurity” in late 2019, issuing new directives to labs (interestingly, a biosafety training was held at the WIV on November 19, 2019, perhaps indicating heightened concern at that time). The Wuhan Institute of Virology, while a world-class research center, had various reported issues (e.g., the head of the lab published articles about needing better safety and there were funding requests to improve wastewater treatment, etc., suggesting suboptimal infrastructure). Thus, the notion of a lab accident in Wuhan is entirely plausible from a safety standpoint – especially if an undetected mistake occurred in late 2019.

Precedent for a Lab-Caused Epidemic: It is often said that no novel pathogen outbreak has ever been proven to start from a lab leak. This is mostly true – with one significant exception: the 1977 H1N1 influenza outbreak (sometimes called “Russian flu”). In 1977, an H1N1 flu strain mysteriously appeared in China, Russia, and elsewhere, causing a global epidemic. This strain was nearly identical to an H1N1 that had circulated two decades prior and then disappeared, which was puzzling. Scientists now widely believe the 1977 flu outbreak was caused by a laboratory or vaccine trial accident – essentially, the old virus was preserved or re-created in a lab and got out. This event, while not a pandemic on the scale of COVID-19 (it mostly affected younger people since older individuals had immunity from the 1950s strain), shows that lab escapes can cause large outbreaks. It stands as the only known instance of a lab release leading to a substantial epidemic. COVID-19, if a lab leak, would be unprecedented in being a novel virus (not previously seen) causing a pandemic; 1977 was an old virus re-emerging. That caveat aside, 1977 demonstrates that lab-related pandemics are not impossible – just very rare.

Comparing Wet Markets vs. Labs: Wet markets (like Huanan) have historically been the source of several zoonotic diseases, including SARS-1 and avian influenza outbreaks. They involve many opportunities for viruses to jump from stressed, caged animals to humans in crowded conditions. Labs, especially ones handling high-risk pathogens, have had accidents, but seldom if ever have those led to community spread because typically the infections are identified and contained. So we have a situation where both culprit scenarios have precedent: the wet market fits a known pattern for coronavirus emergence, and a lab fits the pattern of rare but not impossible accidents with potentially high consequences. The probability of a random lab in the world leaking a virus that sparks a pandemic is extremely low, but when one specific lab (WIV) is known to be working on similar viruses and is located at the epicenter of an outbreak, that conditional probability rises.

Biased Historical Analogies: We should be careful with historical comparisons – COVID-19 might not neatly follow any prior script. For example, some argue “no lab leak has ever caused a pandemic, so this one couldn’t have,” which is a form of inductive reasoning that could fail if this is the first such incident. Conversely, saying “all past pandemics came from nature, so this one must have too” could overlook the unique circumstances here. We must weigh prior likelihoods (which favor zoonotic) against specific evidence in this case (some of which might favor lab). History gives us priors: maybe 90-95% of new diseases come from nature. But history also tells us labs handling dangerous pathogens are not 100% safe (accidents happen every decade or so).

Overall: Historically zoonotic spillovers are the far more common origin of outbreaks, and that sets a strong prior expectation for COVID-19’s origin. The COVID-19 pandemic would be the first of its kind to originate from a laboratory mishap if that hypothesis is true – making it an outlier event (though 1977’s case is a partial parallel). On the other hand, COVID-19 also has an outlier feature – erupting in a city hosting a major virology institute – which history has not seen before for a pandemic. Thus, history alone cannot resolve this; it can only guide our intuition (lean natural) and remind us to consider low-probability, high-impact scenarios (lab accidents) seriously.

REFERENCE: COVID-19 Lab Leak Theory — Wikipedia

VI. Critical Thinking: COVID — Lab Leak vs. Natural Zoonotic Origins

Some additional things I thought about: raccoon-dogs (intermediary proof or nah?), hypothetical most likely lab leak sequence, reasons for CCP coverup that weren’t lab leak related, China’s ability to disprove a lab leak, whether lack of proven intermediary = lab leak, why no conclusive evidence, Germany’s BND claim, etc.

A.) Raccoon Dog Evidence vs. “No Confirmed Intermediate Host”

a. What Is the Raccoon Dog Evidence?

Early 2023 analyses of swabs from the Huanan Seafood Market revealed SARS-CoV-2 genetic material mixed with significant amounts of raccoon dog DNA.
This strongly suggests raccoon dogs (known coronavirus hosts) were present in the immediate vicinity of the virus.
However, investigators never sampled live raccoon dogs during the critical window; by the time researchers arrived, the market was cleared.

b. Why We Still Say “No Confirmed Intermediate Host”

Despite the raccoon dog DNA clue, there is no direct finding of an animal infected prior to the earliest human cases.
A true “smoking gun” would be:
1. Identifying an animal population carrying the exact virus (or near-identical) before it spread widely in humans.
2. Showing a clear path from that infected animal population to the first human patients.
Since SARS-CoV-2 was not isolated from a definitive animal reservoir before widespread human transmission, the raccoon dog link remains highly suggestive but not definitive.

B.) If It Was a Lab Leak: Most Likely Sequence

Should a lab leak have occurred, the plausible chain runs as follows:

Researchers Collecting Viruses: Scientists from the Wuhan Institute of Virology (WIV) bring back bat coronavirus samples from caves or collaborate with other labs that do.
Laboratory Work: They culture or experiment on these viruses (possibly gain-of-function or serial passaging in cell lines/animals).
Accidental Infection: A lab worker is exposed—perhaps through poor biosafety protocols or during risky experiments—and unknowingly becomes Patient Zero (or one of the first).
Local Spread: The infected worker(s) transmits SARS-CoV-2 to family, colleagues, or community contacts.
Market Cluster as “Amplifier”: The Huanan Seafood Market could be a secondary superspreader site if infected individuals visited or worked there. Dense crowds turn a small outbreak into a large one, drawing attention to the market as an early “epicenter.”

Key Point: In this scenario, the market outbreak might overshadow smaller clusters that occurred earlier. Environmental swabs at the market could also find raccoon dog DNA alongside the virus if animals there were secondarily infected by humans (reverse zoonosis).

C.) China Blocking Lab Investigations: Alternative Explanations Beyond a True Lab Leak

China’s intense secrecy does not necessarily prove a lab leak. Other plausible motives include:

Avoiding Reputation Damage: They may fear that open scrutiny would expose inadequate biosafety or ethically contentious research (like gain-of-function)—embarrassing even if the virus did not originate there.
Authoritarian Reflex: Historically, the Chinese government tightly controls any narrative that may show mistakes or vulnerabilities (e.g., initial SARS-1 cover-ups).
Preventing Public Panic: A sensational “lab leak” rumor could fuel internal outrage or harm the CCP’s image of scientific leadership.
Zero Tolerance for External Audits: Inviting foreign investigators might set a precedent for international interference in Chinese internal affairs—something Beijing typically resists.

Overall: Blocking investigations aligns with a cover-up but can also reflect broad authoritarian tendencies to conceal embarrassing lapses—even if the cause is purely zoonotic.

D.) Could the CCP Definitively Prove a Non–Lab Origin?

In principle, yes—if truly zoonotic:

Full Access to WIV Records: Raw lab notebooks, internal databases of viral samples, staff health logs, biosafety reports, etc. If no near-relative virus to SARS-CoV-2 existed there before the outbreak, that would strongly undermine a lab-leak scenario.
Independent Audits: Let neutral international experts inspect freezers, computer logs, experiment protocols, camera footage, etc.
Undeniable Evidence of an Animal Reservoir: E.g., finding a farm or wild-animal population that tested positive before humans were widely infected.

Why Hasn’t This Happened?

China has not produced or permitted a transparent audit. They also haven’t shown a conclusive infected animal population that predated human cases.
This fuels lab-leak suspicions, but it doesn’t necessarily confirm it—authoritarian governments have many reasons (beyond hiding a lab accident) to keep data under wraps.

E.) No Confirmed Intermediate Host: Does That Push Toward Lab Leak?

Pro-Lab Argument

After years of searching—testing tens of thousands of animals—there’s still no direct animal source. This was not the case for SARS-1 or MERS, where civets and camels were identified relatively quickly.
Lab-leak proponents argue the failure to find a reservoir implies the virus might have come from stored lab samples.

Counterpoints

Sometimes it takes years (or never) to pinpoint the exact host, especially if the animals were culled or never sampled in time.
China’s patchy data-sharing or incomplete fieldwork could obscure a natural reservoir, even if it truly exists.
Historical parallel: Ebola virus origins took decades to link confidently to bats, and we still lack perfect clarity on every spillover event.

F.) Why Can’t We Identify Anything Definitively?

Chinese Data Control: The CCP heavily restricts raw epidemiological data, lab records, and even environmental swabs.
Early Cleanup: By the time investigators got to the Huanan market, it was disinfected and cleared of animals, erasing many potential clues.
Complex Wildlife Trade: If the virus spread through illicit or poorly tracked wildlife trafficking, the paper trail is scant or nonexistent.
Possible Lab Record Destruction: If it was a lab leak, relevant logs or samples could have been destroyed or never documented.

Overall: We face incomplete evidence due to secrecy, possible lost or destroyed data, and inherent complexities in tracing zoonotic outbreaks.

G.) Potential (or Likely) Intermediary Species if Natural COVID Spillover Occurred

Raccoon Dogs: Prime suspect based on recent DNA evidence in contaminated market stalls.
Civet Cats: Known culprit in SARS-1, commonly sold in Chinese markets.
Badgers, Mink, Bamboo Rats, Other Small Mammals: Also sold in markets, can harbor coronaviruses.
Multiple Species: It could be a chain (e.g., bat → civet → raccoon dog → human).

H.) Could It Have Been Bats Directly to Humans?

No Evidence of Bats at Huanan Market: Vendors did not typically sell bats there.
Possible “Upstream” Bat-Human Jump: People who handle or research bats might have been exposed in the field (e.g., cave sampling). However, the cluster of early market cases implies at least one infected mammal was physically present or was infecting humans in that environment.
Historical Note: SARS-1 eventually was traced to horseshoe bats but had an immediate amplifier (civets) sold in markets. Direct bat-to-human jumps do happen (e.g., Nipah), but an intermediate mammal is the more typical route.

I.) Best Logic on Origins & Probability

a. Bayesian Thinking

Historical Baseline: Most coronavirus pandemics emerged from wildlife (SARS-1, MERS). Lab leaks causing global pandemics are unprecedented (though lab escapes of pathogens have happened on smaller scales).
Market Evidence: Early infections cluster at Huanan; two lineages (A & B) imply multiple animal-human jumps. Raccoon dog DNA is consistent with an infected mammal source.
CCP Secrecy: Raises suspicion but not definitive proof of a lab leak; authoritarian regimes can hide any embarrassing data—whether it’s a lab accident or wildlife-trade failures.
Intelligence Splits: Agencies like the FBI or Germany’s BND see a plausible lab-leak scenario, but “moderate/low confidence” suggests incomplete evidence.

b. Typical Final Odds

Many experts arrive at roughly 65–70% chance of a natural spillover, 30–35% chance of a lab accident.
These are not exact figures but reflect the net weighting of:
1. Strong epidemiological evidence for the market.
2. Significant intelligence-driven suspicion of a lab mishap.
3. Persistent lack of a definitive “smoking gun” for either side.

J.) The German BND (Bundesnachrichtendienst) Evidence

Reported High Confidence: In 2025, leaks suggested the BND concluded there was an 80–95% probability of a lab origin.
Why the Discrepancy? The BND may have classified human or signals intelligence (e.g., intercepts, insider testimony) pointing to a WIV accident. Other agencies haven’t corroborated this publicly, so the data remain unverified.
Why Many Are Still Skeptical: The BND’s raw evidence isn’t fully public. Strong scientific research still leans toward a natural spillover, absent conclusive lab records. Intelligence can be misread or incomplete, and different agencies weigh evidence differently.

K.) If China Could Have Provided Proof, Why Haven’t They?

They Truly Don’t Have a Perfect “Smoking Gun”: They may never have located the exact infected animal reservoir, and the lab’s research data might not contain a near-relative virus.
Avoiding Blame for Wildlife Mismanagement: A confirmed zoonosis from illegal animal trade still implicates the CCP in failing to regulate wet markets (embarrassing post-SARS-1).
Covering Lab Safety Issues: Even if the lab didn’t cause COVID, an investigation could reveal other risky practices or unauthorized experiments.
Authoritarian Reflex: Reflexive secrecy is typical: Minimizing outside scrutiny helps maintain the narrative that the CCP is in control, regardless of the actual cause.

Key Point: Secrecy can be about broader political control and avoiding “loss of face,” rather than a definitive sign they’re hiding an actual lab leak.

L.) Do the CCP and China’s Leadership Know the True Origin?

They Know It Was a Lab Leak: If a catastrophic lab accident was discovered internally, they have overwhelming incentive to conceal it (avoid global outrage, domestic anger, loss of legitimacy). Their secrecy and destruction of records could reflect a direct cover-up.
They Know It Was a Zoonosis: Possibly they traced it to wildlife but keep quiet to avoid revealing negligence in policing animal trade or early outbreak mismanagement. Even a natural origin can be politically embarrassing (especially post-SARS-1).
They Themselves Aren’t 100% Sure: Authoritarian bureaucracies can be chaotic under crisis: local data may have been destroyed early on, or conflicting evidence never fully resolved. They could have partial or contradictory findings, plus a “best guess” but no definitive proof. Fearing an open audit (which might unearth other failings), they opt for total opacity.

Most Logical Middle Ground: The CCP likely has more information than outsiders, giving them a stronger internal theory—lab or wildlife.

That knowledge might be conclusive or might still have gaps.
Regardless, publicly admitting either cause opens the door to blame or political fallout. Thus, they maintain silence and restrict data access.

Critical Takeaways

Neither Hypothesis Is Completely Ruled Out: The market-focused evidence remains robust and historically consistent with previous coronavirus spillovers. The lab-leak scenario remains viable given the Wuhan Institute of Virology’s work, known biosafety concerns, and China’s secrecy.
Why Many Analysts Still Slightly Favor Zoonosis: Two distinct lineages in early Wuhan. Raccoon dog genetic traces where the virus was detected. Parallels to SARS-1 and MERS (both zoonotic). No publicly confirmed near-identical WIV virus prior to December 2019.
Why ~30–35% Assign a Lab Origin: Intelligence findings (FBI, BND) highlight suspicious lab conditions. Secrecy and inconsistent data from China. The coincidence of a top coronavirus lab in Wuhan.
China’s Secrecy Doesn’t Guarantee a Lab Leak: Authoritarian regimes often suppress anything that undermines the Party line, whether or not the lab was the true source.
Final Probability Range: Reasonable analysts might say ~65–70% natural spillover vs. ~30–35% lab leak, with low-to-moderate confidence. The BND’s reported 80–95% figure is unverified publicly. It shows the debate remains unresolved.
Do They “Know”? They likely have a better idea than the outside world. They may not have airtight proof themselves, depending on lost or never-recorded data. Political motivations and fear of blame could keep them silent, regardless of which origin they suspect.

VII. COVID Origin: Lab Leak vs. Wet Market: Probability Estimates & Confidence Levels

Bringing together all sources of evidence – intelligence findings, scientific data, behavior of actors, expert opinions, and historical context – we now assess the relative likelihood of the 2 origin hypotheses and provide a reasoned probability estimate for each. This is inherently a Bayesian reasoning exercise where we update our prior beliefs with the evidence at hand.

Prior Probability: Based on historical frequency, one would start with a strong prior in favor of a natural zoonotic origin (perhaps on the order of ~90% or more). Novel pathogens almost always come from nature. The prior for a lab leak causing a pandemic is very low (though not zero). However, this prior is significantly altered by one crucial fact: the outbreak occurred in Wuhan, where a lab researching coronaviruses is located. If the outbreak had started in, say, a random village or another city with no such lab, the lab-leak probability would be minuscule. But given Wuhan’s unique situation, our prior for a lab origin must be higher than for an arbitrary location. Some analysts have tried to quantify this: if only a handful of labs in the world were doing similar research, and by chance the outbreak emerged near one of them, that might bump the prior for lab from say 5% to something like 30% or more. We don’t have precise numbers, but qualitatively, Wuhan being the epicenter boosts the lab-leak hypothesis’ plausibility significantly above baseline.
Evidence Update – Intelligence: U.S. and allied intelligence do not have direct proof for either scenario, but notable is the trend of several agencies leaning lab-leak (albeit with low confidence). The BND even estimated up to 95% likelihood for a lab accident. While we should be cautious – these are assessments, not facts – it shows that those with access to classified info see enough smoke to suspect a fire. This pushes probability toward the lab side somewhat. However, the low confidence means the evidence (intercepts, informants, etc.) is not rock solid, so it’s not a decisive update, just an incremental one favoring lab. The fact that some intel agencies and many scientists still consider zoonosis plausible keeps that scenario firmly in play.
Evidence Update – Scientific: The scientific evidence tilts in favor of a natural origin: the strong clustering at the market, multiple early strains, and the raccoon dog DNA findings are difficult to reconcile with a lab leak without invoking a string of coincidences. This evidence significantly boosts the probability of zoonotic spillover. On the other hand, the failure to find an intermediate host despite extensive efforts weighs against the zoonotic hypothesis (though it by no means disproves it, as absence of evidence can be due to limited sampling or Chinese obstruction). The genomic analyses showing no clear sign of engineering neutralize early lab-engineering arguments, which means we treat “lab leak” as likely an accidental release of a naturally evolved virus – a scenario that doesn’t require any unnatural genome properties. So genomics neither strongly helps nor hurts either side now. Overall, I would say the epidemiological data (market case pattern) is one of the most powerful pieces of evidence, and it favors spillover. That likely keeps the natural origin as a leading hypothesis by evidence, despite the uncertainties.
Evidence Update – CCP Actions: China’s non-cooperation and secrecy is an indicator that the true origin might be embarrassing to them. This could mean a lab accident (very embarrassing) or could simply be covering up illicit wildlife trade and a bungled early response (also embarrassing, though perhaps to a lesser degree). The cover-up behavior increases the odds of a lab leak in our assessment, because a lab origin would almost guarantee an intense cover-up, whereas a natural origin might not have required such extreme data suppression (China could have, for example, quickly found and pointed to a source if it existed, to quell rumors – their failure to produce a culprit animal is telling). So the CCP’s actions nudge the probability needle toward the lab side. It’s hard to quantify, but many analysts have said that if China had nothing to hide, they wouldn’t act this way. While not definitive (China might hide data on principle or to avoid any blame), it is a notable factor.
Evidence Update – Independent Analysis: The fact that methodical analyses by third parties (super-forecasters, rationalists) don’t converge on one answer but cluster around, say, 70-30 or 60-40 either way indicates that both hypotheses have merit and neither is overwhelmingly confirmed. This meta-evidence – that informed people disagree – suggests a state of genuine uncertainty. It cautions us against assigning an extreme probability to either scenario. If the evidence strongly favored one side, we’d expect most independent analyses to land there. Instead, we see divisions, which implies roughly comparable arguments on each side. Thus, a reasonable overall probability split might be somewhere in the middle of those extremes.
Historical/Bayesian Perspective: Given the prior and the evidence updates above, we do an informal Bayesian synthesis. Start with perhaps 70% chance natural vs 30% lab (after considering Wuhan’s lab in the prior). Then add the market epidemiology data – that might boost natural to, say, 85%. Add raccoon dog evidence – maybe up to 90%. But then factor in the lack of host found and Chinese obstruction – that might drop natural back down to 70%. Intelligence hints and lab safety concerns might drop it a bit further to around 60%. These are subjective adjustments, but they illustrate how one might arrive at a posterior belief.

Taking all of this into account, the best-supported probability estimate based on current evidence is that a zoonotic spillover from animals (likely at the Huanan market) is slightly more likely than a lab leak, but not by a very large margin.

In numeric terms, one might put it at roughly ~60–70% probability for a natural origin and ~30–40% probability for a lab leak.

My assessment leans toward the zoonotic hypothesis being more probable – perhaps on the order of about 65% vs 35% in favor of zoonotic spillover. This corresponds to odds of roughly 2:1 that the virus emerged from a wildlife source rather than a laboratory accident.

It’s important to attach confidence levels to these estimates. Even a 65% vs 35% split indicates a lot of uncertainty. We would characterize our confidence in either hypothesis as low – meaning new evidence could easily sway the percentages significantly.

For instance, discovery of a SARS-CoV-2 precursor virus in a wild animal farm or cave would shoot the zoonotic probability up to near certainty. Conversely, leaked records of an accident or infected researchers at WIV would push the lab probability way up. Absent such smoking guns, we are left with imperfect information.

Therefore, a credible confidence interval on the lab-leak probability might be wide: perhaps between 20% on the low end (if one puts more stock in the market data) to as high as 50% on the upper end (if one emphasizes intelligence reports and Chinese secrecy). Our point estimate of ~35% reflects a holistic view weighing all the data/evidence.

Final Take: COVID-19’s Origin (2025)

After evaluating intelligence assessments, scientific evidence, Chinese government behavior, expert analyses, and historical precedent, we conclude that both the lab-leak and zoonotic spillover hypotheses remain plausible, with the balance of evidence favoring a natural spillover at the Wuhan market – but only slightly.

The early clustering of cases and genetic data from the market are strongly indicative of an animal origin, yet the coincidence of a major coronavirus lab in Wuhan and the Chinese government’s obfuscation suggest we cannot dismiss the lab-leak scenario.

In probabilistic terms, I’d estimate roughly a 2 in 3 chance that COVID-19 began as a zoonotic jump (likely from a wild mammal, possibly via the wildlife trade), and a 1 in 3 chance that it resulted from a research-related accident in Wuhan.

This estimate is made with low to moderate confidence, given the unresolved gaps in data.

Overall (March 2025): A zoonotic spillover is deemed more likely, but the lab leak hypothesis cannot be ruled out and carries a substantial probability that warrants continued investigation.

Obviously further research and transparency would be beneficial — including access to China’s early outbreak data and lab records… that is the only way to potentially reach a definitive conclusion.

Until then, the origin of COVID-19 will remain an open question marked by evidence-based reasoning, but also by the sobering recognition of how much we still don’t know.

ASAP Drew

Discussion about this post