The Origin of COVID-19: Lessons from the Intelligence Community’s Assessment

Understanding how COVID-19 emerged is of great importance, and the US Intelligence Community (IC) has produced some of the most important assessments of this topic. Much of the recent coverage of the Intelligence Community’s assessment of the origin of COVID-19 has focused on which agencies hold which view. But the IC’s report has a lot more to offer on the subject of the origin of COVID-19. Also, the process used by the IC in its assessment offers insights for how to analyze complex, high-stakes issues. In the business world, as in the world of intelligence, important decisions sometimes require dealing with murky evidence and—since we’re human—the tendency to engage in confirmation bias: interpreting information in a way that confirms or supports our prior beliefs or values. The techniques the IC used to produce the COVID-19 assessment are designed to help analysts think critically and avoid cognitive bias.

Scope Note: The Intelligence Community’s “Updated Assessment on COVID-19 Origins” was released by the Office of the Director of National Intelligence (ODNI) on 29 October 2021. This assessment was reportedly updated in early 2023. As of the date of this writing, the 2023 assessment has not been publicly released, so I will refer mainly to the 2021 assessment, as amended by what has been learned from news reporting about the 2023 assessment. 

The Intelligence Community’s Process

Reflecting the serious and controversial nature of the topic, the managers of the IC assessment took extraordinary care to reduce the chance of cognitive bias and prompt analysts to think critically and objectively. Normal analytical practice is for teams of analysts in each agency to collect information, organize the information, make informed judgments about the evidence, and undergo a quality control process of internal review by peers and by managers before publication. The managers of the IC’s assessment on COVID-19 went considerably beyond these normal procedures by adding a variety of structured analytic techniques. 

1. The National Intelligence Council, which managed the assessment, held a two-day exercise on Analysis of Competing Hypothesis (ACH), during which analysts determined whether existing reporting was consistent or inconsistent with various hypotheses about the origin of COVID-19. This exercise allowed analysts to determine that most reporting was consistent with two hypotheses, and the reporting that was inconsistent was deemed to be not credible.

• ACH emphasizes identifying alternative hypotheses and how they might be disproved, which prompts analysts to think seriously about evidence, explanations, and outcomes that had not previously occurred to them. The process discourages the analyst from choosing one “likely” hypothesis and using evidence to prove its accuracy. Cognitive bias is minimized when all possible hypotheses are considered. 

2. The assessment’s managers hosted an IC-wide Team A/Team B exercise, using a debate-style format to explore how the IC could strengthen either hypothesis. In a Team A/Team B exercise, independent groups are pitted against each other in a competitive analysis. Agencies pulled from these conversations to sharpen their positions. 

3. Key information gaps were identified. These core intelligence gaps were intended to guide future collection. They also forced analysts to understand and articulate the key gaps in their knowledge. 

4. Analysts explicitly stated a standard of proof: what evidence they would require to provide a more definitive explanation about the virus’s origin. The standard of proof articulated in the IC’s assessment was as follows: “The IC judges they will be unable to provide a more definitive explanation for the origin of COVID-19 unless new information allows them to determine the specific pathway for initial natural contact with an animal or to determine that a laboratory in Wuhan was handling SARS-CoV-2 or a close progenitor virus before COVID-19 emerged.” 

5. Analysts expressed their key judgments together with a level of confidence: low, moderate, or high. The Intelligence Community learned from previous failures–such as the National Intelligence Estimate in 2002, Iraq’s Continuing Programs for Weapons of Mass Destruction–that it is crucial to be explicit about the confidence with which key judgments are stated. It is one thing to tell the President “We believe X is most likely,” and another to say “We believe X is most likely with low confidence.” 

6. The draft assessment underwent four rounds of outside review. The process of consulting outsiders helped the drafters improve the explanation of technical and intelligence jargon, and helped sharpen the presentation.    

Key Findings

The resulting IC assessment included several noteworthy themes:

The paucity of evidence and the tentativeness of the resulting judgments. Given Beijing’s lack of cooperation and the lack of diagnostic or conclusive information, the Intelligence Community remains divided on the most likely origin of COVID-19, and most agencies are unable to make higher than low confidence assessments. All agencies assess that two hypotheses are plausible: natural exposure to an infected animal and a laboratory-related incident. The below chart summarizes the various hypotheses the IC considered, and how IC agencies assess their likelihood. 

According to the assessment, variations in analytic views largely stem from differences in how agencies weighed intelligence reporting, scientific publications, and intelligence and scientific gaps. Seven of eight IC elements participating in the assessment said they cannot reach a conclusion or their conclusions were reached with “low confidence.” The Intelligence Community’s definition of confidence levels is:

• Low Confidence generally means that the information is scant, questionable, or very fragmented, so it is difficult to make solid analytic inferences; it could also mean that the IC has significant concerns about or problems with the sources.
• Moderate Confidence generally means that the information could be interpreted in various ways, that the IC has alternative views, or that the information is credible and plausible but not corroborated sufficiently to justify a higher level of confidence.
• High Confidence generally indicates that the IC’s judgments are based on high-quality information or that the nature of the issue makes it possible to develop a solid judgment.

The strength of the IC’s judgment that China’s officials were unaware of the virus before it was isolated in early January 2020.

The assessment cites the lack of foreknowledge on the part of Chinese officials in Beijing and Wuhan as an “area of broad agreement.” When analysts outlined the arguments for COVID-19 arising from a natural zoonotic spillover, they chose to list “lack of foreknowledge” as their first point. Even analysts who favor the laboratory-related incident hypothesis explained how an incident could have occurred without Chinese authorities knowing about it. They noted that an incident could have avoided detection if it involved only a small number of researchers who did not acknowledge or did not know they had an infection.

According to the assessment:

“The IC assesses China’s officials probably did not have foreknowledge that SARS-CoV-2 existed before Wuhan Institute of Virology (WIV) researchers isolated it after public recognition of the virus in the general population. Accordingly, if the pandemic originated from a laboratory-associated incident, they probably were unaware in the initial months that such an incident had occurred. 

Early in the pandemic, the WIV identified that a new virus was responsible for the outbreak in Wuhan. It is therefore assessed that WIV researchers pivoted to COVID-19-related work to address the outbreak and characterize the virus. These activities suggest that WIV personnel were unaware of the existence of SARS-CoV-2 until the outbreak was underway.”

The degree of foreknowledge is important because some researchers claim that certain events that occurred in fall 2019 indicate that Chinese officials knew about, and were reacting at that time, to a laboratory incident. 

• In early March 2023, former CDC Director Robert Redfield said that various events in the WIV in fall 2019–especially letting a contractor re-do the ventilation system in the Wuhan laboratory—led him to conclude that “There is strong evidence there was a significant event in that laboratory in September 2019.” https://oversight.house.gov/release/covid-origins-hearing-wrap-up-facts-science-evidence-point-to-a-wuhan-lab-leak%EF%BF%BC/

• A report released in October 2022 by the minority (Republican) staff of a Senate Committee concluded that “the COVID-19 pandemic was, more likely than not, the result of a research-related incident.” The report is broad ranging, but the authors focused on events that suggest that Chinese authorities knew about a lab event in fall 2019. The authors highlighted, for example, the WIV taking down its online database of viral sequences; procurement and patents to upgrade ventilation equipment at the WIV; the speed with which Chinese scientists developed a vaccine; and a 19 November 2019 special training session run by a senior Chinese Academy of Sciences biosafety/biosecurity official who relayed “important oral and written instructions” from PRC leadership in Beijing to the WIV regarding the “complex and grave situation facing [bio]security work.” https://www.help.senate.gov/imo/media/doc/report_an_analysis_of_the_origins_of_covid-19_102722.pdf

The IC has not publicly addressed the more-recent circumstantial evidence cited by Dr. Redfield and others.

But the IC’s 2021 assessment judged that then-existing claims of abnormal activity at the WIV in fall 2019 were not supported and did not offer diagnostic insight. The assessment emphasized that the consensus view of the Intelligence Community is that Chinese officials in Beijing and WIV probably first learned about the existence of SARS-CoV-2 after it was isolated, which occurred in early January 2020. If the Intelligence Community is correct in their “broad consensus” that Chinese officials did not learn about SARS-CoV-2 until 2020, then circumstantial evidence of activities that occurred in fall 2019 is interesting, but moot; it is not evidence of Chinese officials reacting to knowledge that a laboratory incident had occurred. The discrepancy between the IC’s assessment and other researchers’ findings needs to be resolved. 

The IC assessment also included a number of factors that analysts believe are not diagnostic of either hypothesis:

After more than three years, an animal source for a spillover event has not been determined. Some researchers say that a specific pathway should have been found by now if the COVID-19 pandemic was the result of zoonotic spillover. They note that in the SARS outbreak of 2002-2004, public health officials in China found SARS infections in palm civets and raccoon dogs within six months of the first known human case. But the IC’s assessment makes clear that “in many previous zoonotic outbreaks, the identification of animal sources has taken years, and in some cases, animal sources have not been identified.” It took 14 years for scientists to identify that horseshoe bats were the reservoir species for the SARS epidemic. We still do not know the origins of the 2014 Ebola outbreak.

WIV illnesses in fall 2019.  The IC assessed that information indicating that several WIV researchers reported symptoms consistent with COVID-19 in autumn 2019 is not diagnostic of the pandemic’s origins. Even if the researchers’ hospital admission could be confirmed, this alone is not diagnostic of COVID-19 infection.

WIV researchers had collected a virus called RaTG13 that is about 96% identical to SARS-CoV-2.  The IC’s report says that the WIV collected RaTG13 from a bat in 2013, but the report also says that RaTG13 “… is widely believed to not be a direct ancestor of SARS-CoV-2.”

The refusal of Chinese officials to cooperate in the investigation of the origin of COVID-19. The assessment says: “Beijing’s lack of cooperation on origins [is] not diagnostic of either hypothesis.”

It is clear that more information will be needed to determine the origin of COVID-19, and that China’s cooperation will be key. The Intelligence Community deserves credit for accurately explaining the quality of evidence and the standard of evidence they used for their conclusions, when they have been under immense pressure to side with one theory or another. In my view, the IC is correct: we should be patient and wait for evidence either that there was an animal host for SARS-CoV-2 or that the virus was at the Wuhan Institute of Virology before the outbreak occurred.

Appendix: The Intelligence’s Community’s arguments, in a nutshell

The Case for the Natural Origin Hypothesis

• China’s officials lack of foreknowledge. WIV’s activities in early 2020 related to SARS-CoV-2 are a strong indicator that the WIV lacked foreknowledge of the virus. 
• The precedent of past novel infectious disease outbreaks having zoonotic origins.
• The wide diversity of animals that are susceptible to SARS-CoV-2 infection, and the range of scenarios—to include animal trafficking, farming, sale, and rescue—in China that would enable zoonotic transmission. Historically lax government regulation—and even promotion—of these activities increase the probability that initial transmission occurred along one of these routes. 
• Wuhan markets sold live animals and dozens of species—including raccoon dogs, masked palm civets, and a variety of other mammals, birds, and reptiles—often in poor conditions where viruses can jump among species, facilitating recombination events and the acquisition of novel mutations. 
• It is less likely that a laboratory worker was inadvertently infected while collecting animal specimens than if an infection occurred through numerous hunters, farmers, merchants, and others who have frequent, natural contact with animals. 
• China’s poor infectious disease surveillance system would not have been able to detect the SARS-CoV-2 exposure as quickly as a suspected exposure in a laboratory setting.

The Case for the Laboratory-Associated Incident Hypothesis

• Work on coronaviruses is inherently risky with numerous opportunities for researchers to unwittingly become infected with SARS-CoV-2.
• Academic articles authored by Wuhan Institute of Virology (WIV) employees indicate that WIV scientists conducted research on other coronaviruses under inadequate biosafety conditions that could have led to opportunities for an incident. 
• Academic literature indicates that WIV researchers conducted research with bat coronaviruses or collected samples from species that are known to carry close relatives of SARS-CoV-2.
• Information suggests researchers in China used biosafety practices that increased the risk of exposure to viruses. Academic publications suggest that WIV researchers did not use adequate biosafety precautions at least some of the time. 
• The initial recorded COVID-19 clusters occurred only in Wuhan, and WIV researchers who conducted sampling activity throughout China provided a node for the virus to enter the city.  
• China’s investigations into the pandemic’s origin might not uncover evidence of a laboratory-associated incident if it involved only a small number of researchers who did not acknowledge or have knowledge of a potential infection.