Incommensurability, The Correspondence Principle, and the “Scientists Were Wrong Before” Gambit


One of the intrinsic features of the scientific process is that it leads to modifications to previously accepted knowledge over time. Those modifications come in many forms. They may involve simply tacking on new discoveries to an existing body of accepted knowledge without really contradicting prevailing theoretical frameworks. They may necessitate making subtle refinements or adjustments to existing theories to account for newer data. They may involve the reformulation of the way in which certain things are categorized within a particular field so that the groupings make more sense logically, and/or are more practical to use. In rare cases, scientific theories are replaced entirely and new data can even lead to an overhaul of the entire conceptual framework in terms of which work within a particular discipline is performed. In his famous book, The Structure of Scientific Revolutions, physicist, historian, and philosopher of science, Thomas Kuhn referred to such an event as a “paradigm shift.” [1],[2]. This tendency is a result of efforts to accommodate new information and cultivate as accurate a representation of the world as possible.

The “scientists have been wrong before” argument

However, sometimes opponents of one or more areas of mainstream science attempt to recast this self-correcting characteristic of science as a weakness rather than a strength. Anti-GMO activists, anti-vaxxers, young earth creationists, climate science contrarians, AIDS deniers and many other subscribers to unscientific viewpoints have used this as a talking point. The argument is essentially that the fact that scientists revise and sometimes even eliminate old ideas indicates that scientific knowledge is too unreliable to take seriously. They reframe the act of refinement over time as a form of waffling. Based on this, they conclude that whatever widely accepted scientific conclusions they don’t like should therefore be rejected.

Why the “Scientists Have Been Wrong Before” Gambit Exists

The main function of the “scientists have been wrong before” gambit is to serve as a post-hoc rationalization for embracing ideas that are neither empirically supportable nor rationally defensible, and/or rejecting ones that are. Pseudoscience proponents want to focus on perceived errors in science in order to downplay the successful track record of the scientific method. In doing so, they fail to account for the why and the how of scientific transitions. This is also ironic and hypocritical because pseudoscience has no track record worth speaking of at all. Scientific theories are updated when other scientists better meet their burden of proof, and when doing so serves the goal of better understanding the universe. In contrast, the aforementioned gambit is a self-serving attempt to side step the contrarian’s burden of proof in order to resist change.

The argument is disingenuous for a number of reasons, not least of which is that it ignores the ways in which scientific knowledge typically changes over time. Previous observations place constraints on the specific ways in which scientific explanations can change in response to newer evidence. Old facts don’t just magically go away. In order to serve their purpose, reformulations of scientific theories have to account for both old facts and the new. Otherwise, the change would not be an actual improvement on the older explanation, which presumably accounted for at least the older data, but not the newer.

Facts, Laws, and Theories

Before further unpacking this point, I should clarify my use of terminology: in this context, I’m essentially using the term fact to denote repeatedly observed data points. These are independent of the explanations proposed for their existence. Alternatively, one might say that facts report. Scientific Laws are essentially persistent data trends which specify a mathematically predictable relationship between two or more quantities. On the other hand, Scientific Theories are well-supported explanations for why some aspect of the natural world is the way it is and/or how exactly it works. They are consistent with the currently available evidence and make testable predictions that are corroborated by a substantial body of repeatable evidence. In short, facts and laws describe; theories explain.

For example, evolution is both a fact and a scientific theory. This because the fact that populations evolve and the modern scientific theory of evolution (which describes how it occurs) are separate but related concepts. Evolution is formally defined as a statistically significant change of allele frequency in a population over time. *An allele is just genetics jargon for a variant of a particular gene. That is descent with modification. It happens all the time. We witness it constantly. It’s not hypothetical. It’s not speculation. It’s an empirical fact.

The theory of evolution, on the other hand, is an elaborate explanatory framework which outlines how evolution occurs. This includes the mechanisms of natural selection, genetic drift, gene flow, mutation (and much more), and it makes many testable predictions about a wide range of biological phenomena. In science, a theory provides more information than facts or laws, because it connects them in ways that permit the generation of new knowledge. I’ll say it again: facts and laws describe; theories explain.

The Correspondence Principle

It’s true that scientific ideas can be wrong or incomplete and that scientific theories can change with new evidence. However, the argument that this justifies rejecting well-supported scientific theories just because one doesn’t like their conclusions ignores the constraints that prior experimental results place on the ways in which scientific knowledge can realistically change in the future. People advancing the Scientists have been wrong gambit are typically vague and imprecise in their usage of the term, “wrong.” It is often implied that wrong is being used in the sense of “totally factually wrong,” rather than merely incomplete, which is inconsistent both with scientific epistemology and with the history of science. It’s at odds with scientific epistemology, because knowledge in science is generally conceived of in a fallibilistic and/or probabilistic manner rather than in a binary one [12]. It’s at odds with the history of science because it is not generally the case that the data used to support a theoretical claim is entirely 180 degrees mistaken, but rather that the theory is being replaced by a more complete one which, in many cases, simply looks differently. Sure, theories can be expanded and the meaning and implications of experimental data can be conceptually reframed, but new theories can’t be in direct contradiction with the aspects of the old one whose predictions corresponded with experimental data. Unless it can be shown that all prior data consistent with the predictions of the older theory was either fraudulent or due to systematically faulty measurements, this is simply not a viable option.

Another way to put it is that old facts don’t go away so much as their explanations can change in light of newly discovered ones.

This is reflected in what is called the correspondence principle [8].

A Paraphrasing of Bohr’s conception of the Correspondence Principle

Although originally associated with Niels Bohr and the reconciliation of quantum theory with classical mechanics, it illustrates a concept which applies in all areas of science. Essentially, the correspondence principle says that any modifications made to classical mechanics in order to account for the behavior of matter in the microscopic and submicroscopic realms must agree with the repeatedly verified calculations of classical physics when extended to macroscopic scales [9]. However, the overarching concept of older (yet well-supported) scientific theories becoming limiting cases of newer and broader ones is inextricable from advancement of scientific knowledge more generally.

This is why there exist certain facts that will probably never be totally refuted, even if the theories which explain and account for them are subsequently refined and/or placed within the broader context of newer and more comprehensive explanatory frameworks. This is necessarily the case because any candidate for a new scientific theory which proves inferior to the old framework insofar as accounting for the empirical data would be a step backward (not forward) in terms of the degree to which our leading scientific theories map onto the real world phenomena they purport to represent.

As Isaac Asimov put it:

“John, when people thought the earth was flat, they were wrong. When people thought the earth was spherical, they were wrong. But if you think that thinking the earth is spherical is just as wrong as thinking the earth is flat, then your view is wronger than both of them put together” [16].

The Story of Gravity

Another one of my favorite examples of this is gravity. Our understanding of gravity has undergone multiple changes over the centuries, but none of those updates ever overturned the empirical observation that massive bodies reliably undergo an apparent acceleration towards other massive bodies in a mathematically predictable relationship. Aristotle was wrong about the mass of an object determining the rate at which it fell, and explained it in teleological terms, whereby certain objects were thought to have more “earth-like” properties, so that it was in their nature to belong on the ground [10]. But he didn’t dispute the basic observation that objects fell. Isaac Newton, who developed the inverse square law relationship for gravity, did not develop a theory for why matter behaved this way. He merely described it [11]. Rather than being satisfied with spooky action at a distance, prolific French physicist, astronomer, and mathematician, Pierre-Simone Marquis de Laplace conceptualized gravity in terms of classical field theory, whereby each point in space corresponded to a different value of a gravitational field, such that the field itself was thought of as the thing acting locally on a massive object [5].

The modern theory of gravity (Einstein’s General Relativity) explains it by positing a four dimensional space-time manifold capable of degrees of curvature surrounding massive bodies. In this theory, space-time tells matter how to move, and matter tells space-time how to curve [6]. Like the theory of evolution, general relativity has made many testable and falsifiable predictions that have come to fruition. Moreover, we know that GR cannot be the end of the story either, because the rest of the fundamental forces of physics are better described by quantum field theory (QFT), a formulation to which certain features of GR have notoriously not been amenable [7].

However, not one of these refinements contradicted the basic observations of massive bodies undergoing apparent accelerations in the presence of other massive bodies. Mathematically, it can be shown that Laplace’s formulation was consistent with Newton’s; the difference was in how it was conceptualized. Similarly, in situations involving relatively small masses and velocities, solving the Einstein Field Equations yields predictions that agree with Newton’s and Laplace’s out to several decimal places of precision. And although we don’t yet know for sure what form a successful reconciliation of GR and QFT will ultimately take, we know that it can’t directly contradict the successful predictions that GR and QFT have already made. This exemplifies the point that there exist constraints on the particular ways in which scientific theories can change.

Parsimony and Planetary Motion

I should note that concurrent to the progression of our scientific knowledge of gravity were changes in our understanding of planetary motion, because it demonstrates how the expansion of predictive power is not the only criterion governing theoretical transitions in science. More specifically, the Copernican model of the solar system didn’t actually produce calculations of superior predictive accuracy to the best Geocentric models of his time. Tycho Brahe’s formulation of Ptolemaic astronomy was more accurate. Although Brahe ultimately rejected Heliocentrism, Copernicus’s arguments intrigued him because the his model seemed less mathematically superfluous than the system of epicycles required to make Geocentrism work, yet it yielded results that were more or less in the same ballpark [13]. In other words, what stood out about Copernicus’s model was that, even though it wasn’t quite accurate, it accounted for a lot with a little. It was more parsimonious.

Many of the arguments against the Copernican model had more to do with Aristotelian physics than with the discrepancies in the resulting calculations, some of which were themselves a consequence of Copernicus’s assumption that orbits had to be circular, which was due in part to the philosophical notion that circles were the perfect shape. These problems were of course later resolved by the work of Johannes Kepler and Galileo Galilei; the former used Brahe’s own data to deduce that planets moved in elliptical orbits and swept out equal areas in equal times, whereas the latter formulated the law of inertia and overturned much of the Aristotelian physics upon which many arguments against the Copernican view were based [14]. In combination, Kepler and Galileo laid down much of the groundwork from which Isaac Newton would revolutionize science just a generation later.

The moral of the story, however, is that there are times when parsimony directs the trajectory of further scientific inquiry. It’s not always directed by expanding predictive power. A certain amount of theorizing in science involves what can essentially be understood as a form of data compression. Ultimately, the consistency of theory with empirical reality is the end game, but if a concept can explain more facts more simply and/or with fewer assumptions, then it may be preferred over its leading competitor. It’s certainly preferable to lists of disparate facts lacking any common underlying principles, because science isn’t just about describing empirical phenomena, but about discovering and understanding the rules by which they arise.

This touches on the principle of Occam’s Razor which, insofar as it applies to science, can be roughly paraphrased as the idea that one ought not to multiply theoretical entities beyond that which is needed in order to explain the data [15]. Putting it another way, the more ad hoc assumptions one’s hypothesis requires in order to work, the more likely it is that at least one of them is mistaken.

Or as Newton put it,

We are to admit no more causes of natural things than such as are both true and sufficient to explain their appearances. Therefore, to the same natural effects we must, as far as possible, assign the same causes” [11].


Occam’s Razor is not a rule in science so much as it is a heuristic that sometimes proves useful. Ultimately, our ideas must agree with nature’s results first and foremost. Deference to the empirical world is always paramount, and the universe is under no obligation to meet our arbitrary standards of simplicity or aesthetic preferences, but some prospective theories are better than others at compressing our understanding into more cogent sets of concepts.


In addition to introducing the idea of paradigm shifts in scientific advancement, Kuhn’s The Structure of Scientific Revolutions (TSoSR) also introduced the concept of incommensurability to describe the relationship between newer and older scientific paradigms. Initially, he introduced this as an umbrella term for any and all conceptual, observational, and/or methodological discrepancies between paradigms, as well as semantic differences in the use of specialized terminology. Kuhn’s own conception of incommensurability evolved considerably in the years following his publications of TSoSR, eventually restricting its applicability to problems with the translation of certain terminology common to both paradigms due to semantic differences arising from the transition to a new conceptual framework [3].

However, the basic idea was essentially that the methods, concepts, and modes of communication involved in disparate scientific paradigm are different enough that anyone from one paradigm attempting to communicate with someone from another would necessarily be speaking at cross-purposes, because they lack common measure. Even the observations themselves are thought to be too theory-laden for concepts and problems to be adequately translated across the theoretical boundaries of the pre and post phases of a scientific revolution. Kuhn himself even used the analogy from Gestalt psychology known as a Gestalt shift [4]. Here’s an example:

Hill, W. E. “My Wife and My Mother-in-Law.” Puck 16, 11, Nov. 1915

Do you see a young woman looking away, or an old woman looking down and to your left? Can you switch back and forth between perspectives? The meaning of any reference to the “nose” of the figure depends on whether one is speaking within the young woman or old woman paradigm. The placement and thickness of the lines does not change during gestalt shifts. What changes is the way in which their meaning is understood.

Analogously, the precise meaning of scientific statements depends on the theoretical framework in terms of which they are being made. The empirical facts that the theories seek to explain have not gone away (though newly obtained data may very well be forcing the change). What changes significantly is the way in which the meaning of the data is conceptualized, and the way in which new questions are framed.

Incommensurability as an attack on the scientific method

Some opportunists might seek to co-opt this notion of incommensurabilty to attack the epistemological integrity of the scientific process itself by exaggerating the degree to which new paradigms invalidate previous scientific knowledge, and to downplay their regions of predictive overlap. However, such attacks would necessarily be weakened by having to account for the constraints the correspondence principle places on which aspects of a scientific theory can change and/or be invalidated by a paradigm shift. To conflate a conceptual change in science with the invalidation of all facets of an older theory is to implicitly presuppose an anti-realist relationship between theory and the empirical phenomena to which it refers.

This is circular reasoning.

The unstated assumption is that no meaningful correspondence relationship exists between scientific concepts and the aspects of the empirical world they purport to represent, therefore changes in how terms are used and how problems are conceptualized precludes the preservation of facts and predictions an earlier model got right. As we saw in the earlier examples of the correspondence principle in action, this is demonstrably false. Many facts and predictions of older theories and paradigms are necessarily carried over to and/or modified to be incorporated into newer ones.

Concluding Summary

Scientific knowledge changes over time, but it does so in the net direction of increasing accuracy. This is one of the strengths of the scientific method: not one of its weaknesses. Most attempts to reframe this as a weakness (invariably via the use of specious mental acrobatics) ignore the constraints necessarily placed on the ways in which scientific theories can change or be wrong.

Many important revolutions in science involve conceptual changes which do not contradict all of the facts and predictions of the older theory, but rather reframe them, restrict them to limiting cases, or expand them to more general ones.

The preservation of certain facts and predictions which are carried over from older theories to newer ones (because the older ones also got them right) can be understood in terms of the correspondence principle.

The validity of the concept of incommensurability between temporally adjacent scientific paradigms is restricted to terminological, conceptual, and sometimes methodological differences between pre and post scientific revolution phases, but does not in any way contradict the correspondence principle.

The fact that scientific ideas can be wrong in principle does not mean that the particular ones the contrarian using this gambit dislikes will be among the discarded, nor that the ways in which it could conceivably be wrong could vindicate the contrarian’s desired conclusion.

Consequently, citing the observation that “scientists have been wrong before” is never a rationally defensible basis with which to justify rejection of scientific ideas which are currently well-supported by the weight of the evidence; only bringing new evidence of comparable quality can do that. If the contrarian is not currently in the process of gathering and publishing the evidence that would supposedly revolutionize some area of science, then they are placing their bet on an underdog based on faith in a future outcome over which they have no influence, and for which they have no rational basis for expecting. This is no more reasonable than believing one is going to win the lottery based on the observation that other people have won the lottery before, and then not even bothering to buy a ticket.  

You don’t know what aspects of our current knowledge will turn out to be incorrect, nor which will be preserved. That’s why the maximally rational position is always to calibrate one’s position to the weight of currently available scientific evidence, and then simply leave room for change in the event that newer evidence arises which justifies doing so.


[1] Kuhn, T. S., & Hawkins, D. (1963). The structure of scientific revolutions. American Journal of Physics31(7), 554-555.

[2] Bird, A. (2004). Thomas Retrieved 4 January 2018, from

[3] Sankey, H. (1993). Kuhn’s changing concept of incommensurability. The British Journal for the Philosophy of Science44(4), 759-774.

[4] What Impact Did Gestalt Psychology Have?. (2018). Verywell. Retrieved 4 January 2018, from

[5] Laplace, P. S. A Treatise in Celestial Mechanics, Vol. IV, Book X, Chapter VII (1805), translated by N. Bowditch (Chelsea, New York, 1966).

[6] Astronomy, S. (2017). Einstein’s Theory of General Retrieved 4 January 2018, from

[7] relativity?, A. (2018). A list of inconveniences between quantum mechanics and (general) relativity? Retrieved 4 January 2018, from

[8] Bokulich, A. (2010). Bohr’s Correspondence Retrieved 4 January 2018, from

[9] Bokulich, P., & Bokulich, A. (2005). Niels Bohr’s generalization of classical mechanics. Foundations of Physics35(3), 347-371.

[10] Pedersen, O. (1993). Early physics and astronomy: A historical introduction. CUP Archive.

[11] Newton, I. (1999). The Principia: mathematical principles of natural philosophy. Univ of California Press.

[12] Fallibilism – By Branch / Doctrine – The Basics of Philosophy. (2018). Retrieved 5 January 2018, from

[13] Blair, A. (1990). Tycho Brahe’s critique of Copernicus and the Copernican system. Journal of the History of Ideas51(3), 355-377.

[14] Copernicus, Brahe & Kepler. (2018). Retrieved 5 January 2018, from

[15] What is Occam’s Razor?. (2018). Retrieved 5 January 2018, from

[16] Asimov, I. (1989). The relativity of wrong. The Skeptical Inquirer14(1), 35-44.


Scientific Consensus isn’t a “Part” of the Scientific Method: it’s a Consequence of it

Although conceptually simple, the term “scientific consensus” is often misused and misunderstood. It can get confused with appeals to popular opinion or erroneously conflated with “consensus” in the colloquial sense of the word. These misunderstandings can lead to things like opinion polls, often predominated by unqualified individuals, being misconstrued as evidence that no scientific consensus exists on some topic for which it clearly does, or that it leans towards a different conclusion than it actually does. In some cases, the very concept itself invokes resentment or even retaliatory commentary from people whose views are threatened by its implications. The purpose of this article is to clarify the concept that the term scientific consensus is meant to refer to and address some of the arguments commonly leveled against it.

Defining Scientific Consensus

Just as the term “theory” has a different meaning in science than its colloquial usage, the term scientific consensus means something different than “consensus” in the usual colloquial sense. The latter typically refers to a popular opinion, and needn’t necessarily be based on knowledge or evidence. On the other hand, a scientific consensus is, by definition, an evidence-based consensus. A convergence of the weight of existing evidence is a prerequisite which distinguishes a knowledge-based scientific consensus from mere agreement. This is critical, because the scientific enterprise is essentially a meritocracy. As a result, it doesn’t matter if a few contrarians on the fringe disagree with the conclusions unless they can marshal up evidential justification of comparable weight or explain the existing data better. The weight of the evidence is paramount.

In a nutshell, a consensus in science refers to a convergence of many independent lines of high quality evidence all leading the majority of active scientists in a given field to arrive at the same conclusion and/or complimentary conclusions. It’s not something any scientist necessarily sets out to become a part of as a goal, but is rather something they discover they’re a member of because that’s where their research results led them. The process by which scientific consensus emerges over time can be complicated and tends to vary from case to case, but it is likely to exist whenever scientific knowledge is this best explanation for a given consensus. Furthermore, scientific knowledge is the best explanation for a consensus when the following definitional criteria are satisfied:

Consilience of Evidence: The consensus should be based on varied lines of evidence which independently converge on the same conclusion or set of conclusions [1]. The scientists and their results needn’t necessarily agree on every single minute detail, and the data convergence will typically fall within a set of error bars, but will point to the same general conclusion even if debates still exist on the minutia. This may involve contributions from multiple scientific sub-specialties, each providing different pieces comprising a broader understanding or set of conclusions.

For instance, the scientific consensus in climate science incorporates evidence and expert knowledge from meteorology, geology, geophysics, geochemistry, atmospheric physics, atmospheric chemistry, community and global ecology, astronomy, planetary science, and even stellar astrophysics. Scientists from different specialties study different aspects of the issue and arrive at results comprising a piece of a puzzle whose results are all consistent with the conclusion that the recent warming trend has been largely the result of human activities [38],[39],[40],[41],[42].

Similarly, the scientific consensus on the individual and social benefits of vaccination combines knowledge and evidence from fields such as microbiology, immunology, virology, epidemiology, systems biology, molecular biology, biochemistry and more. Knowledge comes together from these disparate disciplines to create vaccines that significantly reduce the likelihood of their recipients contracting the diseases against which they are designed to protect, whose risks are greater than the minuscule risks of adverse reactions to the vaccines [36],[37],[43].

Social Calibration: The experts involved are mutually committed to employing the same high standards of evidence and formalisms, and have good justifications for those standards [1]. Nobody disputes that carefully collected and reproducible evidence is key in science; the problem is that evidence doesn’t talk. It has to be interpreted by human scientists. The Social Calibration criterion has to do with what the scientific community as a whole accepts as evidence, how they decide what is relevant and significant, and how individual scientists persuade their peers that they are correct.

One of the reasons that certain fringe disciplines are viewed as pseudosciences by mainstream scientists is because they operate under lower and/or inconsistent standards. For example, one of the most important methods of ascertaining the safety and efficacy of a given medical intervention in conventional science-based medicine is the performance of a large randomized double blinded placebo-controlled clinical trial [34]. In contrast, many so-called “alternative” modalities are satisfied to rely on a modality’s ancientness (whether real or merely assumed), weak or non-replicable studies, and/or unverifiable personal testimonies that may or may not reflect how most patients would be affected [32],[33]. In some cases, alternative practitioners persist even after a substantial clinical evidence directly contradicts the premises underlying the modality, such as is the case with homeopathy [47]. That’s not to say that there do not exist certain exceptions, but the overarching pattern is that the standards of evidence agreed upon by mainstream medical researchers is different than the standards deemed acceptable in alt med. The agreed upon standards of evidence in scientific medicine represents what is being referred to here as social calibration. A counter-example would be something like ghost hunting, whereby there do not exist any consistent standards insofar as what is supposed to qualify as evidence for a ghost [48]

Social Diversity: This criterion simply means that the evidence and analyses comprising the scientific consensus should come from varied sources by scientists of varied backgrounds and cultures in order to avoid any systematic bias in the scientific literature [1]. For example, one of the arguments against the international scientific consensus on genetically engineered food safety is based on the misconception that seed companies like Monsanto are the only ones doing the research, or that they dictate who does. That’s not actually true, but if it were, then the evidence would be falling short on this criterion. Instead, the GE food consensus is supported by myriad studies by scientists from different ethnic, cultural and economic backgrounds with varied funding sources from all around the world, and by position statements from the most prestigious scientific organizations on the planet [27],[28],[29],[30]. The Social Diversity criterion ensures that a consensus is not a product of group think, politics, financial incentives, ideological motives, or shared cultural values.

Via John Garrett of Skeptical Science for Denial 101

Scientific Consensus =/= Unanimity

Notice that the aforementioned criteria defining scientific consensus does not preclude normative contestation or outlier viewpoints within the scientific community. That’s actually quite normal and usually fairly benign. Total absolute 100% unanimity among experts is not a prerequisite to a consilience of evidence supporting a particular conclusion.

Not All Disagreements are of Equal Merit

However, there are cases in which that normative contestation and areas of scientific uncertainty becomes exaggerated by groups with ulterior (unscientific) motivations (either financial, political, or ideological) in order to argue against the reliability of extant scientific knowledge and/or to obfuscate public understanding of it. This phenomenon of special interest groups combating scientific conensus is well-documented on topics ranging from the risks of smoking tobacco, anthropogenic global warming, and the safety of GMO foods and conventional vaccine schedules [2],[3],[4],[5],[6],[7],[8],[9],[10],[11],[12],[13],[14],[15],[16],[17],[19].

With a few exceptions, however, the contrarians in these cases are attempting to shed doubt on the existence, strength, and/or legitimacy of the scientific consensus on a particular topic, and are not necessarily directly contesting the very concept of scientific consensus itself. Rather than claiming that there can be no such thing as scientific consensus, most of them instead argue that such a situation does not exist with respect to the particular topic on which they disagree, or that it may exist but is nevertheless simply unfounded. They insert false balance and exploit the normal everyday uncertainties and tentativeness inherent to all scientific knowledge and attempt to amplify them with respect to topics they wish to portray as more contentious than the evidence actually suggests. This is effective because acceptance that a scientific consensus exists has been shown to function as a “gateway belief” to accepting that a set of propositions is true, and manufacturing the appearance of continued legitimate scientific controversy can obscure public perception of its existence [20],[21].

However, there are some exceptions whereby contrarians attempt to undermine the very concept of scientific consensus itself (i.e. here). As you may have guessed, the claimants in such cases invariably misrepresent what scientific consensus actually is. They try to portray it as analogous to arriving at a conclusion by way of an opinion poll, which is an example of the fallacy of equivocation [18]. They will typically argue you that what matters is the scientific evidence, which, although true, ignores the fact that a concilience of evidence is already a non-negotiable prerequisite to scientific consensus [1]. It is therefore nonsensical to speak of evidence and scientific consensus as if the latter was not contingent upon the former, let alone to imply that they are mutually exclusive. Again, this is no more legitimate than arguing against scientific theories by equivocating to a colloquial definition of theory.

Based on the criteria described earlier (consilience of evidence, social calibration, and social diversity), it’s easy to see how scientific consensus will unavoidably emerge on any question for which repeated applications of the scientific method by a diverse group of scientists result in a body of evidence whose results lean heavily towards certain conclusions and away from others. Although it is possible to quantitatively analyze the nodes and connections of various scientific citation networks and how they evolve over time with respect to a given topic, the specific pathway by which scientific consensus emerges tends to vary from case to case.

This paper entitled The Temporal Structure of Scientific Consensus Formation (by Shwed et al) explains some methods by which such scientific citation networks can be analyzed to ascertain degrees of scientific consensus and to partition nodes into salient sub-communities [2]. There are both pros and cons to this approach, but such efforts are designed to minimize dependence on the discretion of individual scientists in the detection of scientific consensus. When a new topic of study first arises, different scientists typically end up in different camps which approach key questions differently, explore different initial hunches, and form different citation networks as more and more studies are produced. As a scientific consensus begins to form, the lines of distinction between the various camps begin to dissolve, and members of each start to converge on certain areas of agreement.

Somewhat counter-intuitively, the authors also discovered that when scientific consensus is achieved, the aforementioned scientific citation networks tend to grow in size, as does the total rate of literature output on the topic. At least, that was the case in the instances they analyzed. The relevance of this observation is that any consensus arrived at on the basis of weak or faulty evidence will tend to quickly dissolve as interest in (and scrutiny towards) the topic increases, whereas conclusions based on stronger evidence will tend to open up follow-up questions whose study results are complimentary to them. This is another key to understanding why scientific consensus represents not the death of scientific inquiry, but rather a scaffolding on which subsequent scientific inquiry can build and grow. It simply represents what we’ve learned so far about some aspect of the universe. In this way, scientific consensus is not so much a final point of arrival but rather a launching point for further inquiry to add to and refine our current knowledge. The authors also identified three trajectories along which scientific consensus emerges, which they characterize as “flat, spiral, and cyclic,” but I’ll leave that for readers compelled to read the original paper [2].

If the scientific consensus is wrong on some topic, then the subsequent exploration of additional questions derived from it should reveal that. There’s a reason why we refer to science as a self-correcting enterprise. It’s not just a catch phrase. Efforts to better understand the universe must build upon existing knowledge. Scientific consensus helps shape the discussion and guide resource allocation with respect to tangential and/or follow-up questions within a particular sub-discipline. Without it, we would simply keep spinning our wheels by re-establishing the same conclusions over and over again without ever attempting to build on that foundation and generate new knowledge.

Why Scientific Consensus Matters

Although often colored by personal values, biases, competing motives, and desires, humans generally make decisions based on what they perceive to be true. This is true both on the individual and group levels. Not everyone can be an expert in a scientific discipline, and nobody can be an expert in all of them. Consequently, people routinely have to assess what is likely to be true in areas on which they are not experts, and make decisions based on it. Scientific Consensus represents the most reliably accurate knowledge available to human beings on a given topic at any given time. It’s far from infallible, but then again, so is every other epistemological framework available to humans (albeit even more so). To reject it on the grounds that it is not infallible in favor of even less reliable approaches to knowledge would be an example of the Nirvana Fallacy [31]. The best available option, even if imperfect, is nevertheless still the best available option. Scientific consensus also serves as a launching point guiding further scientific study of related questions, and helps facilitate the generation and accumulation of new knowledge.

Detecting Scientific Consensus

I think it’s a fairly safe assumption that analyzing scientific citation networks with sophisticated algorithms and statistical methods is not something that the average person is likely to do with respect to every scientific claim they stumble upon. It’s not the be all end all anyway, because it only shows how sub-communities of scientists and their published work converges over time. It does nothing to adjudicate on the quality of individual studies within a citation network or the reliability of their conclusions. Nor does it distinguish between cases where a cited work’s findings are being used as supporting evidence, versus cases where a cited work’s conclusions are being challenged. It’s useful, but it’s not a replacement for actual human experts capable of summarizing the state of affairs in their fields of specialization. Ultimately, becoming an expert in a particular field would be the ideal way to equip oneself to competently assess the current state of the science within it, but that’s not feasible for most people, and nobody is an expert on every topic.

Systematic Reviews as Proxies

Fortunately, however, there are other proxies one can look for to get a ballpark idea of the degree of scientific consensus (or lack thereof) on a particular topic. For example, on many thoroughly studied topics there exist systematic reviews and/or meta-analyses which examine many studies at a time in order to assess what is implied by the weight of the evidence. A systematic review seeks to answer a specific research question by summarizing all the available scientific literature fitting a pre-specified set of eligibility criteria; a meta-analysis seeks to use statistical methods to summarize and analyze the results of such studies. Systematic reviews can vary widely in quality just like other types of studies. You can get an idea of what a good systematic review should entail and how to read one here [22],[23],[35],[44],[45].

Position Statements as Proxies

Alternatively, many reputable scientific organizations will put together position statements on certain topics, which can be a useful proxy for ascertaining the degree of scientific consensus that exists for a given topic. If the majority of prestigious organizations have arrived at similar conclusions, then the chances are that there is a fairly strong scientific consensus on the subject. Obviously this is an imperfect proxy, because there are also front organizations which masquerade as objective scientific organizations, but which are really serving some other agenda, and because it doesn’t provide a clear view of the evidence upon which their conclusions are based.

Other Proxies

It’s advisable to avoid relying too heavily on petitions or surveys of scientists’ opinions as a proxy for or against the existence of a scientific consensus, particularly on topics that tend to be controversial in public discourse. It’s not that they can’t ever be done in such a way that they could convey useful information, but rather that they’re too easy to screw up, or to be manipulated into conveying misleading information. In fact, that’s a common tactic used by people whose goal it is to obfuscate public understanding by disputing the existence of a scientific consensus on particular topics where it exists. They accomplish this by cobbling together signatures and/or statements from people whose views comport with theirs, but whose qualifications are often tangential to the topic under discussion, and/or whose opinions represent a tiny minority of researchers, and are not well-supported by the weight of the evidence in the peer-reviewed literature.

For example, the debunked Oregon Petition Project was an attempt to obscure the weight of the scientific consensus on human-caused climate change [24]. A document assembled by the Discovery Institute which boasted of 100+ scientists who reject the theory of evolution was humorously rebutted by the National Center for Science Education with Project Steve: a list comprised exclusively of scientists named Steve who accept evolution, which nevertheless dwarfed the Discovery Institute’s list [25]. Similarly, anti-GMO campaigners have written things such as the I-SIS letter as an attempt to sew uncertainty and doubt on the mainstream scientific consensus position on the safety of Genetically Engineered food crops [26],[27]. HIV/AIDS denialists have also attempted similar tactics [46].

One possible exception to this rule of thumb would be a survey which groups the participating scientists according to the degree to which their area of specialty pertains to the subject under discussion so that one can see whether the percentage of agreement increases the closer the areas of expertise get to the specific topic. Even then it would have to be based on a representative sample of each sub-category of scientists, and I wouldn’t recommended relying on it as anything more than a complimentary proxy with which to cross-corroborate with other signs of an extant scientific consensus.

Above all, avoid relying on unsourced YouTube conspiracy videos, opinionated people with no relevant scientific education, blogs and other websites that make sensational claims for which they don’t cite credible research, and fake experts whose claims are totally inconsistent with the peer-reviewed scientific literature.

It’s perfectly fine to use a video medium to learn about science, but just understand that there is no vetting process whatsoever, so content creators can say whatever they want with impunity. University lectures are usually fine (and recommended), as are tutorials videos such as Khan Academy, and any videos which cite credible sources in their video description. This should go without saying, but I’m including it for the sake of completeness.


Scientific consensus is not a part of the scientific method so much as it is a consequence of it. It inevitably arises whenever a large body of scientific literature accumulates that points towards similar conclusions. Typically, people who argue otherwise are equivocating due to them either misunderstanding or deliberately misrepresenting the meaning of the term. Scientific consensus is characterized by the co-existence of a consilience of evidence, social calibration, and social diversity, and although not infallible, nevertheless represents the best knowledge currently available on a given scientific question at a given time. Furthermore, it is instrumental to the generation and accumulation of new knowledge in that it directs researchers toward complimentary follow-up questions whose exploration allows humankind to build upon previous knowledge. 


[1] Miller, B. (2013). When is consensus knowledge based? Distinguishing shared knowledge from mere agreement. Synthese190(7), 1293-1316.

[2] Shwed, U., & Bearman, P. S. (2010). The temporal structure of scientific consensus formation. American sociological review75(6), 817-840.

[3] McCright, A. M., & Dunlap, R. E. (2000). Challenging global warming as a social problem: An analysis of the conservative movement’s counter-claims. Social problems47(4), 499-522.

[4] Proctor, R. N. (2012). The history of the discovery of the cigarette–lung cancer link: evidentiary traditions, corporate denial, global toll. Tobacco Control21(2), 87-91.

[5] Lopipero, P. A., & Bero, L. A. (2006). Tobacco interests or the public interest: 20 years of industry strategies to undermine airline smoking restrictions. Tobacco Control15(4), 323.

[6] SCHURMAN, R. (2004). Fighting Frankenfoods: Industry opportunity structures and the efficacy of the anti-biotech movement in Western Europe. Social problems51(2), 243-268.

[7] Wales, C., & Mythen, G. (2002). Risky discourses: the politics of GM foods. Environmental Politics11(2), 121-144.

[8] Anti-GMO Activists Are the One Practicing Tobacco Science. (2015). Food and Farm Discussion Lab. Retrieved 4 August 2017, from

[9] Boykoffa, M. T., & Boykoffb, J. M. (2004). Balance as bias: global warming and the US prestige press$. Global Environmental Change14, 125-136.

[10] Dunlap, R. E., & McCright, A. M. (2008). A Widening Gap: Republican and Democratic Views on Climate Change.

[11] McCright, A. M., & Dunlap, R. E. (2011). THE POLITICIZATION OF CLIMATE CHANGE AND POLARIZATION IN THE AMERICAN PUBLIC’S VIEWS OF GLOBAL WARMING, 2001–2010tsq_1198 155.. 194. The Sociological Quarterly52, 155-194.

[12] Gary Ruskin, GMO Labeling Movement Funded by Anti-Vaxxers | American Council on Science and Health. (2017). Retrieved 4 August 2017, from Attach quote

[13] The Anti-Vaccine And Anti-GMO Movements Are Inextricably Linked And Cause Preventable Suffering. (2017). Retrieved 4 August 2017, from

[14] SCHICK, S. F., & GLANTZ, S. A. (2007). Old ways, new means: tobacco industry funding of academic and private sector scientists since the Master Settlement Agreement. Tobacco control16(3), 157-164.

[15] Kata, A. (2012). Anti-vaccine activists, Web 2.0, and the postmodern paradigm–an overview of tactics and tropes used online by the anti-vaccination movement. Vaccine30(25), 3778.

[16] Offit, P. A. (2015). Deadly choices: How the anti-vaccine movement threatens us all. Basic Books (AZ).

[17] Jolley, D., & Douglas, K. M. (2014). The effects of anti-vaccine conspiracy theories on vaccination intentions. PloS one9(2), e89177.

[18] Equivocation. (2017). Retrieved 5 August 2017, from

[19] Something About Pots and Kettles. (2015). The Skeptical Beard. Retrieved 5 August 2017, from

[20] van der Linden, S. L., Leiserowitz, A. A., Feinberg, G. D., & Maibach, E. W. (2015). The scientific consensus on climate change as a gateway belief: Experimental evidence. PloS one10(2), e0118489.

[21] Lewandowsky, S., Gignac, G. E., & Vaughan, S. (2013). The pivotal role of perceived scientific consensus in acceptance of science. Nature Climate Change3(4), 399.

[22] Systematic reviews and meta-analyses: a step-by-step guide | (2017). Retrieved 7 August 2017, from

[23] Uman, L. S. (2011). Systematic Reviews and Meta-Analyses. Journal of the Canadian Academy of Child and Adolescent Psychiatry20(1), 57–59.

[24] 30,000 Scientists Reject Anthropogenic Climate Change?. (2016). Retrieved 7 August 2017, from

[25] Project Steve | NCSE. (2017). Retrieved 7 August 2017, from

[26] Scientists Declare No Consensus on GMO Safety. (2017). Retrieved 7 August 2017, from

[27] Hulk, C. (2015). The International Scientific Consensus On Genetically Engineered Food SafetyThe Credible Hulk. Retrieved 7 August 2017, from

[28] Sanchez, M. A. (2015). Conflict of interests and evidence base for GM crops food/feed safety research. Nature biotechnology33(2), 135-137.

[29] (GENERA), G. (2014). Source shows half of GMO research is independent | Ag Retrieved 7 August 2017, from

[30] (2017). Retrieved 7 August 2017, from

[31] Man, F. (2016). The nirvana fallacy: An imperfect solution is often better than no solutionThe Logic of Science. Retrieved 8 August 2017, from

[32] We Should Abandon the Concept of “Alternative Medicine”. (2015). Science-Based Medicine. Retrieved 8 August 2017, from

[33] Hunt, K., & Ernst, E. (2009). Evidence-based practice in British complementary and alternative medicine: double standards?. Journal of health services research & policy14(4), 219-223.

[34] Sibbald, B., & Roland, M. (1998). Understanding controlled trials. Why are randomised controlled trials important?. BMJ: British Medical Journal316(7126), 201.

[35] Greenhalgh, T. (1997). Papers that summarise other papers (systematic reviews and meta-analyses). BMJ: British Medical Journal315(7109), 672.

[36] Graphic proof that vaccines work (with sources) – Isabella B. – Medium. (2015). Medium. Retrieved 8 August 2017, from

[37] Vaccines work. Period.. (2013). Science-Based Medicine. Retrieved 8 August 2017, from

[38] Hulk, C. (2017). No, Solar Variations Can’t Account for the Current Global Warming Trend. Here’s Why:The Credible Hulk. Retrieved 8 August 2017, from

[39] Cook, J., Oreskes, N., Doran, P. T., Anderegg, W. R., Verheggen, B., Maibach, E. W., … & Nuccitelli, D. (2016). Consensus on consensus: a synthesis of consensus estimates on human-caused global warming. Environmental Research Letters11(4), 048002.

[40] Oreskes, N. (2004). The scientific consensus on climate change. Science306(5702), 1686-1686.

[41] Doran, P. T., & Zimmerman, M. K. (2009). Examining the scientific consensus on climate change. Eos, Transactions American Geophysical Union90(3), 22-23.

[42] →, V. (2015). Is There a Consensus about Climate Change?Thoughtscapism. Retrieved 8 August 2017, from

[43] Comparison of Effects of Diseases and Vaccines – Canadian Immunization Guide. (2012). Retrieved 8 August 2017, from

[44] Shea, B. J., Grimshaw, J. M., Wells, G. A., Boers, M., Andersson, N., Hamel, C., … Bouter, L. M. (2007). Development of AMSTAR: a measurement tool to assess the methodological quality of systematic reviews. BMC Medical Research Methodology7, 10.

[45] Liberati, A., Altman, D. G., Tetzlaff, J., Mulrow, C., Gøtzsche, P. C., Ioannidis, J. P., … & Moher, D. (2009). The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS medicine6(7), e1000100.

[46] Schüklenk, U. (2004). Professional responsibilities of biomedical scientists in public discourse. Journal of medical ethics30(1), 53-60.

[47] Hulk, C. (2015). Money For Nothing: Why Homeopathy Is Still Pseudoscientific Nonsense That Does Not WorkThe Credible Hulk. Retrieved 13 August 2017, from

[48] Ghost-Hunting Mistakes: Science and Pseudoscience in Ghost Investigations – CSI. (2016). Retrieved 13 August 2017, from


Science has been wrong before, therefore I can make up whatever bullshit I want.

Some of the historical instances which typically get characterized as “science having been wrong” can be better understood as incomplete theories/models being conceptually re-framed in order to account for both the facts explained by the prior theory/model, as well as whatever (more recently acquired facts) made the modification necessary. In such cases, the “meaning” of the known facts gets viewed through a new and better lens. When it’s done successfully, the new conceptual framework also makes additional predictions that, if accurate, can expand human knowledge beyond merely summarizing the phenomena already known at the time of its inception.

One of my favorite historical examples of this is the transition from Newton’s Universal Gravitation to Einstein’s General Theory of relativity. The fact that apples can be readily observed moving from tree branches to the ground wasn’t overturned by General Relativity because Newton wasn’t wrong about that. It was a fact long before Newton, and remains a fact today. What changed was the way in which such occurrences were conceptualized and explained.

Newton didn’t have a comprehensive theory, but he had an equation, m*d2r/dt2 = G*M*m/r^2, which seemed to imply non-local action at a distance. General Relativity on the other hand, which involved considerably busier mathematics (involving tensors and differential geometry), conceived of the falling apple as moving naturally along a geodesic path on a 4 dimensional spacetime manifold which had been curved by the presence of a large mass (Earth in this case). It also made other predictions that people may never have thought of otherwise  and which expanded our knowledge (i.e. gravitational lensing for example)., the he point being that it didn’t so much refute the basic facts so much as it explained them in a new way and explained other facts that the Newtonian model didn’t account for.

Similarly, we know that General Relativity cannot be the end of the story because it produces absurd results at the outer limits of its predictive power (such as the prediction of actual singularities), and it doesn’t play well with another extremely well-tested theory, quantum mechanics (and quantum field theory). We do know that any candidate for replacing it will have to (at minimum) account for the facts that the current theories do predict and account for, or else it would be a downgrade rather than an improvement.

This type of reiterative evolutionary process represents the scientific process at its best, and recognizing that can go a long way in distinguishing between varying degrees of relative truths (as described in Asimov’s famous “the relativity of wrong” essay), and assigning fair and reasonable degrees of confidence to various aspects of our current scientific understanding of a given subject.

Asimov famously summarized it thusly:

“John, when people thought the earth was flat, they were wrong. When people thought the earth was spherical, they were wrong. But if you think that thinking the earth is spherical is just as wrong as thinking the earth is flat, then your view is wronger than both of them put together.”

Although improving public awareness of this may not be sufficient in and of itself to ameliorate the problem of public science denial, since such attitudes are often fueled by ulterior motivations rather than by lack of knowledge alone, it is nevertheless important insofar as understanding the way in which cumulative knowledge progresses in science, because understanding that makes it clear why nihilistic dismissal of science on the grounds that it can be wrong is an untenable and irrational position.

Yet, we see those sorts of rationalizations constantly by people who reject evolution, climate science, vaccine science, science-based medicine and the science of genetically engineered foods.

So to summarize, the take away message here is twofold:

1) The phenomenon of science correcting mistakes is one of its strengths: not one of its weaknesses.


2) Just because newer theories or models incorporate and explain additional facts and information doesn’t mean that the facts explained by the old theories and models aren’t true, or that the new ones don’t have to account for them. Often times the improved theory or model is inclusive of the old one, but merely explains the facts in a different way and/or accounts for additional information that the old one didn’t cover. – Credible Hulk unnamed