بسم الله الرحمن الرحيم

Before dealing with the philosophical conundrums that the HCM exposes its adherents to, we should deal with the oft-invoked refrain of “common sense probability and plausibility.”

As discussed in the previous post, hard sciences and social sciences validate theses based on experiments or, when experimentation is not possible, statistical analysis of historical data. HCM proponents often argue that their inferences about history are based on probability.

Quasi-Probabilistic Approach

When an HCM researcher looks at available evidence, they then attempt to draw a conclusion as to the most probable and plausible explanation – without relying on unjustified external explanations from religion or tradition. Probabilities are invoked as common-sense, it is simply intuitive to utilize naturalistic explanations based on available data and lead to the most “plausible” inference. These inferences are asserted to best explain the data, being closest to the truth. According to Mukharji, Aroop, and Richard Zeckhauser,

“Not all historical events are difficult to explain, of course. Physical events, say a fire, can be diagnosed after the fact. Their source, say a faulty electrical wire or a dropped cigarette, can often be found. Events that occur multiple times are easier to assess. Often, indeed, analysts do employ probabilities to describe such happenings.”[i] ["Bound to Happen: Explanation Bias in Historical Analysis." Faculty Research Working Paper Series, Harvard Kennedy School, Oct. 2019, updated Nov. 2019, RWP19-032. Working paper]

Certainly, these kinds of small inferences are necessary. However, do they translate to the scale and scope that HCM researchers engage? HCM researchers invoke probability in such a way as to vastly oversimplify the complexities of history.

It should be noted that there is a categorical demarcation in method here between faith-based history and naturalistic history. Followers of Abrahamic faiths believe that observed history was guided by God intentionally. The Qur'an often encourages believers to reflect on the meaning of past events. However, from a naturalistic paradigm, we must assume that all historical events have no inherent meaning imbibed from on high and that they are the product of random forces. Therefore, the following critique only applies to historiographies that assume history has no guiding force and is a product of random natural forces.

The forward procession of events into the future is not linear. There are potentially infinite possibilities at any time, with a similar degree of possible outcomes. Quantifying every possible future scenario is analytically impossible. Yet, when observing the past, only one historical outcome is observed in reality. The exact reason for this outcome can never be known, yet HCM scholars will attribute inferences and probabilities to these outcomes.

Plausible Interpretations

Plausibility, probability, and science are highly severely conflated in discourses around the HCM. I have demonstrated in the previous part that the HCM is neither a hard science nor a social science, but most akin to an interpretative science.

A common misconception is that plausibility and truth are correlated or even causal. However, plausibility has nothing to do with actual truth, it simply describes whether something subjectively seems reasonable. The plausibility fallacy, as articulated by Dr. Mitchell Handelsman and Nassim Taleb, is a logical error of assuming that the more plausible an explanation is, the more likely it is to be true. Oftentimes, the most implausible explanations are true.

Additionally, crafting plausible theories does not beget knowledge. To many people historically, the most plausible explanation as to the shape of the earth was their experience of it; that it was flat. The more implausible explanation, that the Earth is round, was determined initially by trigonometry. Scientists such as al-Biruni performed calculations, looking at the angle of the horizon from atop a mountain. These are a reproducible methods with clearly defined parameters and variables. HCM scholars claim that the more plausible an explanation, the more likely it is to be true. However, plausibility is a measure of perceived coherence and intelligibility, not truth. Merriam-Webster,

superficially fair, reasonable, or valuable but often deceptively so

My issue is that even the provisional conclusions, especially the ones around which there is more confidence in the field, are unreliable. The problem with plausibility is not that plausibility is evaluated. As you said, it's somewhat inevitable. The problem is that plausibility is one of the main, if not the main, measure of evaluation. It is wholly unreliable for that. Any field that uses plausibility to evaluate hypothesis or theories will be fundamentally detached from reality. This is what traditional Chinese medicine or Marxist theory or literary analysis does.

Is there value in TCM? Of course. Does Marxist theory lead to fascinating insights? Yes. Is literary analysis important? Definitely. Do any of them independently lead to reliable conclusions that approach objectivity? Absolutely not.

If plausibility is the mechanism by which we evaluate theories then we will necessarily undervalue less-plausible theories (which may be objectively true) until new evidence makes them more plausible. This may never occur, especially dealing with settled history. Meaning we can develop a historiography with "scientific" fetters that claims any thesis can be challenged with evidence – while actually creating interpretative dogma that is not aligned with reality. Therefore, enterprises that are based on creating plausible theories must have other mechanisms to arrive at truth. Mukharji, Aroop, and Richard Zeckhauser,

“First, many historians write works that are permeated with deeply-held theoretical assumptions about how the world works. A person who believes that market capitalism is a primary determinant of human behavior, will probably be more likely to think market capitalism was a primary determinant of any particular historical event. The same might go for rational actor theory, racial theories, gender theories, or any number of other theories of behavior that appeal to particular historians. But it is implausible that a single theory of human interaction could explain the multiple events that a single historian might address in a career. Some events should be explained by some reasons in a hierarchy of causation, and others by a different set in a different hierarchy. We would thus expect explanation bias to more acutely afflict more ideologically driven individuals.”

As will be demonstrated in later posts, the HCM is latent with the assumptions of modern secularism. These assumptions color all interpretations. Therefore, some objective mechanism must be required so as to parse "plausibility" from actuality. When advocates of the HCM claim it is “probabilistic", they do not describe how they have set down the parameters & variables to achieve reproducibility. This leaves them highly subject to imposing their deeply held views about the world, informed by a secular framework; then calling it objectivity.

It may seem intuitive to cross compare evidence, investigate plausibility, and arrive at the best "common sense" solutions. However, this is itself a "plausible" method and is not rigorous.

It should be noted that this is a problem of historiography and not of history. And it is a problem of secular empirical historiography, not religious historiography that believes history is not random. Documentation of history as simply collecting & retelling of facts is not the same as historiography.

Least Deserving Outcomes

As we observe modern historical events proceeding in real time, oftentimes outcomes deemed the least probable or implausible occur. This is despite the consensus of experts and vehement certitude of scholars. History regularly surprises us going into the future, yet we assume the generator of historical events in the past is linear enough to dissect. Very few pundits or analysts have any kind of remarkable track record with predicting the future, and the few that do fall within the statistical variance of the survivorship bias. Simply put, there is no Nostradamus.

From an empirical perspective, historical events are the product of random sequences of events without a deterministic mechanism that we know of. This does not mean that human actors do not have intentions or goals, or that there are no discernible behaviors. It simply means that there are no decipherable consistent historiographical laws, unlike physics.

Yet, the HCM method purports to be able to look at these events and “infer” the most probable conclusions as to why events occurred or how religious texts were produced. The method of attributing meaning to these data points by inference and rough "probability" occurs without creating sample paths for invisible unrealized histories.

This is a fatal flaw. Since pure randomness and coincidence are not accounted for by calculating other possible permutations, it is never clear whether the inferred explanations are simply a byproduct of historical noise or indicate something substantial. HCM academics do not investigate sequences of unrealized scenarios over time, meaning that HCM historiographers are only ever dealing with a control group. This fails to account for the fact that the most correct explanation for observed data or for history may be the most opaque and least intelligible. Mukharji, Aroop, and Richard Zeckhauser,

“A third source of explanation bias could relate to discomfort from the cognitive dissonance encountered when giving fair hearing to competing theories. Humans generally have a hard time juggling competing hypotheses. A single explanation, especially one explained narratively, can be preferable precisely because individuals have an easier time understanding and being persuaded by it. While this could help account for explanation bias among producers of history (of any kind), it may better explain the strength of explanation bias's adoption among its consumers. Consumers of history who have less knowledge than its purveyors may also be less aware of competing theories and, therefore, even more susceptible to explanation bias.”

Back-Testing

HCM scholars frame the method as probabilistically back testing historical data, unfettered by religion or tradition. This is intended to be a claim of objectivity and accuracy. Back-testing itself is not a foolproof method. While it may provide valuable information in some fields, like cosmology, we also see the abject failure of back-testing in other disciplines. Applications of back-testing on certain historical medicine trials demonstrate that some trials succeed simply due to randomness. There are documented cases of miraculous recoveries from cancer which can be arbitrarily pieced together to form a nonexistent pattern by back testing. Back-testing regularly produces stock market trading algorithms that can be projected to produce tens of millions of dollars, only to go negative within hours of trading in real world scenarios.

The English mathematician Karl Pearson discusses the reference case problem, that true randomness is unattainable because we experience only one sample history. Therefore, we are unable to create proper control and variable groups on past back-testing. This calls into question methods of empirical research into history.

To reiterate, this is a non-issue for those who believe God designs our single observed history with intention and calls upon us to draw lessons from the past. Muslims do not believe history is the outcome of random naturalistic forces, but rather a selected chosen outcome of God that He intends for us to draw inferences from. However, it is a huge issue when the assumption is that history is naturalistic, random, with no guiding force. Life is nonlinear. This is observed in all domains of empirical inquiry. If we take a sandcastle and continue adding a single grain of sand in a linear fashion until it collapses, the collapse will be non-linear. In fact, it is practically impossible to predict where any particular grain of sand will end. Oftentimes the final outcomes of historical sequences or events, or the explanations between events, are the ones that are least path-dependent. They are the “least deserving” from a human perspective, occurring due to extraneous network externalities (see the Matthew Effect or the Gambler’s Fallacy). The results of future random processes are often unaffected by past results. This calls into question any empirical attempt to derive historical rules.

Additionally, the current success of empirical sciences also does not independently indicate truth. The famous Charles Dawkins fallacy, "It's true because it works" fails to take into account the problem of induction (which will be discussed later). Stock traders often have a successful market strategy for many years that produces billions of dollars and is therefore "true" due it's perceived efficacy until their fund explodes overnight. This problem is severely compounded when no statistical analysis or experimentation can be done. Then we cannot even know if a theory is true within a short time window, and nothing is falsifiable.

Complexity in History

The three-body problem and wider complexity theory illustrate this kind of quantitative nonlinearity. If we have two balls in a pool table, we can model their exact trajectories with astonishing accuracy if they are hit at any particular angle by a billiard cue. However, the moment a third ball is added, the complexity of potential outcomes becomes incalculable. Why a particular ball will end up where is so statistically difficult to compute that it is beyond the computation power humans currently possess. Yet, when looking at past data from an empirical lens, Historical-Critical scholars attempt to justify and explain why or how certain events are correlated.

A fascinating example of this is Qur’anic parallels with Judeo-Christian traditions at the time of Muhammad’s mission. Despite a dearth of direct causal evidence, the standard position of the HCM field is that the Qur’an is a byproduct of the intellectual, cultural, and religious milieu of seventh century Arabia. This conclusion is informed by parallels in mythology, Targums, Homilies, romances, poetry, the Torah, the Talmud, Christian writings, and other sources. Yet the logical leap between apparent correlation, which is unquantifiable empirically, to causality is totally unexplainable given the extant evidence. These kinds of conclusions would require computational capabilities trillions of times more powerful than computing the location of a third billiard ball upon being hit by a cue; something currently impossible.

Normal Distributions

These quasi-probabilistic claims of HCM scholars fail to account for typical statistical measures like significance, the normal range of fluctuation in a range, and data distributions. If a range of data has normal range of fluctuation, anything within that range is insignificant. In fact, it would be difficult for an HCM scholar to prove that much of the field cannot distinguish itself from statistical noise. HCM’s conclusions often rely on distributions with well-defined moments (e.g. finite mean, finite variance). However, reality rarely models according to a normal distribution. Randomness is often based on power-law or fat tail distributions. Especially in the realms of social science.

The assumption that something would be extremely abnormal or even impossible today in terms of people is less reliable in history than when it comes to physics because human social dynamics are more exposed to fat-tail distributions. This is further compounded when we are more critical about the science we are putting out about human behavior. The problem stems from reliance on models that assume stability in systems that are too complex to be accounted for in such a fashion. HCM starts by assuming that empirical reality contains a typical or an average that we can then analogize from, and draw explanations according to that baseline “reality”. This is assuming that these matters are distributed with "well-defined moments (e.g. finite mean, finite variance) which is misleading and inaccurate.

Social sciences and even history deal with non-ergodic systems. They don't convene to predictable averages overtime in any demonstrable fashion. They are described more accurately by models that consider wild swings or extreme singular events, which makes uniformitarianism problematic. It isn't just that studies have a problem of reproducibility, these methods are too fragile to draw inferences from.

Power-Law Randomness

The problem with well-defined movement distributions (e.g. Gaussian or bell curve) is that you can't model some kind of randomness with it. The problem is that real world phenomenon – especially pertaining to historical events – actually follow power-law distributions where there are extreme exceptions that are sparsely distributed. These are called “fat-tails” and these fat-tails dominate the course of history. Extreme rare events have an outsize impact and a normal Gaussian distribution would fail to account for that even though they are the story. The kind of assumption of normality you're suggesting here is problematic because it underestimates extreme deviations.

The problem is that HCM researchers assume a type of Gaussian randomness when history is defined moreso by power-law randomness. In non-linear, complex systems (such as markets, economies, societies, human history as a whole), local observations and empirical consistency do not necessarily generalize on a grand scale or project backwards. That's why in science theories are tested – and even that leads us to efficacy not truth. We can't run backwards experiments on history so our conclusions are much more fraught. An empirical approach to history should indicate that things for which there are no evidence should not be considered. However, things that have never happened before (for which there is no prior evidence) occur routinely in history. As will be discussed later when critiquing naturalism, outliers are important because of the problem of induction. Since there is no repeatability in a closed environment for history, secular empirical methods of explaining history cannot result in expertise.

[i] Mukharji, Aroop, and Richard Zeckhauser. "Bound to Happen: Explanation Bias in Historical Analysis." Faculty Research Working Paper Series, Harvard Kennedy School, Oct. 2019, updated Nov. 2019, RWP19-032. Working paper

Are HCM Inferences Bayesian?

Admitting for a degree of uncertainty doesn't make the reasoning Bayesian. According to Daniel Kahneman Bayesian reasoning has two pillars,

"Anchor your judgment of the probability of an outcome on a plausible base rate. Question the diagnosticity of your evidence."

Where do you find either in any HCM work?

Faith based historiography typically assumes that the single observed history was intentionally chosen or guided by an All-Knowing All-Powerful God. Therefore, the procession of historical events is both causal and deterministic.

Naturalistic historiography typically assumes that the single observed history occurred due to a confluence of random factors and is unintentional. Therefore, the procession of historical events is neither causal nor deterministic.

The key difference here between faith based historiography & naturalist historiography is the introduction of randomness. Naturalists have to account for randomness in a way faith based historiography does not. While believers may need to and certainly should account for randomness and uncertainty into the future, the concept of Qadr means observed history has a meaning. This does not exist in a naturalistic framework.

In hard & social sciences, randomness, noise, and statistical distributions are accounted for when analyzing data. If the HCM is more than an interpretative science (Freudian/Jungian psychoanalysis, Marxist/Austrian economics, etc.) as HCM academics claim, then it should account for the same kind of naturalistic randomness.

IF history is a product of naturalistic forces THEN naturalistic forces are subject to randomness, statistical distributions, & noise THEREFORE explanations of history must account for that OR the explanations are merely interpretative and not actually explanatory.

By Ali Shukri Amin

بسم الله الرحمن الرحيم

Two Muslim academics recently published an article entitled, The Historical-Critical Method and Why Muslims Should Use It: Response to R.A Jabal in which they advocate for Muslims to apply the Historical Critical Method (hereafter referred to as the HCM) in Islamic studies. Their article was itself a response to another article by Rafay Jabal entitled, What is the Historical-Critical Method and Why Muslims Should Care. In Ahmad & Rasheed’s article, they argue that Jabal has not proposed a solution to his critique of the HCM,

Ultimately, while Rafay proposes that what he is doing is pointing out the biases and limitations of the method, he falls into a trap that many critics of the method fall into. He does not actually provide a firm solution to the problem.

Rafay’s critique was heavily focused on the epistemological commitments the HCM has and how this, in his view, biases the method to the point of unreliability. Ahmad & Rasheed argue that the HCM is valuable as it has produced historical knowledge “from pre-historic times to societies of modern history” and, therefore, Jabal’s deconstructive arguments and prescription to shed the HCM are unwarranted. Ahmad & Rasheed conclude that Muslims should utilize the HCM.

I will dissect their arguments in several parts, dealing in turn with the entire assumption stack of the HCM.

Pseudo-Scientific Approach to History

Ahmad & Rasheed, like other HCM scholars, often find the metaphysical and epistemological critiques leveled at the HCM to be baroque. This betrays a lack of appreciation as to the depth and importance of these arguments. However, I have chosen to begin with a critique which was wholly undiscussed by Rafay, Ahmad, or Rasheed. This critique operates within the accepted parameters and assumption stack of the HCM. I begin with a critique that is compatible with presupposing the entire secular modernist assumption stack for the purpose of demonstrating evident flaws in the HCM as a method which are unaddressed, before expanding outwards to increasingly higher levels of inquiry.

The HCM is advertised as a secular, evidence-based, unbiased method for approaching history. Indeed, it is even advertised as scientific. Dr. Joshua Little, eminent Historical Critical scholar and Research Fellow of Islamic Origins at National Chengchi University, says about the HCM,

“There’s a systematic use of science and established background knowledge. When we’re evaluating a report, a historical report, we evaluate it in light of modern science.” [The Origins of Hadith: Traditional and Modern Perspectives! | Dr. Joshua Little, 1:40:56 https://www.youtube.com/watch?v=BZGw3wiPs1Q ]

Yet, by definition, if you cannot test a hypothesis and run controlled experiments or statistical analysis – a method is not scientific. Does that preclude the HCM from being a science?

Modern academia respects a few fields of research as being reliable for producing truth. These consist of the natural sciences, engineering and technology, medical and health sciences, agricultural sciences, and social sciences. Humanities/arts are not considered to be self-reliant methods of accomplishing truth, but seen as valuable for human expression and administration.

So, within the secular method, not everything must be a hard science to produce objective truth. However, if the HCM is not a natural science or social science, then it becomes pseudoscientific when proponents claim it can produce that truth. HCM proponents do claim that it is the best method for arriving at objective historical truth, although they may temper that claim with the fact that all historical truth is “probabilistic.” Yet, that caveat does not deal with a major problem with the Historical-Critical Method has. The HCM cannot validate or justify its own existence. Upon what basis does a proponent of the method justify that the HCM is superior to the historiographical method of ibn Khaldun? What is the methodological evidence that one approach & its conclusions are superior to another?

In hard sciences, such as evolutionary biology, there are evidence-based methods by which scientists may determine that other methods of scientific inquiry lead to erroneous conclusions in the past and due to these facts this other method is better. If this is the case then the HCM is not a scientific method. With a scientific method, we would be able to isolate the exact variables and have reproducibility of results.

In fact, proponents of HCM offer no way to measure historiographical frameworks for their value in explanatory function – since history is an environment where experiments are impossible and therefore there is no reproducibility & scientific conclusions cannot be achieved. However, the concept of nonexperimental sciences is oxymoronic. If neither experiments nor statistical analysis can be done, it's not a hard science. An example is Freudian psychotherapy. To some psychologists, psychotherapy is a very useful method. However, it cannot be validated experimentally though it may produce invaluable insights. The value of such insights is totally subjective.

This is contrasted with some of the hard sciences that are almost entirely based in past data. An example may be cosmology, which is based on observation & statistical analysis of datasets. There can be no controlled experiments but there are theoretical experiments that can be validated or disproven. Furthermore, cosmology and other similar sciences, such as geology, are heavily based on statistical analysis; not on the subjective opinions of researchers on increasingly far-fetched “parallels” (which betray the causality-correlation fallacy). Historiography is categorically different than cosmology. This renders the entire field more like psychotherapy. A researcher cannot come up with a formula to validate a thesis in historiography that can be back projected on historical data. That is not scientific.

If the HCM is not claimed to be a hard science, can it be claimed to be a social science? Proponents of the HCM may claim that HCM is a probabilistic social science. However, they offer no mechanism by which probabilities are being calculated. There are no variables being assigned or underlying rules. Instead, we find narratives based on assumptions that researchers arbitrarily deem most “probable” that are then projected onto the data. Until some statistical framework is proposed and actually utilized, the idea that the HCM is “probabilistic” is not a serious claim.

The HCM is best understood as an “interpretative science” along the lines of Freudian psychotherapy or Marxist economic theory. This renders all of historiography including the HCM, from a secular framework, as a pseudoscience. Since the "engine of historical events" that contains any potential laws of history is invisible to us, if such an engine exists and if such laws exist, we cannot study it in an empirical naturalistic fashion. From a secular perspective, that renders all of historiography no better than subjective interpretation.

Since HCM is not a hard science, it has no basis upon which to validate competing methods or conclusions for objectivity. Experiments and statistical analysis do not apply. How the HCM ensures that it is, itself, a legitimate method, and that it is superior to other methods, and that it can be used to reliably arrive at truth? Why do scholars of the HCM trust the HCM if not for experiments, statistics, and the other accepted tools of modern academia?

The only recourse to determining this would be to raise to the “higher sciences” which were handwaved by Ahmad & Rasheed in their article. Indeed, this is where we find our answer.

The Reproducibility Problem in Science

Science itself is not an infallible method of arriving at truth methodologically. In fact, modern science is replete with methodological challenges. Since the mid-20th century, prescient scientists have come to acknowledge that, despite the best efforts towards scientific rigor, many published research findings are either irreproducible or outright false. Stanley Schoar published a groundbreaking paper in the Journal of the American Medical Association in 1966 in which he concluded that nearly 73% of conclusions in published scientific papers had unjustifiable conclusions. Schoar had identified the “replication crisis”, a now widely recognized problem especially in psychology and medicine. As a result, over the last nearly 60 years, a new field of “metascience” has emerged to conduct meta-analyses of papers within specific disciplines to evaluate validity over time.

In 2005, Dr. John Ioannidis published a profound essay entitled Why Most Published Research Findings Are False. In it, he implicated a large number – if not the majority – of peer-reviewed medical research papers as being irreproducible. Using statistical significance to screen papers, he build a statistical model that indicated most of the conclusions were drawn from what could be considered statistical noise or false positives.

The findings of meta-researchers such as Schoar and Ioannidis go beyond quaint intrigue. Metascience has been used to upend widespread convention about medicine that has affected millions of people. As an example, since 1986, tens of millions of people have taken some form of fluxotine (generic Prozac). Prozac is a chemical treatment for depression in pill form that works by inhibiting serotonin receptors in the brain. is so prevalent that it has been observed in American public water supplies and affects wildlife (https://pmc.ncbi.nlm.nih.gov/articles/PMC5677497/). Yet, despite the academic consensus that depression may be medically described as reduced levels of serotonin in the brain, in 2022 several researchers challenged this conclusion. They published an article in esteemed scientific research journal Nature under molecular psychiatry. A widespread systematic review entitled The Serotonin Theory of Depression: a Systematic Umbrella Review of the Evidence. Utilizing a meta-analysis of published peer-reviewed studies, they concluded that there was no basis to assert that depression may be defined as insufficient serotonin as treatment utilizing serotonin inhibitors (SSRI’s) demonstrated no statistical significance.

Modern scientific conclusions, while often insightful and incredibly beneficial, are also fraught and often unreliable. This is withstanding the strict methodological guardrails that exist for scientific research. No such guardrails exist within interpretative methods like the Historical Critical Method, further throwing into doubt any conclusions HCM researchers may arrive at.

By Ali Shukri Amin