What Cass Sunstein's latest book can tell us about scientific consensus.
A striking feature of cosmology is the ostensibly broad consensus among scientists about the core of the current standard model - the Big Bang theory itself.
Sure, there is some (though not much) room for debate and competing theories to inventions like inflation, dark matter, and dark energy. But questioning the basis of the General Relativity universe and the Big Bang expansion model is almost unheard of, both inside the field and in popular science writing.
This consensus is frequently invoked to imply there is solid scientific evidence for the theory. Big Bang proponents often make claims to "overwhelming observational evidence" and casually dismiss alternative explanations as crackpot fringe theories.
But looking more closely at the foundations of cosmology reveals a different story.
As I outlined in Scientific American and in my reply to Forbes' science writer, what is claimed as evidence in favor of Big Bang theory, and the framework for interpreting this evidence, is really much more problematic than it is made to appear.
Compared to other sciences, the logical and empirical basis for Big Bang theory is in fact remarkably weak. As I argue in my book, the Big Bang is a metaphysical hypothesis that has succeeded in becoming the only acceptable scientific model of the universe.
How could the theory become so widely accepted unless it reflected irrefutable scientific evidence?
Leave aside for a second your own judgment on whether Big Bang theory is right.
For the sake of a simple thought experiment, imagine a future in which the theory is considered to be wrong and there has been a paradigm shift. Science historians would then be grappling with how this consensus could have formed and held sway for so long.
In his new book, Conformity: The Power of Social Influences, Cass Sunstein brings a useful perspective. Drawing on well-established social experiments, Sunstein's analysis, highly relevant for our social media age, applies beyond law and politics to dynamics in the sciences.
Sunstein concludes:
"A 'majority consensus' is 'often capable of misleading individuals into inaccurate, irrational, or unjustified judgments.' Such a consensus 'can also produce heightened confidence in such judgments as well. It follows that 'so long as the judgments are difficult or ambiguous, and the influencing agents are united and confident, increasing the importance of accuracy will heighten confidence as well as conformity—a dangerous combination.'"
Social influence factors are rarely considered seriously by scientists, who tend to operate with an ideal of rational individualism and objectivity. But their pursuit undoubtedly takes place in a social context and is liable to the mechanisms that Sunstein analyzes.
Conformity dynamics become particularly powerful and pronounced when it comes to dealing with very difficult problems. Social experiments in a multitude of contexts clearly show that when the problem before us is particularly hard to answer on our own, we tend to follow the crowd. Is there any problem harder than the nature of the entire universe?
When faced with difficult questions and limited knowledge, we tend to defer to those we perceive as authorities on the matter, and typically these are recognized experts in the field. But experts, as Sunstein explains, can be led astray and this can happen at a collective level despite the rational intention of every single one of its actors.
A key mechanism in this collective effect is so-called cascades. In an informational cascade, people primarily rely on the signals conveyed by others rather than independent information. "Once this happens, the subsequent statements or actions of few or many others add no new information. They are just following their predecessors."
Because nobody really knows on their own, the reference to the authority of someone else takes precedence, and in turn, that authority may really defer to another authority (who might in some cases even be dead).
Loud confidence signals on ambiguous questions and the absence of non-confidence signals from others can swiftly move a group into a consensus view.
As Sunstein notes, "a particular problem arises if people think the large number of individuals who say or do something are acting on independent knowledge; this can make it very hard to stop the cascade."
Indeed, "both outsiders and contributors to cascades often seem to mistake a cascade for a series of separate and independent judgments."
This can easily happen inside a field of experts. As one review of Sunstein's book notes,
"deliberation among the likeminded creates an ideological echo chamber where moderately held beliefs become dogma... Groups act as affect multipliers. They increase the credibility and acceptability of certain ideas held by those in the group."
In popular science writing, cascades also easily occur because each story tends to restate propositions made by others, and for each reference back to the same claim, a semblance of conviction grows. The constant repetition (that makes up so much of science history) can mask a situation in which, as Sunstein puts it,
"most signatories lack reliable information on the issue in question and are simply following the apparently reliable but actually uninformative judgment of numerous others."
But constant repetition of the same information does not create a solid foundation, and especially not when potentially contradictory information is omitted or withheld.
This can happen in a reputational cascade, which comes into play when actors are concerned to preserve their standing in the field. They tend to go along with the crowd and keep any objections or conflicting viewpoints to themselves.
A reputational cascade develops along with an informational one, and reinforces it, when it simply becomes too risky to go up against core consensus.
Several astrophysicists and cosmologists who work inside the field privately tell me they know all too well that pursuing alternative theories or challenging the core of the Big Bang theory is tantamount to career suicide. It is very hard to make a living as a scientist without privileged access to research funding and to major experiments.
In politics, dissent may be a way to stand out and make a mark, but in science, dissenting against core orthodoxy is an inherently risky move. In this way, as Sunstein notes, social influences mirror underlying economic conditions:
"cascade effects, and blunders, are significantly increased if people are rewarded not for correct decisions but for decisions that conform to the decisions made by most people."
Obviously, this is not an exhaustive analysis. But Sunstein's perspective raises some intriguing questions. His focus on law and judgment is not irrelevant to the sciences and especially not to physics, which is heavily concerned with laws of its own. After all, physics is the only science, as philosopher of science Isabelle Stengers puts it, that is engaged in judgment of phenomena by submitting them to a rational ideal.
In my view, cosmology is particularly liable to conformity problems because as a science, it operates at the limits of knowledge. Based on an operative hypothesis that hovers in obscurity, its subject matter is strictly incomparable in scope to other sciences.
Cosmology suffers both from limited means of observation, reliance on the framework within which data are interpreted, and privileged access to authorized resources. These conditions taken together would make it a breeding ground for cascading dynamics.
Of course, it is possible that the Big Bang theory prevails simply because it is true. But when proponents loudly claim it is the only possible alternative, beware the conformist tendency toward heightened confidence in problematic judgments.
Sunstein's work is interesting for demonstrating how large groups and institutions involving experts can be led astray, and how an emerging consensus on a very difficult problem can harden into dogma and orthodoxy.
Reference
Sunstein, Cass R.. Conformity: The Power of Social Influences. NYU Press, 2019. Kindle edition.