Societal Thinking Fast & Slow: A Better Way Now Exists
Daniel Kahnemann was a renegade social scientist who revolutionized economics with an expanded model of human (ir)rationality.
March 18, 2024
by Brigham Adams, PhD
Societies, like individuals, think via two separate, fast and slow, systems. And we now have the opportunity to deploy a third, with the best elements of each.
In Thinking Fast & Slow, Daniel Kahnemman explains how individual cognition occurs through two different systems. System 1 is fast, allowing for rapid action. It is intuitive, drawing on broad, but imprecise impressions, and is often informed by recollections of felt emotions, and even neural circuitry located in the heart and gut. This is a very embodied kind of thinking, directly tied to our actions and experiential history, and ready to move at the speed of life whether we are playing a fast-paced sport or engaging in an escalating emotional argument.
System 2 is quite different. It is slow, orderly, and precise. System 2 thinking occurs predominantly in the prefrontal cortex, the brain we developed to better manage and direct the more ancient brain systems we share with so many animals from our common ancestors. System 2 allows us to read, write, listen, and speak linear streams of ideas via languages, to deliberatively compare and hypothetically test the implications of ideas and scenarios. It is far slower than System 1 thinking, but this ponderous thinking mode helps us wisely evaluate competing choices, carefully express our imaginations and emotions, and circumspectly take weighty decisions.
These systems can be used separately or in tandem, their relative utility often depending on the temporal window in which our thinking must be completed. Playing sports, decisions must be made rapidly, in split seconds. Do we pass the ball or shoot on the goal? System 2 hasn’t enough time to decide. But System 1 can. Later, watching video of the match, System 2 might conclude that a pass would have been better than a shot. Both systems are valuable.
Society, too, I posit, thinks through two separate systems. Our fast System 1, ready for quick action, is our news and social media. It refreshes everyday, many times per day, actually. It evaluates vast amounts of information quickly, outputting concise and actionable summaries. And the media (legacy and new) are very close to the action, very connected to the agents (people and organizations) who make consequential moves in the world at the speed of the daily news and its fortnightly cycles.
Our System 2 for societal-scale thinking, on the other hand, is slow just like the individual-scale System 2. Society ponders social reality, and all of reality, at a glacial pace. Scientists require a year or more to observe some phenomena in the world and set up the rigorous data collection, study designs, statistical analyses, visualizations, and reporting to submit some new thinking for publication. Then, another year may pass as peer scientists review the idea, request refinements to the research, and receive improved work fit for publication. This is a very deliberate process flowing through the narrow cognitive channels of a tiny set of researchers at a time, designed not for speed, but for rigor and certainty. To protect the integrity of this process, practitioners of System 2 societal thinking separate themselves from the rest of society both physically and socially, on campuses where their mode of societal thinking is relatively ‘unpolluted’ by Societal System 1 thinking. They even eschew close interaction with the people and organizations that regularly attempt to act rapidly at the societal level (journalist and politicians), maintaining their focus on providing the greatest long-term thinking they can muster.
As at the individual-level, both the fast and slow modalities of societal thinking are important and valuable. But, I contend that both systems are probably underappreciating the other, and anyway, there is now a third way that adapts the best features of both systems to achieve a new human collective superintelligence that is both fast and rigorous, careful and effective. This new system, Societal System 3 we could call it (though I gently deplore the scientific penchant for mystifying its important concepts with arbitrary and generic naming conventions), moves at the speed of the news cycle, and lives in the online halls of informational and discursive power. It transparently and speedily applies the rigor of science’s coherent, honest, intersubjective epistemology to System 1 content, upgrading its informational value, and upgrading the critical thinking capabilities of society in the process.
Studies have shown that “Crowds Can Effectively Identify Misinformation at Scale” (https://journals.sagepub.com/eprint/NMKPE6FS32EDG7BJMIBV/full). However, the world does not know if this capability can be harnessed and deployed at scale as a new Societal Thinking System 3. However, we have developed, and propose a test of, crowds’ capacity for misinformation detection using a newly scale-ready societal thinking system called Public Editor.
This misinformation identification and labeling system provides granular and specific labels directly onto the words and phrases of online content highlighting 50+ different reasoning errors, cognitive biases, inferential mistakes, and rhetorical manipulations that often cause people to learn erroneous beliefs and misunderstandings. Newsreaders (i.e. people who read the news) view the labels overlaid onto articles as they read daily news content. They also see, in the margin, a display of definitions and additional examples of each error. The labels provide newsreaders with targeted, modest, and instructive doubt about mis-reasoned phrases and claims, so they don’t learn false facts/claims as they read. And as they compare any error to its definition and other examples in the margin, their neural circuitry potentiates the learning. Via iteration, newsreaders learn to discern misinforming reasoning and rhetoric for themselves. (We are performing functional-cognitive tests of these hypotheses with an independent researcher from Oxford and Stanford.)
A prior study with the EU Commission’s own disinformation experts (EEAS-ESTF) found the system’s outputs useful. They fed a set of 250 articles they had previously debunked into (an earlier version of) the system operated by lightly-trained online annotators, and found that the system identified & labeled the same dis-/mis-information their experts had in 94% of articles. In 86% of the articles, the (crowd using the) system found additional misinformation the EU experts had not identified in their debunking reports.
The system is being deployed online and via classrooms and civil society organizations in Europe and the US starting April 5th. But, the system needs a dozen or more researchers measuring and stress-testing its performance and processes to continually improve its functioning, accuracy, legitimacy, and impact. From user interfaces to consensus processing to bias-mitigation algorithms, and annotation-instrument tuning, this system will improve with researcher engagement. This intersubjective epistemology system will generate many thousands of examples of human reasoning errors, and will need researchers to study how those errors proliferate throughout internet channels, platforms, human communities, networks, and media. Plausible interventions will need to be tested before the EU’s DSA and USA’s CDA-S230 clauses will trigger enforcement actions deploying the system into social media platforms. The data output by this system will be capable of training AIs that can label misreason in any content. It will be able to train ‘reasoning engines’ that ensure LLMs output lucid cleanly-reasoned text. This system and its data have many surfaces that require careful research and continual improvement.
Particularly because the system can feed inputs into AI to scale quickly and broadly, the system can have a very large impact touching millions of people. To function fully, optimally, and responsibly, it requires the careful and creative attention of many researchers and engineers. Philanthropists interested in delivering this new capacity to human society, can contact Public Editor’s project lead directly: brigham@goodlylabs.org.