By Arthur M. Diamond, Jr.
First Posted to The Institute of Arts and Ideas web site on Weds., March 25, 2026
An experienced nurse in the neonate intensive care unit is mostly focused on the infant who is her main responsibility. But she notices that a nearby infant is cycling through minor changes in skin color. That infant’s primary nurse sees it too. The infant then turns blue-black. The experienced nurse knows it is pneumopericardium, where air pressure around the heart keeps the heart from sending blood into the infant’s body. She knows it because she had been the nurse for an infant who died from pneumopericardium. The heart monitor misleadingly seems to show that the heart is still beating, so the infant’s primary nurse thinks the problem is a collapsed lung. As the chief doctor arrives the experienced nurse “slaps a syringe in his hand” and tells him to “stick the heart” to release the air. An x-ray tech confirms the diagnosis, the doctor acts, and the infant lives.
The experienced nurse was out of line. The infant who lived was not her responsibility and it was not her job to tell the chief doctor what to do. She could have been punished, but she took a chance and acted on intuitive knowledge that she could not immediately articulate. She had intuition that proved correct and her acting on her intuition was literally a matter of life and death. But we set up barriers to discourage ourselves and others from acting on our intuition. We establish regulations, credentials, protocols, manuals.
___
Breakthrough innovative entrepreneurs often have limited formal theoretical knowledge, but high levels of informal unarticulated knowledge.
___
Dr Min Chiu Li was an oncologist at the National Cancer Institute in the US in the early days of administering methotrexate chemotherapy to women who had choriocarcinoma cancer. The NCI protocol mandated that when the visible symptoms of cancer were gone, he should stop chemotherapy. But he had an intuition that small amounts of cancer still lurked after the visible symptoms were gone. So he violated the protocol and gave his patients a longer course of chemotherapy. The administrators at the NCI fired Min Chiu Li for violating the protocol, but later were surprised to observe that the patients treated according to the protocol were dead, and the patients treated by Min Chiu Li were alive.
In my book Openness to Creative Destruction I argue that breakthrough innovative entrepreneurs often have limited formal theoretical knowledge, but high levels of informal unarticulated knowledge. Henry Ford, Bill Gates, and Steve Jobs did not graduate from college. What is true of innovative entrepreneurs is also often true of innovative scientists. When James Watson and Francis Crick had lunch with famous biochemist Erwin Chargaff, Crick could not remember the well-known chemical details of the four bases of DNA. Chargaff dismissed the pair with contempt. Watson and Crick did not excel in the memorization of theory but had intuition that allowed them to see the double-helix structure of DNA. AI expert Melanie Mitchell and cognitive psychologist Gary Klein agree that we have more unarticulated knowledge than articulated knowledge.
But we too seldom ask: how useful is it? Or an even better question is: how useful could it be if we sought to make good use of it rather than sought to ignore or block it? Unarticulated knowledge deserves deeper study, and Klein is one of those who has made a start. Over his career he has modified his taxonomy of the types of unarticulated knowledge. What I am calling “unarticulated knowledge” he calls “tacit knowledge,” a label I prefer to reserve for the kind of muscle-memory bike-riding example the phrase’s originator, Michael Polanyi, made famous. In one of his later efforts, Klein distinguishes five types of unarticulated knowledge: Perceptual, Conceptual, Embodied, Social, and Metacognitive. The type I am most concerned with in this article is the Conceptual, within which he includes: “pattern recognition; mental models; expectancies; mindsets; noticing the absence of expected events; imagining antecedents and anticipating consequences; seeing affordances.”
The size and importance of unarticulated knowledge has implications for the current worries that the growth of AI will create widespread job loss. If worker productivity depends importantly on their unarticulated knowledge, and if AI models are trained solely on databases of articulated knowledge, then we have built-in limits on the extent to which AI can replace humans in the labor market.
The level of regulations in the US has steadily increased over many decades, at the same time that the number of breakthrough innovations has fallen (see also Graeber 2012; Huebner 2005). We may be wrong to rely so much on regulations, credentials, protocols, and manuals, but we do not do so out of simple stupidity or evil intent. We do so for several plausible reasons.
One of the reasons is because unarticulated knowledge is often called (including by Klein and me) “intuition” and we associate intuition with mysticism, knowing that mystics have often made predictions that proved false. We also know that our intuition is sometimes systematically biased in a variety of ways. Daniel Kahneman has many examples in his Thinking, Fast and Slow, including, for example, the anchoring effect, confirmation bias, and loss aversion.
But Klein thinks we sell ourselves short if we dismiss intuition. The intuition that he defends is based on experienced patterns, not mystical epiphany. This kind of intuition is on solid ground partly because it often can be articulated when we have enough time to do so, and when it is worth the time to do so.
In a life-and-death case, Lieutenant Commander Michael Riley on the British HMS Gloucester had about 90 seconds to decide if the object coming towards them on the radar was a friendly American plane or a hostile Silkworm missile. Riley was sure the object was hostile, and at the last second shot it down. He could not explain how he had known and why he was sure. By asking a series of shrewd probing questions, Klein teased out how Riley had known that the blip was hostile. Although the plane and the Silkworm flew at different altitudes, the radar did not directly report altitude. But experienced and focused users of the radar could infer altitude from the distance from the shore when a blip first became visible. In this case, Riley did not have the time to articulate the unarticulated, but later Klein, with Riley’s help, proved that it could be done.
Another reason we rely so much on regulations, credentials, protocols, and manuals is that we worry that we have no good way of judging other people’s claims to have unarticulated knowledge. So we worry that the unscrupulous might take advantage of us. This worry often arises in situations subject to what economists call “the principal-agent problem.” The problem arises when the principal pays the agent to do a task; then the agent takes the money but doesn’t do the task.
The principal-agent problem often exists even when we are dealing with articulated knowledge. An increasing number of scientific journal articles and grant proposals are fraudulent. The journals and the grant agencies are paying (in terms of resume entries and money grants) for bogus research. A prominent sad example is Alzheimer’s research. Charles Piller, a journalist at the distinguished journal Science, has expanded his exposé articles into the book Doctored, documenting that much of the leading research has been fraudulent, helping to explain why progress against this major disease has been so limited. The victims include first and foremost those suffering from Alzheimer’s, but also the taxpayers who fund government research grants, and the Alzheimer’s researchers whose honest but modest results have been rejected for publication and grants because they falsely seem inferior to the fraudulent results.
So we guard against unarticulated knowledge because we worry that if we can be so extensively defrauded when we are dealing with claims of articulated knowledge, how much more extensively will we be defrauded if we do not protect ourselves against claims of unarticulated knowledge?
The principal-agent problem is even more severe in common situations where the principal is acting as a fiduciary for others. So a government grant-giver has a moral duty to act prudently since he is acting as a fiduciary for the taxpayer. And a venture-capital fund investor has a moral duty to act prudently since he is acting as a fiduciary for the investors in his fund.
___
We should seek opportunities to fund on the basis of performance, not based on committee evaluation of written proposals.
___
This contrasts with angel investors who invest their own money and so can morally take greater risks based on more tenuous hunches. When the Omaha billionaire Walter Scott spoke to one of my classes I asked if he had been aware of the technological concerns George Gilder had about the Level 3 fiber optics network firm in which Scott had heavily invested. His somewhat gruff response was that he didn’t know technology, but he did know Jim Crowe, the founder of Level 3. Scott was spending his own money so he was not violating any fiduciary responsibility in mistakenly investing in Level 3.
When the principal and the agent are the same person, the principal-agent problem disappears. When the principal is spending their own money, the principal-agent problem is at least mitigated.
We can avoid the principal-agent problem by making it easier for entrepreneurs and scientists to self-fund their ventures and research, thus avoiding the principal-agent problem. For entrepreneurs this can be done by letting them keep the funds that they earn through successful entrepreneurship. Those who have given us the fullest proof of the value of their innovation by succeeding in the marketplace, are allowed to keep the wealth they thereby earn, so they can try it again. These are the serial innovative entrepreneurs like Commodore Vanderbilt, Steve Jobs, and Elon Musk. (The builders of a new computer in Tracy Kidder’s The Soul of a New Machine, compared what they were doing to the game of pinball, where the reward for doing it well is the chance to do it again.)
New York Times financial columnist Andrew Ross Sorkin wrote a column criticizing Steve Jobs for not signing onto the Bill Gates Foundation pledge to give the foundation a large part of his wealth. Jobs was famously known for his intuition about which new products would be “insanely great.” By retaining wealth from previous successes, he could quickly pivot to the next “insanely great” product as a new idea emerged, without having to articulate and sell the idea to a board of directors or to venture capitalists or to Wall Street. So we should encourage successful innovative entrepreneurs to reject the advice of Andrew Ross Sorkin, and instead hold onto their wealth. And we should oppose legislation being proposed in the US Senate to tax all substantial wealth, including that of deserving serial innovative entrepreneurs.
If a successful innovative entrepreneur runs out of new ideas himself, then rather than use his wealth for general charity, he should try to find and invest in other would-be-innovative entrepreneurs who share the traits that enabled the innovative entrepreneur’s own success. (PayPal entrepreneur Peter Thiel and Netscape entrepreneur Marc Andreesen are following this advice.)
We should seek opportunities to fund on the basis of performance, not based on committee evaluation of written proposals. George Stephenson had no formal education and was not very articulate. He could not give a good explanation of why the safety lamp he invented would prevent miners from dying of gas explosions. But he proved it by entering a mine with the lamp and walking toward a chamber known to contain gas. Later and more famously, Stephenson’s Rocket locomotive was not the sleekest looking in the Rainhill Trials contest, and Stephenson was not the most articulate defender of his entry, but unlike the other locomotives that in one way or another broke down, the Rocket kept chugging along. DARPA is one of the more successful government funders of new technology. They often fund based on contests. The X-prizes, founded by Peter Diamandis, are a private-sector effort to fund based on performance.
To reduce the principal-agent problem in science, we should be more open to citizen scientists self-funding their own research, as was commonly done in an earlier period of science, and as has recently been done by neuroscientist Jeff Hawkins, who earned his wealth by being the entrepreneur who developed the successful PalmPilot personal data assistant. The motto of the first scientific society, the Royal Society of England, was Nullius in verba (take no one’s word for it), meaning that anyone who was willing to show the evidence for their findings could participate in science. Make citizen science respectable again. Even today, not all successful innovative scientists rise through the Ivy league or through Oxford and Cambridge.
We should also experiment to find better ways to fund science where self-funding is not possible. We should consider Robin Hanson’s institutional innovation of a betting market where would-be scientists could bet on scientific propositions. Besides finding ways for would-be scientists to self-fund, we should find ways to reduce the amount of funds needed to participate. Universities could be made more efficient. The costs of entry to doing science in some disciplines is already low; citizen scientists make important contributions to astronomy, archeology, and botany. And the costs of contributing to science in other areas should be reduced by reducing regulations.
More broadly we can encourage managers at all levels to give decision rights to their employees. Assign them domains of action where they will not be micro-managed, where they can be alert to patterns and act on the patterns they observe, where they can make use of their unarticulated intuition. Within those domains the employee is not second-guessed by a micro-managing boss or a detailed operational manual.
BIBLIOGRAPHY (not posted in IAI online version):
Barber, Charles. In the Blood: How Two Outsiders Solved a Centuries-Old Medical Mystery and Took on the Us Army. New York: Grand Central Publishing, 2023.
Christensen, Clayton M., and Henry J. Eyring. The Innovative University: Changing the DNA of Higher Education from the inside Out. San Francisco, CA: Jossey-Bass, 2011.
Cowen, Tyler. The Great Stagnation: How America Ate All the Low-Hanging Fruit of Modern History, Got Sick, and Will (Eventually) Feel Better. New York: Dutton Adult, 2011.
DeVita, Vincent T., and Elizabeth DeVita-Raeburn. The Death of Cancer: After Fifty Years on the Front Lines of Medicine, a Pioneering Oncologist Reveals Why the War on Cancer Is Winnable–and How We Can Get There. New York: Sarah Crichton Books, 2015.
Diamandis, Peter H., and Steven Kotler. Bold: How to Go Big, Create Wealth and Impact the World. New York: Simon & Schuster, 2015.
Diamond, Arthur M., Jr. “How to Cure Cancer: Unbinding Entrepreneurs in Medicine.” Journal of Entrepreneurship and Public Policy 7, no. 1 (2018): 62–73.
Diamond, Arthur M., Jr. Openness to Creative Destruction: Sustaining Innovative Dynamism. New York: Oxford University Press, 2019.
Gigerenzer, Gerd. Gut Feelings: The Intelligence of the Unconscious. New York: Penguin Books, 2007.
Graeber, David. “Of Flying Cars and the Declining Rate of Profit.” The Baffler, no. 19 (2012). https://thebaffler.com/salvos/of-flying-cars-and-the-declining-rate-of-profit
Hanson, Robin. “Could Gambling Save Science? Encouraging an Honest Consensus.” Social Epistemology 9, no. 1 (Jan.-March 1995): 3–33.
Hawkins, Jeff. A Thousand Brains: A New Theory of Intelligence. New York: Basic Books, 2021.
Hawkins, Jeff, and Sandra Blakeslee. On Intelligence. New York: Times Books, 2004.
Huebner, Jonathan. “A Possible Declining Trend for Worldwide Innovation.” Technological Forecasting and Social Change 72, no. 8 (Oct. 2005): 980–86.
Jena, Anupam B., and Christopher M. Worsham. Random Acts of Medicine: The Hidden Forces That Sway Doctors, Impact Patients, and Shape Our Health. New York: Doubleday, 2023.
Jenkins, Tania M. Doctors’ Orders: The Making of Status Hierarchies in an Elite Profession. New York: Columbia University Press, 2020.
Kahneman, Daniel. Thinking, Fast and Slow. New York: Farrar, Straus and Giroux, 2011.
Kidder, Tracy. The Soul of a New Machine. 1st ed. Boston: Little, Brown and Co., 1981.
Klein, Gary A. Sources of Power: How People Make Decisions. 20th Anniversary ed. Cambridge, MA: The MIT Press, [1997] 2017.
Klein, Gary A. “Unpacking Tacit Knowledge; Applying the Tacit Knowledge Concept More Effectively.” Psychology Today, 2023. https://www.psychologytoday.com/us/blog/seeing-what-others-dont/202307/unpacking-tacit-knowledge
Landes, David S. “Why Europe and the West? Why Not China?” Journal of Economic Perspectives 20, no. 2 (Spring 2006): 3–22.
McLaughlin, Patrick A., and Oliver Sherouse. “The Impact of Federal Regulation on the 50 States, 2016 Edition.” Arlington, VA: Mercatus Center, 2016.
Mitchell, Melanie. Melanie Mitchell on Artificial Intelligence. EconTalk, interviewed by Russ Roberts, Jan. 6. 2020. https://www.econtalk.org/melanie-mitchell-on-artificial-intelligence/
Park, Michael, Erin Leahey, and Russell J. Funk. “Papers and Patents Are Becoming Less Disruptive over Time.” Nature 613, no. 7942 (Jan. 2023): 138–44.
Piller, Charles. Doctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimer’s. New York: Atria/One Signal Publishers, 2025.
Polanyi, Michael. Personal Knowledge: Towards a Post-Critical Philosophy. Chicago: University of Chicago Press, 1958.
Polanyi, Michael. The Tacit Dimension. Garden City, New York: Doubleday & Co., 1966.
Prentice, Claire. “Miracle at Coney Island: How a Sideshow Doctor Saved Thousands of Babies and Transformed American Medicine.” Kindle Single, 2016.
Richardson, Reese A. K., Spencer S. Hong, Jennifer A. Byrne, Thomas Stoeger, and Luís A. Nunes Amaral. “The Entities Enabling Scientific Fraud at Scale Are Large, Resilient, and Growing Rapidly.” Proceedings of the National Academy of Sciences 122, no. 32 (2025): e2420092122.
Rosen, William. The Most Powerful Idea in the World: A Story of Steam, Industry, and Invention. New York: Random House, 2010.
Smiles, Samuel. The Locomotive: George and Robert Stephenson. New and Revised ed, Lives of the Engineers. London: John Murray, Albemarle Street, 1879.
Sorkin, Andrew Ross. “Dealbook; the Mystery of Steve Jobs’s Public Giving.” The New York Times (Tues., Aug. 30, 2011): B1 & B4.
Watson, James D. The Double Helix: A Personal Account of the Discovery of the Structure of DNA. New York: Scribner Classics, [1968] 2011.
The article above was published behind a paywall on the web site of The Institute of Art and Ideas. I retain the copyright, so I am reposting the article here. I submitted a bibliography and internal parenthetical references, but following their usual formatting, they did not post those, but instead incorporated select web links to some of the sources. My submitted title was “Making the Most of Unarticulated Knowledge.” IAI did not like that title, so they chose: “Scientific Knowledge Can Lie Beyond Language.” I did not veto their title although I regretted that it neglected the practical implications of my article, which to me are as important as the scientific implications. The citation for the original posting of the article on IAI is:
Diamond, Arthur M. “Scientific Knowledge Can Lie Beyond Language.” Posted on March 25, 2026. The Institute of Arts and Ideas. Available from https://iai.tv/articles/scientific-knowledge-can-lie-beyond-language-auid-3530.
