The Scientific Revolution Killed the Polymath. It’s Time to Bring the Generalist Back

I noticed recently that books with the phrase “The Last Man Who Knew Everything” all share in common that their subjects lived during the period close to the Scientific Revolution, roughly between 1550 to 1700. (The examples I own are…

I noticed recently that books with the phrase “The Last Man Who Knew Everything” all share in common that their subjects lived during the period close to the Scientific Revolution, roughly between 1550 to 1700. (The examples I own are about Athanasius Kircher, a Jesuit priest born in 1602; Thomas Young, who studied topics such as optics and philology and was born in 1773; and Philadelphia area professor Joseph Leidy, who was born in 1823.)

It’s as if the Scientific Revolution — and the knowledge it spawned — killed the ability to Know Everything. Before then, it was not only possible to be a generalist or polymath (someone with a wide range of expertise) — but the weaving together of different disciplines was actually rather unexceptional. The Ancients discussed topics such as ethics, biology, and metaphysics alongside each other. The Babylonian Talmud discusses everything from astronomy and biology to morality and law, weaving them together into a single compendium.
So what changed? Scientific knowledge exploded in size, mainly due to the application of the scientific method to our surroundings. As that knowledge base and its domain experts grew exponentially, we began classifying and ordering all that we understood — from the classification taxonomy of Carl Linnaeus to manuals for categorizing mental disease. We made sense of our world by dividing information into manageable portions and distinct areas of proficiency.
But as people began to specialize, knowledge became fragmented. We chose to know more and more about less and less. We may have expanded what we as a society know — but it was at the price of no single individual being able to truly know it all.
We may have expanded what we as a society know — but it was at the price of no one being able to truly know it all.
Now, we obviously require specialized experts (as opposed to dilettantes) to solve specific problems; think about the field of medicine, for example. Yet the most exciting inventions occur at the boundaries of disciplines, among those who can bring different ideas from different fields together. As Robert Twigger noted, “Invention fights specialisation at every turn.”
In fact, some of the most exciting advancements in computing right now come from the field of deep learning — which itself draws from multiple fields: neuroscience, cognitive psychology, machine learning, natural language/ linguistics, computer vision, mathematics — to make the next step of AI possible. Companies such as Facebook, Google, IBM, and Microsoft are all involved.
But frankly, this kind of interdisciplinary approach isn’t happening more broadly in corporations, let alone in academia. There are institutional barriers (nearly all training, and data, lives in silos) as well as cognitive and biological ones. Even though the information storage capacity in our brains is vast (multiple petabytes), we eventually bump up against what we can truly understand (what some call The End of Insight) — or we just can’t hold all the relevant knowledge in our heads.
Still, we needn’t despair. There are ways to foster a culture of interdisciplinarity in a fragmented world.
We Need to Focus on the Tools, Not the Fields
Several years ago, a team of scientists examined hundreds of millions of clicks on scientific papers in order to discern the “clickstream” — the path readers take from one page to the next.

Samuel Arbesman

Samuel Arbesman is an applied mathematician and network scientist. He is currently a senior scholar at the Ewing Marion Kauffman Foundation and a fellow at the Institute for Quantitative Social Science at Harvard University. Arbesman is the author of The Half-Life of Facts and blogs for Wired Science on Social Dimension.

This data revealed patterns of how people moved from one subject area to the next. For example, nursing connects medicine to the fields of psychology and education. Organic chemistry bridges physical chemistry and analytic chemistry; economics is tightly intertwined with sociology and law; and the field of music stands quite distinct.
Of course, these are oversimplifications. Music incorporates concepts from physics and psychology while economics draws heavily from mathematics. But it’s one way to explore the interconnected nature of ideas, and it reminds us that we need to identify the tools necessary to bridge different domains and place them into a connected framework.
Let’s take a simple analogy. What do the following things have in common: doing Sudoku, constructing crossword puzzles, conducting logistics for large companies, playing Super Mario Brothers?
Well, in content terms, not much. They appear to be a collection of tasks that are easy to understand but not master. And it turns out that they’re all hard in a specific way: They’re what are known in theoretical computer science as NP-complete problems. Knowing this means each of these problems can be converted into a version of the other — I can construct a Sudoku puzzle that, if solved, could potentially shed light on how Walmart should route its delivery trucks.
Simply put, there are fields that have a certain generalizability, and their organizing ideas and tools can be used to find relationships between disparate areas. The most basic example of such a field is mathematics. As Eugene Wigner stated in his 1960 paper The Unreasonably Effectiveness of Mathematics in the Natural Sciences, “The miracle of the appropriateness of the language of mathematics for the formulation of the laws of physics is a wonderful gift which we neither understand nor deserve.”
Mathematics is a gift, an unbelievably useful tool for understanding our surroundings.
We Need To Think in Terms of Modules and Protocols
Take the science of complexity. It’s an attempt to abstract complex systems to their relevant interacting components, and then create a mathematical formalism that can explain the phenomenon being examined.

A complex system often has many interconnecting units that are themselves made up of many pieces. These larger units, which often have a certain degree of independence and internal sophistication, are known as modules. The property of modularity is a hallmark of many complex systems, from those in biology to programming. But an additional feature of these modular systems — often more abstract than the individual components — is how the pieces interact.
LEGO pieces can be combined in multiple ways. But what allows them to interact effectively is the shape and structure of the bricks — the bundle of properties that allow them to snap together easily. Similarly, multitudes of personal computers, massive servers, phones, and appliances can all connect to the Internet. What allows them to do this is the use of a common protocol, in this case the Internet Protocol (IP).
Whether referred to as protocols, standards, or interfaces, modules can vary, but can only interact — and be interoperable — if they use a common set of protocols.
Such modularity is not just a feature of physical systems. We need it for information, too. Think of the usefulness of websites like If This Then That (“Lincoln logs for your online life”) allowing “ingredients” like email, photos, RSS feeds, notes, weather updates, calendars, activity, and now location to be connected into meaningful recipes.
IFTTT is important because information is most useful when modules can be connected. And the same is true with knowledge. Distinct fields act like modules: complex, intricate, and complete with their own terminology and jargon. These features act as hurdles to interaction, and we can only interconnect the domains by building a set of common protocols.
This is exactly what the tools of mathematics and complex systems are: protocols. Not only do such tools allow someone to work in multiple disciplines — making it possible, once again, to be a generalist — they demand that similarities between different domains be made explicit.
This suggests that learning how to code is not enough to change how we think. Yes, coding does provide a certain structure to one’s thoughts. But there is a more important — and often ignored aspect — behind programming: through code, and the recognition that algorithmic similarity occurs over and over, we can see the similarities between different spheres of knowledge.
Far from being a tech-centric perspective, coding connects ideas across fields.

And We Need to Embrace the Machines
Charlie Munger, Warren Buffett’s investing partner, refers to the mental models required to understand the world — and that can be plugged into different situations — as a “latticework of models”. When suitably abstracted, these models can provide a powerful way of understanding many phenomena that might on the surface seem unrelated. Though an expert is a good guide along the way, these models are the tools that allow us to jump from field to field.
And machines can help, acting as partners in generalism.
Some people are not happy with this idea. But we need to welcome the tools that will allow us to more effectively manage the rapid growth of knowledge and prevent the balkanization of fields. As knowledge grows, we must increasingly rely on computers. This is not a new insight; in 1945, Vannevar Bush wrote the seminal “As We May Think” essay in The Atlantic describing the need for a machine:
But there is increased evidence that we are being bogged down today as specialization extends. The investigator is staggered by the findings and conclusions of thousands of other workers — conclusions which he cannot find time to grasp, much less to remember, as they appear. Yet specialization becomes increasingly necessary for progress, and the effort to bridge between disciplines is correspondingly superficial…
The difficulty seems to be… not so much that we publish unduly in view of the extent and variety of present day interests, but rather that publication has been extended far beyond our present ability to make real use of the record. The summation of human experience is being expanded at a prodigious rate, and the means we use for threading through the consequent maze to the momentarily important item is the same as was used in the days of square-rigged ships.
The problem of hidden knowledge (also discussed in my book The Half-Life of Facts) continues to grow. And now we have the Internet, and search, and big data which both surface, and hide, knowledge. As a way of addressing this problem of growing knowledge, Bush proposed a “memex” device, a type of rudimentary web browser.
But we can go further than browsing. Computers can help us generate new knowledge. It could be in proving mathematical theorems. It could be in finding papers that, when combined, yield new discoveries. It could be in taking different people’s annotations and finding unexpected connections between them. No matter what forms such discovery takes, though, it is clear that the crafts and tools of mathematics and computing will finally allow for the return of the generalist.
* * *

Where are all the generalists, anyway? They’re not really thriving in academia; for the most part, they’ve gone elsewhere to find their place, and one of these places is business. In the realm of data science at least, the startup world is beating academics at their own game when we consider examples such as Google and Facebook or and Misfit Wearables.
Videogame companies are also promoting this stitching-together of fields. Maxis (a subsidiary of Electronics Arts), the company that makes SimCity, Spore, and the Sims, is full of people who bounce from topic to topic, incorporating information from seemingly unrelated fields. Want to know why the steepest incline of streets in the newest version of SimCity is a certain number of degrees? It’s because the developers took the time to examine the steepest inclines in the world and based their coding of this information on that knowledge.
More generally, the world of business and entrepreneurship actively encourages those who see connections between disciplines. One who can recognize a relationship between two disparate fields of ideas will more likely come up with the next, big, new thing. That’s investment gold.
So how do we train people for this kind of thinking? The Girl Scouts once offered a fascinating merit badge: the Dabbler badge. This allowed a young scout who wanted to do a little bit of everything to not only generalize, but to be recognized for that achievement. Perhaps it’s time for the academic and business equivalent of the Dabbler badge: a way to acknowledge and foster those dabbling in different ideas, all the way from gradeschool to late career.
Specialization is clearly on the rise. It’s time for the generalist and the polymath to rise once again. Society needs to make a place for these Last Men and Women to Know Everything, and we need to go beyond the rhetoric of education reform to focus on the right tools that will make this happen.