The case against specialization

While there aren’t any certain predictions on what the next decade or beyond looks like, the people who will build the fields of the future are those who can see across disciplines.

Early in my philosophy of science classes, I learned that the world’s endeared physicists were really natural philosophers. Newton’s Principia was a devotion to the search for a Deity – alongside his laws of motion and his work on light was an equally, if not more, devoted theologian and alchemist. The famous “I think, therefore I am” came from the same man who gave us the Cartesian coordinate system. While Descartes built the foundations of analytical geometry, he was equally consumed by demonstrating the existence of God and the distinction between the human soul and the body.

During the time of natural philosophers, these subjects were in their infancy. A handful of names were freely studying across and building the foundations of mathematics, philosophy, natural sciences, and theology. Newton’s optical experiments were written on the same pages as his alchemy recipes. What’s since been carved into separate departments and given individual identities was, at its core, an undisciplined study of the natural world. Natural philosophy was “a method for exploring it, and an ethos or way of existing within it.”

It didn’t last. Eventually, the steam engine came to being, industrialization took off, and with it came the greatest flywheel of the modern era: the pursuit of specialization. Scientific societies, the exponential increase of academic journals, and the creation of research laboratories, all cultivated the conditions for specialized knowledge. The incentives were (and remain) obvious: go deep enough in one area, receive funding, publish your work, and gain social credit to build your identity. As the 20th-century progressed, the flywheel reinforced itself and there became more avenues in which you could specialize…and everyone wants to be special.

For the past several decades, there’s been not only a benefit but a true necessity for specialists. Our greatest advances have been the result of large-scale coordination efforts between individuals with specialized knowledge who could see what were fundamental natural limitations versus solvable technology problems. The Apollo program took the collective effort of 400,000 engineers, scientists, and technicians across 20,000 companies to achieve safe human spaceflight to the Moon. Today, we can’t get any closer to mapping the human brain without connectomics researchers who specialize in microscopy techniques. It’s these people who (we expect) will invent better or faster or more efficient labelling, slicing, and expansion methods.

At a time when information was scarce and knowledge production was slow, depth was the most reliable path to progress. The world was largely driven away from being a jack of all trades, and focused deeply on being a master of one.

Yet, in the past <5 years, information access has become abundant and easier to prompt for than ever before. The cost of intellectual expansion has dropped dramatically, to about $20/month. And so far, this shift is proving to be much more meaningful for individuals than organizations. The rising use of AI tools isn’t only attributed to increased information access, but the ability for these tools to handle a meaningful portion of the execution layer. They’re not necessarily replacing specialization, but they’re compressing the time it takes a curious person to work effectively at the frontier of some adjacent field.

Take drug development. A synthetic chemist with a promising molecule has historically needed to outsource every step: a regulatory consultant to map the FDA pathway, a clinical advisor to design the trials, a market specialist to understand whether insurers would ever pay for it. Today, the same chemist can develop a working understanding of how drugs move through regulators, model a Phase I and II trial timeline, draft a reimbursement argument, and pressure test their assumptions against the competitive landscape. It’s not necessarily about achieving the level of a specialist, but positioning yourself to move through the field more effectively than you could have before.

We’re at a point where the return on investing in your peripheral domains has gone up enormously.

There’s genuine questions about who’s positioned to capture the returns of these tools and the anticipated future. The common case is well-established: the specialists at the top survive, because they’ve built enough foundational knowledge to use these tools to go further rather than be replaced by them. The entry-level cohorts are vulnerable. It’s potentially because AI can do their work, but also their external incentives don’t necessarily reward developing the foundational knowledge of their field that makes real innovation possible. That’s a different, harder problem.

There’s a second case I’m beginning to notice, that I believe will have a lot of value. The people who will build the fields of the future aren’t those who abandoned specialization, but those who build peripheral understanding around it. We anticipate a future with new things to do and to want, with an accelerated rate of discovery unlike anything before. The people who will contribute to the fabric of reality will be the ones who can pull on threads from adjacent fields and see further down the road in their own, as a result. We’re cycling back, in a way, to the time of natural philosophers.

The people I’ve met who seem positioned to build what comes next share this knowledge of adjacent territory, including a common feeling about the importance of accumulating context that didn’t have an obvious payoff at the time. In the moment, these experiences can feel like detours. I sense that we’re approaching an inflection point where that kind of peripheral knowledge stops looking like a distraction and will be massively rewarded.

This is starting to become the case with skills that initially seemed based on intuition. Technologists have begun chasing taste. They’ve recognized the importance of developing discernment for what’s worth building, funding, and scrapping. But taste emerges from across domains, not within one. Fashion understood this long before Silicon Valley’s technologists got ahold of it. A great collection isn’t about clothes. It’s a study of history, anthropology, and a political moment in which it’s presented. The designers who produced work that remains relevant to the zeitgeist are those who paid attention to things that had no obvious relevance to their craft.

There’s no map for which peripheral domains will matter to the future of science and technology, but I think Rick Rubin’s “look for what you notice but no one else sees” has merit and has become more possible to act on. Whether that’s science communication, philosophy, or an entirely different technical field, leveraging the right peripheral domains will allow you to engage with a changing and accelerating world. Where we’ve long rewarded specialization, there will soon be reward for range.