What EDM Reveals About the Twin Structure of Scientific Innovation
EDM is not just a better academic metric. It points to a deeper truth: real breakthroughs often emerge through parallel convergence, not singular genius.
A new paper on the Embedding Disruptiveness Measure (EDM) — a method designed to detect truly disruptive science — raises an old question: why do we remember breakthroughs as if they belonged to one person?
In recent weeks, much of my writing has focused on AI infrastructure, supply-chain leverage, and the hidden concentration of power beneath the semiconductor stack. But behind those industrial bottlenecks lies a deeper upstream question that has stayed with me: how do real breakthroughs actually emerge, and why are they so often remembered as the victory of a single name?
At first glance, EDM looks like a bibliometrics paper about a better way to detect disruptive research. But read more carefully, it points to something larger: many of the most important scientific breakthroughs may not be solitary acts of genius at all, but moments when an entire knowledge system becomes mature enough for several people to arrive at nearly the same answer.
Science is fascinating not only because it produces truth, but because it allocates glory.
We are taught to understand breakthroughs through heroic narratives: Darwin proposed evolution, Newton invented calculus, a Nobel laureate changed an entire field, and one paper altered the course of history. These stories are elegant, memorable, and institutionally convenient. They fit the way science is taught, celebrated, and archived.
But the real history of knowledge is rarely that clean.
Many important breakthroughs are not singular eruptions from isolated genius. They emerge when multiple researchers, in different places and under similar intellectual conditions, converge on the same turning point at nearly the same time. That is why a recent study on the Embedding Disruptiveness Measure, or EDM, is so intriguing. Its importance is not simply that it offers a new metric. It asks a deeper question: what kind of work truly redirects the future of knowledge, and what if such redirection is often not a single-point event at all?
A New Metric, but Also a Larger ProblemEDM attempts to measure whether a paper or patent merely extends an existing trajectory or pushes subsequent work onto a genuinely different path. Rather than relying only on direct citation links, it uses the broader citation network to compare the “past context” a work inherits with the “future context” it generates. The greater the distance between those two, the more disruptive the work is presumed to be.
At first glance, this sounds like a technical refinement.
In reality, it touches something much larger:
Our traditional ways of measuring breakthroughs may have been conceptually too narrow from the beginning.
Conventional disruption indices rely heavily on local, one-hop citation structures. That makes them easy to compute, but also limited. They often miss indirect effects, delayed influence, and multi-generational shifts in knowledge trajectories. Worse, they tend to compress the scores of most papers toward zero, weakening their ability to distinguish genuinely transformative work from ordinary continuation.
EDM matters because it tries to move beyond this local view.
But its most interesting implication, in my view, lies elsewhere: its handling of simultaneous discoveries.
Real Breakthroughs Are Often Not Isolated Peaks, but Twin PeaksThis is where the study becomes more than a bibliometric improvement. Traditional disruption measures struggle when two or more researchers independently arrive at comparable breakthroughs around the same time. In such cases, overlapping references and mutual citation patterns can dilute the very disruptiveness that should have been recognised. EDM, by using broader structural information, appears better able to detect these “twin breakthroughs.”
That matters because it challenges one of the deepest habits in the way we narrate science.
Take Darwin and Wallace. Public memory remembers Darwin. History books remember On the Origin of Species. But the emergence of evolutionary theory was not the product of one mind acting in a vacuum. It was made possible by natural history, geology, imperial-era specimen collection, population thinking, and a broader nineteenth-century intellectual environment. Wallace’s parallel arrival does not diminish Darwin. It reveals something more important:
Breakthroughs often happen when a field has become mature enough to produce more than one answer to the same question at the same time.
The same logic applies to cases such as the parallel discovery of the J/ψ particle or the independent development of the Higgs mechanism. These were not merely contests over priority. There were signs that an entire knowledge system had reached a phase transition.
In that sense, major breakthroughs are often not solitary peaks.
They are twin peaks, or even entire storm fronts.
If EDM truly captures this better than earlier measures, then its value is not merely predictive. It becomes interpretive. It helps us see scientific innovation not as isolated flashes of brilliance, but as structural emergence.
Nobel Prizes Matter, but They May Also Distort the Shape of BreakthroughThe study highlights EDM’s stronger performance in identifying Nobel Prize-winning papers and milestone publications. That is impressive. But perhaps the deeper message is not that EDM helps us find genius more accurately. It is that our existing narratives of genius may already be distorted by how prizes, institutions, and history compress collective transformations into singular names.
Nobel Prizes are extraordinary institutions, but they have built-in limits. They need identifiable people, bounded contributions, and a manageable story of credit. They can honour greatness, but they cannot fully represent the distributed, overlapping, and often simultaneous character of real scientific change.
That is why I find EDM intellectually interesting. It does not merely propose a better metric. It invites a more honest narrative.
It suggests that what we call “breakthrough” may often be less about heroic singularity and more about structural readiness. Data infrastructures mature. Instruments improve. Conceptual tensions accumulate. Failed attempts reveal the right questions. And then, suddenly, multiple researchers arrive at nearly the same place.
Not because one copied another.
But because the knowledge system itself had become ready.
The Policy Temptation — and the RiskThere is, of course, a policy temptation here. If a measure can identify genuinely disruptive work more effectively, funding agencies may want to use it to guide scientific investment. That is understandable. But it is also dangerous.
Once disruptiveness becomes quantified and institutionalised, it risks becoming another target for optimisation. Researchers may begin to perform for the metric rather than pursue important work whose value unfolds slowly, cumulatively, or infrastructurally.
Science does not advance through disruptive work alone.
It also depends on integrative work, careful validation, incremental refinement, and patient consolidation.
So EDM may be valuable as a lens.
It would be far less valuable as an autopilot.
Why This Matters Even More NowIn today’s era of AI, massive compute, large-scale data infrastructures, and increasingly industrialised research systems, this pattern of “twin breakthroughs” may become more common, not less.
As knowledge systems mature faster, and as more researchers work at the frontier with comparable tools, comparable datasets, and comparable questions, the distance between one “first mover” and another may narrow. Under those conditions, the old lone-genius story becomes even less adequate.
The more important question may no longer be who got there first.
It may be whether we are willing to recognise that real scientific progress is often the visible surface of a much deeper collective accumulation.
What Matters Most Is Not EDM ItselfIn the end, the most important takeaway is not the metric itself.
It is the deeper picture of innovation that the metric hints at:
Real breakthroughs are often not one person’s victory. There are moments when an entire knowledge system has matured enough to allow several people to reach the same answer at nearly the same time.
That may be less romantic than the lone-genius myth.
But it is probably closer to the truth of how science actually moves.
Closing PunchlineReal breakthroughs are often not one person’s victory.
They are what happens when a knowledge system has finally become ready.
This essay sits adjacent to my recent writing on AI infrastructure and supply-chain power, but turns to a more upstream question: how scientific breakthroughs are formed, recognised, and remembered.
About the AuthorSinclair Huang is an independent researcher and advisor focused on AI infrastructure, industrial strategy, and the structure of technological power. His recent writing explores how value, leverage, and bottlenecks emerge across semiconductor supply chains, capital cycles, and scientific innovation.
More at sinclairhuang.org