Two scientists sitting on stools in an interior wide shot of Chemical Laboratory, part of the Central Cancer Research laboratories. Credit: National Cancer Institute
The Simple Metric Mistake That’s Been Costing Scientists Credit
In A Nutshell
- A new tool called the Embedding Disruptiveness Measure maps how research papers influence each other, going beyond simple citation counts to catch breakthroughs the old method missed.
- The most common way scientists measure “disruptive” research can completely miss it when two teams discover the same thing independently and cite each other.
- Researchers tested 80 pairs of highly cited papers and found 64 of them, or 80%, were cases of separate teams reaching the same discovery at the same time.
- Famous scientific “multiples,” like Darwin and Wallace both discovering evolution or two teams simultaneously finding a new particle in 1974, show how easily credit can get lost.
Sometimes, two scientists on opposite sides of the world crack the same problem at nearly the same time, and only one gets the credit. History tends to remember whoever got there first, or had the bigger spotlight, while equally important work fades into obscurity. Now, a team of researchers has built a new way to measure just how influential a paper really is, and in doing so, uncovered overlooked breakthroughs happening simultaneously, right under everyone’s noses.
Published in the journal Science Advances, the study tackles a deceptively simple question: how does anyone measure whether research is truly disruptive, not just important or widely cited, but work that redirects an entire field? Researchers Munjung Kim, Sadamori Kojaku, and Yong-Yeol Ahn argue the tools scientists have used to answer that are flawed, so they built a new one. Tested on massive publication datasets, it confirmed well-known breakthroughs and surfaced simultaneous discoveries traditional methods had missed entirely.
Correctly identifying who made a field-changing contribution shapes Nobel decisions and how the public understands scientific progress. If commonly used disruption measures can miss certain breakthroughs, that’s a problem worth fixing.
Citation Tallies Hide a Paper’s Real Impact
A paper can rack up thousands of citations without changing the course of its field. It’s the difference between a popular song everyone hums for a summer and an album that reinvents a genre. Both get attention, but only one reshapes the territory.
Previous attempts to measure disruptiveness relied on tracking how papers cite one another. A truly disruptive paper gets cited instead of the older work it built upon, with the field moving on to treat it as the new foundation. An incremental paper keeps getting cited alongside the work it referenced, because nothing has fundamentally shifted.
Scientific influence, according to the study’s authors, doesn’t travel in straight lines. A paper’s ideas often ripple outward through several layers of later research, so a paper’s true impact can show up in work that never directly cites it. That kind of ripple effect is real but hidden from standard citation-tracking. Discoveries are also often spread across multiple papers rather than one publication, making them harder to detect.
The New Tool Correctly Flags Nobel-Level Breakthroughs
To get around these blind spots, the team built a new approach. Instead of counting which papers cite which, their method maps papers into a mathematical space based on broader patterns in how knowledge flows through the literature. Papers that fundamentally redirect a field show up differently here than papers that simply add another brick to an existing wall.
Applied to large-scale publication data, the new tool, called the Embedding Disruptiveness Measure, reliably flagged papers already recognized as important, including Nobel Prize-winning research. That validation matters: a new tool must confirm what experts already know before it can be trusted with what they don’t.
Beyond confirming the usual suspects, the measurement also identified “simultaneous disruptions,” cases where multiple teams independently produced breakthrough work at the same time but the standard index missed some of them. When two rival papers cite each other, as often happens when competing teams learn of each other before publication, that older index can swing wildly, rating the same paper as fully disruptive or not at all depending on a single citation link. Known landmark papers weren’t flagged as disruptive under that older measure, even though scientists considered them field-changing.
The New Measure Flags 80% of Tested Rival Discoveries
Simultaneous discovery is one of the most fascinating patterns in science history. Charles Darwin and Alfred Russel Wallace independently developed the theory of evolution. Isaac Newton and Gottfried Wilhelm Leibniz both invented calculus. Sociologists call the phenomenon “multiples”: when the right combination of prior knowledge and unanswered questions has accumulated, people often converge on the same answer independently.
Recognizing simultaneous breakthroughs after the fact is surprisingly difficult, though. Citation patterns consolidate attention on whichever paper gains early traction, while the parallel discovery languishes with fewer references.
This new measurement appears to cut through the distortion. In one test, researchers manually examined 80 highly cited papers the method had paired closely together. Sixty-four, or 80%, turned out to be simultaneous or closely linked discoveries, including the 1974 discovery of the J/psi particle, announced independently by two teams the same day in the same journal issue.
Better Attribution Would Correct Decades of Uneven Credit
If tools like this reveal more simultaneous discoveries than standard measures capture, that says something about innovation: breakthroughs may emerge not from lone genius but from a field’s collective knowledge reaching a tipping point. Neglected contributors to major discoveries have sparked controversy before, and measures of scientific impact shape how those debates over credit play out.
Even so, the authors caution that “disruptiveness” isn’t one uniform quality. The method can’t reliably catch simultaneous discoveries when papers receive very few or no citations, and the two measures may capture different facets of influence.
Science likes to think of itself as a meritocracy where the best ideas win. The reality is messier. Credit flows unevenly, attention concentrates on a few well-positioned papers, and genuine breakthroughs slip through the cracks. A more robust measuring tool won’t fix all of that, but it offers a clearer view of work older measures miss.
Disclaimer: This article is based on findings published in a peer-reviewed journal. It is intended for general informational purposes and reflects the study’s methods and conclusions as reported by its authors. It does not constitute an endorsement of any individual researcher, institution, or scientific claim beyond what the cited study demonstrates.
Paper Notes
Limitations
The authors note that their method has limitations. Because the Embedding Disruptiveness Measure relies on citation patterns, it cannot reliably identify simultaneous discoveries when the papers involved receive very few or no citations, since those papers are excluded from the underlying analysis entirely. The authors also caution that measuring how disruptiveness changes over time is computationally demanding with their approach, and that “disruptiveness” itself is not a single, universally defined quality, meaning different measurement tools may capture different aspects of scientific influence.
Funding and Disclosures
This work was supported by the Air Force Office of Scientific Research Grant FA9550-19-1-0391 and by the National Science Foundation Grant 2404109. The authors declare no competing interests.
Publication Details
The paper, titled “Uncovering simultaneous breakthroughs with a robust measure of disruptiveness,” was authored by Munjung Kim, Sadamori Kojaku, and Yong-Yeol Ahn. It was published in Science Advances (Vol. 12, eadx3420) on April 1, 2026, with the DOI: 10.1126/sciadv.adx3420.
