When MIT announced course 2.847—"Foundations of Dexterity Science," named for the study—it expected 60 enrollments. It got 1,900 and moved the lectures to Kresge Auditorium. When Seoul National University announced its program, the application portal failed within an hour. The University of São Paulo, ETH Zürich, IIT Bombay, the University of Lagos, and 209 other institutions launched dexterity science degrees, tracks, or departments this semester.
Two hundred fourteen universities. One semester. Historians of science, consulted for this article, could not find a precedent. Computer science took roughly twenty years to spread from three universities to two hundred. Molecular biology took fifteen. Dexterity science did it in under two years from the publication of the manifold paper—the fastest field formation in the recorded history of the academy.
The reason is embarrassingly practical: the field arrived with its dataset already built. A new discipline usually spends decades assembling the observational base that makes real science possible. Astronomy needed star catalogs. Genomics needed sequenced genomes. Dexterity science was born with the corpus—hundreds of millions of hours of verified, structured, real-world human manipulation data, openly licensable for research at rates the Council deliberately set near zero. ("Charge the robot companies. Not the grad students." The proposal passed with 97 percent.)
The result is a gold rush of low-hanging fruit that professional scientists describe with open giddiness. The manifold has been mapped in broad strokes, but its provinces are unexplored. Why does professional dexterity converge across cultures while amateur dexterity diverges? Why do the hands of the blind organize touch differently at the manifold level? What exactly is happening in the two-tenths of a second when a slipping wine glass is caught without conscious thought? Every one of these is a career, and every one of them is answerable, this year, with data that already exists.
The academic infrastructure is scrambling to keep up. Three journals launched this quarter. The first International Congress of Dexterity Science, planned for a 400-person conference center in Zürich, has relocated twice and will now fill a convention hall built for 11,000. The keynote speaker is not a professor. She is a heart surgeon from Tokyo whose suturing data underpins the field's most-cited paper, listed in the program by her preferred credential: her ☜handle.
Enrollment demographics are unlike anything else in the sciences. Applicants include roboticists and neuroscientists, as expected—but also physical therapists, sign language interpreters, chefs, and a notable contingent of master craftspeople in their fifties and sixties, arriving with thirty years of hand knowledge and a desire to understand what it is they know. The University of Bologna has waived its standard prerequisites for applicants with 2,000+ hours of high-quality-score corpus contributions. Other universities are copying the policy. Hands, it turns out, are transcripts.
An oversubscribed field, a bottomless dataset, careers of open questions, and a pipeline of students who arrive already carrying the data on their own bodies. Sixty years of robotics produced the block-stacking benchmark. Two years of the corpus produced a science.
Somewhere in this semester's 214 cohorts is the person who will answer the question every syllabus saves for the final lecture: not "how do hands move," but "why do they move so well?" The corpus is waiting. ☜