AI traces 25M academic lineages across 9 centuries
Leibniz dominates 53.4 downstream academic paths vs 5.3 average in a 9-century AI map.
A new study reconstructs roughly nine centuries of mentor–student relationships in mathematics using Wikidata data from the Mathematics Genealogy Project and MacTutor Archive. From about 470,000 mentor–student assertions, the authors built a directed graph of 372,853 people. Using all 64 historical Fields Medalists as a starting set, backward tracing uncovered about 25.5 million distinct paths spanning 57 generations. The analysis highlights a striking hourglass structure around Leibniz: thin upstream connections averaging 5.3 paths per node, thick downstream connections averaging 53.4—a nearly 10-to-1 ratio. Newton, by contrast, appears on only four of the 64 lineages. The researchers also report that 54 of the 64 lineages converge on the same five 12th- and 13th-century Islamic and Byzantine scholars, ending at an 11th-century boundary they call the Monastery Wall. Notably, the traversal engine is algebraically reversible, meaning every ranking decision it makes can be reconstructed and audited after the fact.
- AI mapped 25.5M academic mentor-student paths from 372,853 scholars across 9 centuries using 470K Wikidata assertions
- Leibniz's academic influence shows a 10:1 downstream-to-upstream path ratio (53.4 vs 5.3 nodes) forming a bottleneck
- Traversal engine is algebraically reversible for auditability; 54/64 Fields Medalist lineages converge on five pre-12th-century scholars
Why It Matters
Reveals Leibniz’s outsized academic influence and provides auditable AI tools for studying centuries of knowledge transmission.