The research behind MAPS
The following research and figures are supported by academic studies from institutions like Michigan State University and professional analysis from the McKinsey Global Institute.
1. Range vs. Specialization in "Wicked" Environments
David Epstein's work draws on research by psychologist Robin Hogarth, who distinguished between "kind" and "wicked" learning environments.
- Kind environments (e.g. Chess, Golf): Patterns repeat and feedback is immediate — early specialisation and deliberate practice are highly effective.
- Wicked environments (e.g. Business, Science, Medicine): Rules are unclear, patterns do not repeat, and feedback is delayed. In these domains, generalists with "range" consistently outperform specialists because they can apply analogies and concepts across fields.
Source: Epstein, D. — Range: Why Generalists Triumph in a Specialized World (2019)
2. Polymathy and Success — The Michigan State Study
Research by Dr. Robert Root-Bernstein at Michigan State University found that high-impact scientists are significantly more likely to engage in "serious leisure" or artistic avocations.
- Nobel Laureates are 22 times more likely to be performers (actors, dancers, magicians), 12 times more likely to be writers (poetry, fiction), and 7 times more likely to be visual artists compared to average scientists.
- These diverse pursuits are not distractions — they are "tools for thinking" that foster the creative leaps necessary for scientific breakthroughs.
3. Cross-Domain Thinking — Geoffrey Hinton
Geoffrey Hinton, the "Godfather of AI," is a primary example of cross-domain innovation.
- Background: Hinton holds a degree in Experimental Psychology from Cambridge.
- Impact: He treated machine learning not as a logic-rule exercise but as a biological and psychological one — mimicking the human brain's neural networks. This cross-domain thinking pioneered deep learning and the backpropagation algorithm.
Source: Encyclopædia Britannica — Geoffrey Hinton biography
4. AI and the Specialized Labor Market
Reports from the McKinsey Global Institute document a massive shift in how work is automated.
- Automation potential: McKinsey estimates that 57% of US work hours could be automated with currently existing technologies (as of late 2025).
- The specialisation ladder: Junior-level specialist tasks — coding, basic research, data processing — are the most susceptible to AI automation.
- Adaptability: Career resilience now depends on AI fluency and human-centric skills like social-emotional intelligence and cross-domain reasoning, which remain the hardest for AI to replicate.
Source: McKinsey Global Institute — The Future of Work reports
5. The M-Shape Career Model
While the "M-shape" terminology is a contemporary professional framework, it is structurally supported by the academic concept of Polymathy.
- Structural advantage: Unlike T-shaped individuals (one deep skill), M-shaped individuals maintain two deep pillars. This allows for "match quality" — the degree of fit between a person's diverse talents and the needs of a complex role — which Epstein argues is the ultimate driver of career satisfaction and impact.
Source: Michigan State University — Root-Bernstein polymathy research