The article uses "new field" as a relative term -- a field that the researcher is not an expert in.
> A question [BD14] had to address in this connection is: what constitutes a
brand-new direction? Again MathSciNet guides the answer. For each pair of the 73
Mathematics Subject Classification numbers, the authors worked out the likelihood
that a paper in one area is referred to by a paper in another area. Thus, for instance, they see that 35 (Partial Differential Equations) is closest to 58 (Global Analysis)and 76 (Fluid Mechanics) but furthest from 08 (General Algebraic Systems) and 19 (K-theory). Borjas and Doran deem a topic brand-new if it is not among the 15 closest to the researcher’s original area. This is a conservative choice and probably underestimates the cognitive mobility
> A question [BD14] had to address in this connection is: what constitutes a brand-new direction? Again MathSciNet guides the answer. For each pair of the 73 Mathematics Subject Classification numbers, the authors worked out the likelihood that a paper in one area is referred to by a paper in another area. Thus, for instance, they see that 35 (Partial Differential Equations) is closest to 58 (Global Analysis)and 76 (Fluid Mechanics) but furthest from 08 (General Algebraic Systems) and 19 (K-theory). Borjas and Doran deem a topic brand-new if it is not among the 15 closest to the researcher’s original area. This is a conservative choice and probably underestimates the cognitive mobility