Spinning sheet metal parts successfully requires the design of a complex toolpath to avoid workpiece failure. Notwithstanding decades of research, industrial CNC spinning of new parts today still relies on trial and error, which results in considerable time waste and does not guarantee optimisation of the toolpath. Yet, hand spinning artisans in non-industrial settings can successfully spin new parts first time thanks to the skill accumulated in years of practice. So this paper asks: is there a way to capture and parameterise human skill to drive research into automation? What rules do hand spinners follow? A haptic spinning system is implemented to capture and parameterise the skill of six spinning artisans. Their speech and actions are recorded in a set of over 70 experimental trials and a database of toolpaths is created including information on the trajectory of the roller, applied forces and shape of the workpiece, alongside their speech. The database is analysed, and seven principles for toolpath design are formulated: 1. Take small bites, 2. Stay on the mandrel, 3. Use forward and backward passes, 4. Push in the right place, 5. Go at the right speed, 6. Keep the flange at the right angle, 7. Use a draft. Quantitative toolpath parameters are developed to parameterise the first six principles: the average plastic strain in each toolpass, the fraction of workpiece shape on the mandrel, the fraction of backward passes, the position of the force peak, the feed ratio and the concavity of the toolpass. In trials on thinner and larger blanks than previously investigated in the literature, these parameters appear to be a good predictor of success. This suggests that displacement control alone is not sufficient to automate toolpath design, and that new parameters relating to force and shape control must be employed.