An effective energy technology strategy has to balance between setting a stable long term framework for innovation, while also responding to more immediate changes in technology cost and performance. Over the last decade, rather than a steady progression along an established learning curve, PV costs and prices have been volatile, with increases or plateaus followed by rapid reductions. The paper describes, and considers the causes of, recent changes in PV costs and prices at module and system level, both international trends and more place-specific contexts. It finds that both module and system costs and price trends have reflected multiple overlapping forces. Established forecasting methods experience curves and engineering assessments have limited ability to capture key learning effects behind recent PV cost and price trends: production scale effects, industrial re-organization and shakeouts, international trade practices and national market dynamics. These forces are likely to remain prominent aspect of technology learning effects in the foreseeable future and so are in need of improved, more explicit representation in energy technology forecasting.