The sections above provide evidence that PV has improved through both the accumulation of incremental improvements to existing designs, as well as through larger steps, that involved more substantial changes to existing designs and processes. Learning curves are much more adept at characterizing the former than the latter. Nonincremental changes are difficult to predict; they sometimes appear to occur as random serendipitous events. But we know that important technological advances occur. If forecasting over several years or decades, we are pretty sure to be wrong if we exclude the possibility ofhard-to-predict nonincremental changes. If we are to more reliably inform decisions about investments and policies, predictions of future costs need to somehow incorporate insights about important, nonincremental technological changes. In particular, we are interested in when they are likely to occur, how big an impact they will have on costs, and conversely the consequences on costs of a sustained period without such changes.
A recent Master’s thesis sought to inform our ability to model such changes by assessing the historical record of them. Husmann (2011) studied historical silicon PV breakthroughs under the expectation that past breakthroughs can illustrate the causes and effects of nonincremental changes on a still-developing technology. This section describes a novel methodology for identifying nonincremental changes (“breakthroughs”) and provides some preliminary results. Ultimately, the identification and evaluation of specific technical improvements will be useful for informing the type of modeling described in the next section.