He ran an energy model with built-in learning curves and showed that with exactly the same model (same equations, all same data inputs, costs, etc.) you can get radically different energy pathways with similar total costs:
I'd also be interested in the question posed the other way round: what is the value of "firm" clean generation in the presence of VRE, batteries and long-term storage?
E.g. how much does system cost reduce by adding nuclear/CCS etc. to wind+solar+batteries+hydrogen.
Either way, the system costs of all these options is in a similar ball park, which throws up the question:
What do we actually want?
What can we build quickly, with wide public approval?
- end of coal
- efficiency
- electrification
- renewables (he developed first hydro power)
- open data
- technological learning ("tendency of progress is to quicken progress")