@bhahne @infobeautiful
The simple answer might be that the IEA's interest is mostly in forecasting the large numbers, which are still the fossil fuels. For that, it doesn't matter if PV is 0.02% or 2% of the total, both are lost in the noise. So their PV model was just a placeholder, instead of making a difficult model that would still likely be wrong.
That might not work anymore for the next decade, but it has worked for a full decade after that Vox article.