Known Unknowns

Assembling an Earth Model means to combine pieces of causal reasoning that we hold true (using whichever modeling technique: linear constraints, state differential equations, fragments of code, etc.) with assumptions, which are hypotheses that we want to evaluate or policies that we want to optimize. For instance, all the model mentioned in this paper are built on the causal models about climate change described in the IPCC reports. In the case of the Energy/Economy/Climate coupling, there are (at least) five “known unknowns”:


·       How much energy will be available in the future? At which costs? This question is well known for fossil fuels and is related to the number of accessible inventories. For instance, the introduction of shale oil and gas has changed our perspective between 2000 and now. On the other hand, the cost of extraction for future resources is difficult to foresee. This question also applies to renewable clean source of energy. We now have a better understanding of cost evolution (although it is a matter of debate) but our capacity to execute, from material resources such as metals for wind turbine to manufacturing capabilities, means that the rate at which we can deploy these renewable energy plants is a “known unknown”.


·       How much energy is needed and affordable for the economy at a given cost?  The energy intensity (amount of W.h, that is Watt x hour, to produce a dollar of GDP) is decreasing, but it is unclear to see how fast or how long this trend will last. If energy becomes rare (and/or too expensive), which activities will adapt (because they create enough value to afford a more expensive energy supply) and which ones will have to stop? The “hot question” of the need for energy subsidies – when a government helps some human activities to have access to a lower energy price – is part of this second issue.


·       How fast can we substitute one form of primary energy to another? A key factor to manage global warming is to accelerate the transition to clean sources of energy. The first question addressed our capacity to produce this clean energy, this third question addresses the capacity to switch to one form to another, because all sources are not equivalent because of energy density, mobility, intermittence, etc. This a complex question since the answer is different for each type of industry (Gates, 2022).


·       Which GDP growth can be expected from investment, technology, energy and workforce? Most integrated energy/economy/climate models are based on an implicit “economy growth engine”, which is then adjusted to reflect the lack of energy or the loss of productive capacities. What the economy growth trajectory would be without these impediments is a “known unknown” (mostly, the “natural rate of growth”). It is easy to calibrate that rate from what was observed in the past decades, but this is mostly an act of faith.


·       What will be the economical and societal consequences from the IPCCs predicted global warming? There are many unknowns here. First the amount of loss of productive capacities due to global warming impacts is a topic of debate, as shown by the previous section (it is the most differentiating factors of all the models derived from DICE that have been published in the past decade). Second, considering the catastrophic nature of the impact (Wallace-Wells, 2019), there are many other indirect impacts that will add to “capacity losses”. If the temperature rises above +2C, fear and pain may create all kinds of bifurcations from the “modeled path”.

 

These are “known unknowns”, as popularized by United States Secretary of Defense Donald Rumsfeld in a famous 2002 speech, in the sense that the issues are well understood and documented, but there is no consensus about what the answers might be. In the remainder of the paper, we call these “known unknowns” beliefs to emphasize the lack of consensus (and/or the variation of opinions over the past decades, as shown by the Energy resource example). We make these beliefs “first-class citizens” of the CCEM model, which means that we use a closed abstraction (most often a table or a single-parameter function) to represent the answer to the questions that we just listed. We want to make beliefs explicit in CCEM because we find that other studies with very different conclusions differ, not on their earth model principles, but on their unspoken hypothesis about some of these known unknowns. For instance, the main difference between the various DICE-related models presented in the previous section is foremost about the answer to the fifth unknown.