The NY Times has published a fun "Room for Debate" focused on climate change fiction and the role of such work in shaping public opinion. To say something new here, let's think about this topic through the lens of the modern field experiment literature. I will embrace the essential heterogeneity model.
Suppose that climate change authors could randomly assign their books to adults and that adults were required to read them. Note that this is a typical field experiment research design. The random assignment of books removes concern that readers "self select" what books to read such that a Conservative Republican only watches Fox News. The random assignment of the book means that this selection concern vanishes.
Suppose that Mr J and Mr Q and Ms M are randomly assigned book #3. Do they change their worldview because of their exposure to this treatment (the book)?
Suppose that Mr J. views all of climate change as a big hoax and simply giggles as he reads a liberal Doom and Gloom climate change fiction book while Mr Q's imagination is expanded as he thinks about scenarios that never cross his mind before. Suppose that Ms M is a lover of all things Al Gore and eagerly reads the book.
Note the essential heterogeneity here. J Q and M are three different types of people. They know their worldviews. To an economist, only Q matters. He is the only one of the 3 who is "at the margin". Mr. J dismisses climate change as a hoax. Ms. M already is a true "believer".
To a statistician, the key issue is what % of the population resembles Mr. Q? For this "marginal" group, how responsive are they to new information? If this share is small and if this group tends not to update its priors, then such climate change fiction can't change the world because opinion is already set.
To really make a difference with their message, the climate change literature experts must target the Q sub-group and figure out how to offer them a price discount for reading these books.
The same issue arises for the anti-carbon tax coalition. Mr Q. is the swing voter. They need to target such guys and get their message out to him.
Returning to the field experiment design, the book seller does not know who in the population is J, Q or M and so the seller gives out 33% to type J , 33% to type Q and 33% to type M.
The statistician observes the population change in environmentalism and this equals:
.33*change in environmentalism of group Q given that each Q read the book.
Note that for the remaining 66% of the population who are types J and M that the treatment (reading the book) has no effect on them because they have already made up their mind.
Without targeting the treatment to the subgroup who is most susceptible to its influence, the book seller has achieved a low impact because most people are not affected by the treatment.