In the interview with Rebecca Henderson, she mentions a phenomenon known as “Feedback Effects”. Upon hearing it, I thought it would have something to do with climate modeling; that models are perhaps intentionally altered in response to negative feedback. However, feedback effects are actually more similar to the concept of “unknown unknowns” discussed in Nate Silver’s book. Henderson explains that feedback effects are unforeseen outcomes of an event, and while they could be perceived as possible, they are difficult to predict. She gives an example of greenhouse gases that are currently trapped under ice sheets. If these sheets melt due to climate change, they will release an explosion of additional greenhouse gases into the atmosphere. This will in turn increase the rate of global heating, which will cause ice melting and rising sea levels to occur even faster than currently predicted. However, if we act before then, it is unlikely that the release of these gases will occur. Henderson describes these kinds of instances as “discontinuity”. While we can make predictions in a way that can be described as linear, our response to that prediction can go in a variety of non-linear directions, and feedback effects exist too far down these non-linear paths to foresee.
https://www.labxchange.org/library/items/lb:HarvardX:99683ddd:lx_simulation:1
I also thought that view on positive feedbacks was really interesting, and something I had not really thought about before -- and it makes total sense! Generally in mathematical models, we see negative feedback, because it is near the regions with negative feedback that the system is stable, and can be properly linearized. However, once we go into non-stable regions, which have some form of positive feedback, linearization becomes way more complicated. Depending on our time resolution, we really would just observe a discontinuity or discontinuities, until the system stabilizes again. It is honestly terrifying.