4.4.5.1 Predicting tipping points

In some circumstances, tipping points in climate and ecological systems may be preceded by specific statistical signals, termed early warning signals (EWS) (see Chapter 1.6). These provide some indication that a system is losing resilience and a self-propelling transition may be approaching. Chapter 2.5 discusses where these EWS may be applied to negative social-ecological tipping points, and here we expand upon this by considering how they may relate to positive social tipping points and illustrate this with a case study of the EV transition.

EWS are often observable as a consequence of critical slowing down (CSD), which occurs in a system as it loses resilience before a tipping point. When a resilient system with strong restorative feedbacks experiences some perturbation, it will return quickly to its equilibrium state (i.e. a healthy forest recovering from a drought). However, as the system loses resilience, these restorative feedbacks weaken, and the system takes longer to return to equilibrium following a shock. This changing response can be measured to indicate the system’s resilience, by measuring the declining return rate (Wissel, 1984). This change can also be measured over time with an increase in the lag-1 autocorrelation (AR(1)), in addition to an expected increase in variance prior to a tipping point (see Chapters 1.6 and 2.5 for further details of this method and other EWS).

While measuring EWS with empirical data is most common in ecological and climate systems, it is not exclusive to these domains and a number of studies have applied this approach to alternate systems, such as health, economics and online social discourse (Dakos et al., 2023). In health sciences, attempts have been made to identify generic EWS prior to disease re-emergence (Proverbio et al., 2022). Several studies have attempted to detect EWS prior to economic shock events, with varying levels of success (Tan and Cheong., 2014; Diks et al., 2019; Wen et al., 2018; see Chapter 2.5). Social media data has also been employed to detect EWS before transitions in online discourse (Pananos et al., 2017) and could be applied to online radicalisation (see Chapter 2.5). These studies often focus on negative shocks, where the shift occurring is to a less desirable alternate state, but it is also possible that these statistical indicators may be present prior to a rapid transition to a more desirable state.

As discussed in the rest of Section 4, positive tipping points may occur in different elements of social systems and across different nested scales. For example, in socio-technical systems, development of a technology may have positive feedback loops which allow it to scale rapidly, reduce in cost and improve in quality: thus becoming more accessible (Sharpe and Lenton, 2021; Farmer and Lafond, 2016; Lam and Mercure, 2022). Rapid changes in social behaviour or perspective may be required to enable this transition. In these complex systems it is likely that social and technical change will be interlinked, with each affecting the other. Consequently, for some systems it may be possible to measure changes in resilience within the social sub-system and in the technical or ecological sub-system. There are also likely to be exogenous shocks due to policy decisions or external economic factors which will show up in the system and may enable us to measure some element of its resilience. We sketch out these intersecting feedback loops as they may apply to the EV transition in Figure 4.4.8.

There are therefore two potential ways that we might measure the resilience of social systems; i) the return rate from a known perturbation or event or ii) the long-term changes in the resilience from a longer-term forcing on the system, which can be measured with AR(1). These approaches could be applied to multiple elements or indicators of these systems, either to detect decreasing resilience of an incumbent system, or to detect increasing resilience in a new, positive social or technological innovation. Here we refer to these indicators as EOIs.