APPLICATION OF THE FENCE PARADOX TO REAL-WORLD POLICIES (thread)
1/ The Fence Paradox holds that, whenever a barrier blocks harm from small stressors, humans take more risks, increasing their exposure to large-harm events.
(image below)
1/ The Fence Paradox holds that, whenever a barrier blocks harm from small stressors, humans take more risks, increasing their exposure to large-harm events.
(image below)
2/ Some applications of the Fence Paradox:
- Student loans act like the Fence above, making it possible for students to take increased risks (incur debt)
- Painkillers hide symptoms allowing patients to continue their unhealthy behaviors
- (continues below…)
- Student loans act like the Fence above, making it possible for students to take increased risks (incur debt)
- Painkillers hide symptoms allowing patients to continue their unhealthy behaviors
- (continues below…)
3/ The ABS (🚗 Anti-Blocking System) prevents harm from small mistakes, inviting people to speed, leading to rarer yet more harmful incidents
(counterintuitively, data shows that the ABS *increased* the number of deaths at the wheel, though it reduced the number of incidents)
(counterintuitively, data shows that the ABS *increased* the number of deaths at the wheel, though it reduced the number of incidents)
4/ The ABS would work if, say, it slapped the driver every time it engages. Safety protections only work if they limit "big harm" while letting "small harm" pass.
Otherwise, they invite risk-taking which, in a non-linear world, inevitably leads to fatal consequences.
Otherwise, they invite risk-taking which, in a non-linear world, inevitably leads to fatal consequences.
5/ Protections work in the measure they cap fat-tailed consequences (increasing ergodicity),
and fail in the measure they prevent small consequences (preventing antifragilization and increasing risk-taking, ultimately leading to non-ergodicity, i.e. exposure to fatal risks).
and fail in the measure they prevent small consequences (preventing antifragilization and increasing risk-taking, ultimately leading to non-ergodicity, i.e. exposure to fatal risks).
6/ In large measure, policies should be judged based on whether they increase ergodicity (moral) or they decrease it (immoral).
Explanation in the quoted tweet
Explanation in the quoted tweet
7/ This means that, in general, policies work in the measure they cap fat-tailed consequences (e.g. widows get financial support) and fail in the measure they prevent small consequences (e.g. unconditional income).
The latter leads to undesired second-order effects.
The latter leads to undesired second-order effects.
8/ The best interventions indirectly protect the population by allowing them to model their risk-taking models to the real world.
The worst ones obfuscate the real world by absorbing small stressors, letting populations adapt to a non-representative subset of the environment.
The worst ones obfuscate the real world by absorbing small stressors, letting populations adapt to a non-representative subset of the environment.
9/ Indeed, when "fence policies" prevent harm from small stressors, populations adapt to the lack of harm, undergoing fragilization and becoming more vulnerable to rare events
10/ I talk about the Fence Paradox, policies and fragilization in my book "The Power of Adaptation" gum.co/powerofa
11/ One issue of "fence policies" is that proposing them sends stronger signals of "protecting the population", even if the net effect is negative.
Doing the former opens a policy-maker to attacks from the intolerant minority for which signaling is more important than solutions
Doing the former opens a policy-maker to attacks from the intolerant minority for which signaling is more important than solutions
12/ (For many intolerant minorities,
signaling from authorities → legitimacy → relevancy & power,
whereas solutions → lack of problems → lack of relevancy.)
signaling from authorities → legitimacy → relevancy & power,
whereas solutions → lack of problems → lack of relevancy.)
13/ A second issue is that 1st order interventions which provide a false sense of security are more expensive (→ policy makers receive more budget → more power) & stronger signaling (→ more votes & loyalty).
The status- & power-seeking favor them spectator.co.uk/2019/10/plumbe…
The status- & power-seeking favor them spectator.co.uk/2019/10/plumbe…
14/ A third issue is that such 1st order interventions (aka "Fence Policies") work in the short-term, i.e. as long as the sample is small (before extreme events have a chance to surface).
Therefore, with a small sample, they seem "rational improvements" (sigh).
Therefore, with a small sample, they seem "rational improvements" (sigh).
15/ For the three reasons above, centralized decision-makers are exposed to an environment where "fence policies" appear to make sense.
Moreover, they often lack the skin in the game to suffer the long-term consequences. Hence terrible policies.
Moreover, they often lack the skin in the game to suffer the long-term consequences. Hence terrible policies.
16/ How to solve the problem?
Re-establish skin in the game. Insist to metaphorically slap decision makers (and receive a slap yourself) every time a small stressor hits, and only seek protection from fat-tailed events.
Look for fractal organizations (where slaps propagate).
Re-establish skin in the game. Insist to metaphorically slap decision makers (and receive a slap yourself) every time a small stressor hits, and only seek protection from fat-tailed events.
Look for fractal organizations (where slaps propagate).
