On: Embracing Complexity for Systemic Interventions
Transmission Series Ep. 5
Episode: 31
Date: May 5, 2020
Evolutionary biologist.
Review of 5 Transmission Essays.
- John Krakauer and Michelle Carlson on COVID and Spiraling Frailty Syndrome
- Stefani Crabtree on What History Can Teach Us About Resilience
- Van Savage on The Informational Pitfalls of Selective Testing
- David Tuckett, Lenny Smith, Gerd Gigerenzer, and Jürgen Jost on Making Good Decisions Under Uncertainty
- Cristopher Moore on The Heavy Tail of Outbreaks
Key Takeaways
- Complex problems.
- Broad averages are misleading.
- Outcomes can be much more volatile than expected.
- Need to understand local variation.
- Need to present all possible outcomes.
- Complex problems require complex interventions.
- We tend to collapse causality into single causes.
- Important problems tend to be complex and rarely have simple causes.
- We poorly understand how to resolve complex problems.
- We typically don’t look for complex solutions as we prefer simple ones.
Key Concepts
Cristopher Moore on The Heavy Tail of Outbreaks
- Danger of averages.
- Consider the variance, as well as the mean.
- System averages are especially misleading if there is large local variance.
- Example in COVID-19: impact of super-spreaders on R0.
- Averages can hide heavy tails.
- Probability of outliers can be much higher than you anticipate.
- See “The End of Average“.
Van Savage on The Informational Pitfalls of Selective Testing
- Constraints: testing capacity.
- Reporting the number of tests instead of incidence of cases.
- Can happen when true cases outnumber the testing capacity.
- Reporting the growth in testing instead of the growth in cases.
- Can happen when both testing and incidence of cases expands rapidly.
- Reporting the number of tests instead of incidence of cases.
- Solution: massively expand testing and/or test randomly.
David Tuckett, Lenny Smith, Gerd Gigerenzer, and Jürgen Jost on Making Good Decisions Under Uncertainty
- Decision making under high uncertainty.
- Data is not sufficient to support probabilities.
- Solution: present all possible outcomes: understand potential variance.
- Counter-factuals, simulations, null models.
- Recognize the limitation of narratives (subject to cognitive biases).
John Krakauer and Michelle Carlson on COVID and Spiraling Frailty Syndrome
- Complex problems require complex solutions
- Example: frailty:
- Complex syndrome that increases after age 65.
- Physical activity provides complex solution.
- Less costly than drugs and other mono-therapies.
- See “Interventions for Human Frailty: Physical Activity as a Model“.
- Example: frailty:
- Difficult to implement, convince people.
- We are energy-minimizing cognitive systems.
- If we can find dominant mono-causal explanations, we prefer them.
- Medicine needs to evolve to allow for more complex interventions.
- Examples: physical activity, sleep.
Stefani Crabtree on What History Can Teach Us About Resilience
- All civilizations have experienced and learnt how to cope with pandemics.
- Often, pandemics have contributed to the downfall of empires.
- Rarely the case that interesting or dramatic phenomena have simple causes.
- Always tempting to collapse causality into a single cause.
- Example COVID-19:
- Outcomes are not only driven by virus.
- Pre-existing inequalities hugely affect severity in pathology.
- Complex causality requires complex control or complex intervention.
- Things are connected in subtle ways.
- There are non-linearities.
- There are different phases and critical transitions.
- We have a poor understanding on how to intervene into complex systems.
- Last thing we want to do is pretend that they are simple systems.