Human interventions to control the COVID-19 pandemic have created complexity. People’s reaction to the interventions may be counter-intuitive and seemingly irrational – witness Donald Trump’s behaviour.
There is delay in obtaining information on the virus, the disease and the pandemic – therefore we are always playing catchup.
Human societies are inherently policy resistant and so the interventions are likely to be delayed, diluted or defeated. For example, the prohibition on sale of cigarettes has replaced by a black market: consequently 95% of smokers are still smoking … and the government revenue from sale of cigarettes has declined.
Mindsets and both conscious and unconscious biases in the scientists and decision-makers influences the information they seek, the way the gather it, their interpretation and how they make decisions. Can NDZ reassure herself and us that her recommendations and/or decisions on sale of tobacco products and alcohol are not influenced by her mindset and biases but only by the empirical evidence on what will ‘flatten the curve’?
The coronavirus that causes COVID-19 is new. We do not understand well the virus; nor its behaviour in human beings; nor the human body’s responses to the virus. For example, we do not know if people who have recovered from the disease have lasting immunity – and if they do not, then, for example, the notion of herd immunity in nonsensical.
The pandemic is a ‘complexity mess’: every part of the problem interacts with other problems and is therefore part of a set of interrelated problems; simple cause and effect are replaced by effects having multiple causes and the effects are also causes of other problems.
As a consequence of effects becoming causes, the nature of the mess changes as we try to deal with it.
In this situation, large scale interventions may have small effects on the pandemic and seemingly small interventions may have big effects. For example, enforced lockdown of an entire society may not stop exponential spread of the disease or even adequately ‘flatten the curve’. But mutual care for one’s fellow human’s being’s health might.
All models are wrong. All models of real-world complex systems are faulty imitations of reality – anyone who has drawn up and had to perform against a business budget understands this.
This does not mean that models are of no utilitarian value, on the contrary they can be useful, especially in revealing what we do not know or understand.
However, we need to be humble about our knowledge and understanding as we engage with complex problems. Be able to provide different and valuable perspectives on our data. As we learn, we may also need to change our decisions in the hope that the adaptation better will allow us to deal effectively with the present and the unknowable future.
A few key points on models and forecasts were highlighted in Dance with Chance:
- Extrapolating historical patterns into the future will be inaccurate; the future is never the same as the past;
- Complex statistical models fit historical data well, but don’t necessarily forecast well;
- Simple statistical models often outperform complex models because they just forecast the basic trend forward;
- Human judgment is a notoriously bad form of prediction, even when the judgments are made by “experts”;
- The real extent of uncertainty is not captured in either statistical models or human judgments; modellers are surprised by the large errors and events not captured in their forecasts.
There is no need to be fatalistic, despite the challenges of truly complex problems such as COVID-19 pandemic. We might not be able to predict with accuracy the occurrence of chance events or how they will evolve. That does not mean we must be fatalistic. There are ways of becoming prepared for an uncertain future:
- Accept that You Operate in an Uncertain World.
- Assess the Level of Uncertainty you Face – Realistically.
- Augment the Range of Uncertainty you’ve Realistically Assessed. A simple but practical way of achieving this is to develop a range of certainty around any prediction you make. Human beings are just not good at imagining just how bad or well the future might be. Therefore, the authors of Dance with Chance suggest taking the difference between the largest and smallest value in your data set and doubling it to set a range around your predicted value. This should cover about 95 percent of all possible outcomes.
- Prepare: The next step is to consider how one might prepare for such extremes of change. Prepare for the worst and post-mortem critics will say you over-reacted; prepare for the best and ‘you should have known it would be worse’. We are already witnessing this as COVID unfolds, as more data become available and as prediction models are refined.
George W Bush was the butt of may jokes, but he prepared the USA for an influenza epidemic in 2005; his administration’s detailed plan last was updated in 2017 – it is why there is a federal stockpile of equipment for a respiratory epidemic! The plan was not implemented by the Trump administration. Bill Gates made similar proposals in 2015 after the W Africa Ebola epidemic – he was ignored. Here is a view from Fareed Zachaia on what COVID will mean: are we prepared …. For the worst or best possible outcome?
| Roger Stewart. April 2020 |
