Tuesday, August 18, 2020

Problem-solving approaches: clock-fixing vs cloud-fixing

 

“Car breakdown? Internet not working? Boss or spouse upset? Garbage everywhere? Long commute times? Corruption? Poverty?” How to solve it? Thus begins the blog I wrote six years ago titled “Four approaches to problem solving”. Today, we could add Covid to the list of problems. The four approaches presented in the blog are system-centric, problem-centric, solution-centric, and solver-centric. Which approach is applicable in which context? I suggested that as the social complexity of the problem increases, the role of solution-centric and solver-centric approaches increases. Recently, I came across a framework that sheds some light on this hypothesis. It involves understanding the difference between clock-fixing and cloud-fixing. Before understanding the fixing part, let’s understand what the analogies of clock and cloud mean.

Dividing the world between clocks and clouds is attributed to Karl Popper, the philosopher of science, who wrote the article “On clocks and clouds” in 1966. Clocks are predictable while clouds are unpredictable. When a clock malfunctions it is possible to dismantle it into smaller parts, diagnose the problem, replace the malfunctioning parts, and put it all back. In contrast, if a cloud doesn’t give enough rainwater, we don’t know how to fix it – at least not yet.

As a system, a cloud is considered more complex than a clock2. What makes a cloud more complex than a clock? One characteristic could be its degree of openness. A clock is a relatively closed system i.e. its behaviour doesn’t depend much on the environment while a cloud is a relatively open system i.e. its shape and content may be undergoing continuous change due to the interaction with the surrounding environment. A few more properties of complex systems are Non-linearity (Can a small change cause disproportionately big impact?), Emergence (e.g. a cloud can suddenly turn into a tornado but a clock can’t transform itself spontaneously). There is no precise definition of a complex system but we can get some idea by contrasting clocks and clouds (see complex system page from Wikipedia for more details). Could this distinction between cloud vs clock or clock-ness vs cloud-ness of the problem help us decide on the approach of problem-solving?

When a person undergoes hip-replacement surgery, the process looks closer to clock-fixing. A part of the hip-bone gets replaced and the patient is able to walk again with a high probability. However, when a depression patient undergoes a psychotherapy session, it looks closer to cloud-fixing. You don’t know how many sessions he might need and even after that, it may not work. When a few second long video goes viral on the Internet and mobilizes huge crowds across multiple cities into protests, it certainly looks closer to clouds turning into a tornado. When someone says I want to control thoughts through a brain-machine interface, doubts get raised as to whether a clock-fixing approach is being applied to a cloud-like system2.

Instead of putting each problem into a clock vs cloud bucket, how about if we look at different dimensions of the problem and solve it using an appropriate approach? Let’s take Covid-19 as an example. One could just take maximizing sanitizer usage as a goal, especially in shops and malls. And then apply a system-centric approach, break-down the process of entering a shop, and introduce a mandatory step of using sanitizer at the entry point. Alternately, one could apply the problem-centric approach, do the causal analysis, work towards the Covid vaccine, or in the interim find drugs with sufficient efficacy. Causal analysis can also be carried out through computer simulations of networks and infer probabilities of various causes in a regional cluster. 

We could move closer to cloud-like dimensions and ask questions like, “Why is certain population not even susceptible to Covid despite exposure?” And then one could look at bright spots “people who are tested negative despite sustained Covid exposure” and see if that data gives any clues. Finally, we could just look at the depression wave that is coming as a side-effect of the pandemic. And look at psychotherapeutic or introspective approaches. This is where the dimension gets closest to the cloud. Awareness of the nature of the problem may help us in predicting the chance of success.

In short, we saw that knowing the nature of the challenge – whether it is closer to a clock or a cloud, may help us follow an appropriate approach in creating the response. Alternately, we could look at the clock-like or cloud-like dimension of a challenge and try to respond appropriately.

Notes:

1. This analogy has been used by others to differentiate reductive vs non-reductive approaches of analyzing and fixing systems. For example, see Robert Sapolsky, Stanford professor of biology, neuroscience and neurosurgery use this analogy in his May 19, 2010 lecture “21. Chaos and reductionism” at around 44:18.

2. Karl Friston, Professor of computational neuroscience at University College London, compares brain-computer interfacing to solve psychopathic problems to satellites perturbing weather and changing it meaningfully in the interview “Karl Friston: Neuroscience and the free energy principle: AI podcast #99 with Lex Fridman” at 39:12.