This post lays out the foundation of what we might want to change, in order to not end up in a worse situation than where we started from.

We want to change things in order to improve them, that much is obvious – unless we’re conducting sabotage, in which case we’re trying to change someone else’s system to make our own situation better in some way. But it’s underappreciated, how easy it is to make systems worse off.

As is often the case, the dear child has many names. Iatrogenics, harm caused by the healer is a major concern in medicine, expressed by the Hippocratic Oath (“first, do no harm”). When considering larger scale changes, economists talk about negative externalities. As Michael Sanders, executive director of the UK What Works Centre for Children’s Social Care put it in his Helsinki talk, “we wouldn’t run a drug trial that didn’t look at side-effects” (but we routinely run policy trials that fail to do this). Layperson’s parlance might be to talk about the unintended consequences caused by one’s well-meaning actions.

We can also use the term second-order effects. Here, the highlight is on what happens after the initial action. We could, for example, reduce variation in a system by aggressively putting out small forest fires, or protecting our children from all emotional shocks. The first-order effect is a calmer forest, or a child that feels a little bit safer. The second-order effect would be, when the debris that would’ve been cleared up by small fires, piles up and produces a fire of catastrophic magnitude, or when the child finds themself unable to deal with anything that doesn’t go their way.

Can we make sure our change efforts won’t be wasted and turn against our goals? Probably not entirely. But happily, we have some guidelines as to what should and what shouldn’t be meddled with. To understand the logic, we need to make a small foray in complex systems science.

A complex system is some delineated part of the universe, which contains many interacting small-scale parts, that produce emergent and often unexpected large-scale phenomena (see slides V-VII of this course). Examples range from ants building nests without central control, to people semi-accidentally building societies, and rain forests emerging from the cycle of myriad life forms.

There are some good news and some bad news, and I’ll start with the latter. The big problem for conventional research into these systems is, that each part is interdependent with the others, and you can’t just take the system apart, measure each individual part, and add the effects together to get the full picture. If you could, you’d be able to perform so-called Gaussian analyses (where you can simplify by assuming large deviations from the mean quickly become practically impossible), and you wouldn’t need to worry about runaway cascades or domino-effects. Now, the good news is that all complex systems share universal features, so we can learn from one system and—to an extent—generalise to another.

What does this have to do with anything? Well, thing is that if you try to fix something which isn’t all that broken, it’s difficult to ascertain that things will eventually turn out better than they were: The interconnectedness ensures that things are probably too complex to get a full grasp of. Like any organism, we have also coevolved with the environment, and (as a great personal disappointment to yours truly) divergence to modern fads leads to long-term problems.

In this kind of a situation, there are some rules of thumb (or “heuristics”) regarding change efforts:

1. We want to stop taking risks that have a miniscule—or so we think—chance of a truly unacceptable outcome.
2. We want to increase the amount of choice we have in the future, because it’s impossible to know in advance what the best course of action is.
3. In order to change the system, we first need to destabilise it, and when it’s destabilised, care must be taken so that it won’t run off in an unfavourable direction.
4. Removing something is safer than adding something.
5. Removing new or modern things is safer than removing old or present-in-evolutionary-history things.
6. Whenever feasible, avoid centralisation (“don’t put all your eggs in the same basket”).
7. Efficiency and optimisation are great, when future is like the past. Redundancy and slack are great, when unexpected things can happen.

So, you might want to rather quit smoking, than start taking the newest dietary supplement pill marketed as “scientifically proven”. In the same vein, given the mess that nutrition science is, it is safer to ignore the latest diets and opt for fasting instead. Almost any modern person would also benefit from replacing sedentary time with physical activity – both light (e.g. slow walks) and vigorous (e.g. high-intensity interval training).

As a bonus, here are seven questions to ask yourself, when evaluating what you want to change in the first place (in no particular order):

1. ”Which actions will give me the most options/alternatives in the future?”
2. ”How do I have the most choices when it turns out I was wrong?”
3. ”How do I make failure less costly?”
4. ”How can I make many small bets (or try many times)?”
5. ”Which redundancies and buffers can I build in to de-optimise?”
6. ”Can I have more ’unnecessary’ slack in my life?”
7. “What may happen if I take no action?”

Tread carefully, but remember that the world moves, and thus staying put is also an active choice with consequences!

References and further reading

On managing complex systems:

Bar-Yam, Y. (2004). Making things work: solving complex problems in a complex world. Cambridge, Massachusetts: NECSI/Knowledge Press.

Siegenfeld, A. F., & Bar-Yam, Y. (2019). An Introduction to Complex Systems Science and its Applications. ArXiv:1912.05088 [Physics]. Retrieved from http://arxiv.org/abs/1912.05088

On dangers of dampening natural variability, and opportunities for striving under uncertainty:

Taleb, N. N., & Blyth, M. (2011, June). The Black Swan of Cairo. Foreign Affairs. Available from: https://fooledbyrandomness.com/ForeignAffairs.pdf

Taleb, N. N. (2012). Antifragile: things that gain from disorder (1st ed). New York: Random House.