[wodderam]

This is what the field of complex systems and complexity economics tries to deal with but it doesn't seem like complexity economics has taken off.

With economics too there is a problem that a correct prediction that is acted on will be gamed and become a self defeating seemingly "wrong" prediction.

I think the 2022 recession calls were probably a good example.

If everyone believes there is going to be a recession in the future and everyone acts to avoid the recession we can't know the counter factual in the economy of not predicting the recession to know what was the right prediction.

I have read this process referred to reflexivity but even a best selling trading book by George Soros himself didn't get this idea to stick in the popular mind.

It seems like prediction is not even the right word to use or we should have a different word for a prediction that can influence the outcome of what it is predicting. Then there is even strong and weak versions of this process that we haven't bothered to separate from the proper word prediction in the sense of weather prediction. That the prediction is basically independent of what it is trying to predict.

[api]

Complexity research hasn’t given us prediction for complex living systems so much as explain why it’s hard or in some cases impossible.

Unfortunately people keep buying linear models for this.

[mike_hearn]

Sure it took off. There's no conceptual problem with your predictions affecting people such that your inputs change.

This is why I was careful in my post to talk about public sector models (academic, government). There are lots of private sector actors successfully modelling aspects of society and the economy: actuaries, quants, even day traders. They're all acting on the predictions of a private model in a dynamic and even adversarial environment, yet they succeed. The problems here are not on the conceptual level, they're to do with the prevalence of sub-zero intellectual standards in the public sector. Nobody forces academics to make a big fuss over models that have never been validated, or which have massive CIs, or which generate so many or such long term 'scenarios' that they're effectively unfalsifiable.

All these things are the kind of pseudo-scientific modelling practices that result in firings or bankruptcies when done in the "real world". Government subsidies keep the low standards going in the public sector. Ultimately, ending this farce requires the social studies to be defunded, as is now happening in NZ and hopefully soon in the rest of the world.