Economic Modelling

 Economic Modelling is a good example of the Four Pieces Limit.

A highly respected SMH columnist makes the case for Semantic AI without mentioning it

SMH Article

We are unlikely to get economic modellers to work harder, but we may be able to get them to work smarter, by understanding their human limitations.

Humans, including modellers, no matter how industrious they are, have a Four Pieces Limit - they can't handle more than four things in play at once. So it is not surprising their output is pretty woeful on something complex. What to do? Use a machine that can read English and understand equations as equations - structures that modify the values flowing in their connections, no matter the direction. The machine can connect everything together, so a change in one area affects other areas, which cause change in the initial area. What machine does that? A machine running Semantic AI - not LLM, which is great for Search Engines, useless for much else. Humans use their Unconscious Mind to understand complex text - it is too complex for their Conscious Mind. You will see it when someone says something, but has no idea where the thought came from. The trouble with doing it unconsciously is that the unconscious part is hard to change - Ross complains about economists living in the past, and being very reluctant to change their mental models. There is a good reason why they do this, but Semantic AI blows it away, as it blows away similar limits in Law and Specification-writing.

There is a push on to have Complex Systems Engineers do economic modelling (we are CSEs) but it would have the same problem - people can only think about a small part of a complex problem at any one time - it needs a machine. No shame - particularly when you see the amount of wastage - billions (MRH-90), or hundreds of billions (F-35), or economic predictions that are far from accurate year after yeare - giving "safe" predictions is very different from giving accurate ones..

General website

Defence-related website