Predicting Economics and Evolution

2022 Jul 23

I came across an article which is about a problem in economics. There is a standard model of economics based on rational human behavior, but the model works poorly. To compensate, human cognitive biases are added to the model to better align predictions with reality. However, this does not make a better model; this actually demonstrates that the model is fundamentally wrong.

We don’t have a hundred biases, we have the wrong model

Astronomy

First the article tells the story of the scientific transition from the geocentric orbital model to the heliocentric model. The first model did not correctly predict movements of planets and other bodies in the sky. Replacing the geocentric model with heliocentrism eliminated this problem.

The old model needed a large collection of accommodations to be added to it before it would match to observations. They were called epicycles, and were individual compensations to fix the model. The model itself failed to make correct predictions and therefore it didn’t really explain anything. However, the new heliocentric model did not need individual adjustments and this meant that it was predictive for new observations and therefore that it had real explanatory power.

Economics

The article above goes on to argue that because the established economics model it is discussing is similarly non-predictive, therefore it is wrong! Addition of the biases (like epicycles) do not make a better model. These are just accommodations and show the fundamental failures of the model. The model needs to not be adjusted anymore, but be wholly replaced.

Biology

Biological evolution is a theoretical model with similar characteristics. note It claims to answer the question of the arrival of the fittest. However, by the evidence of experiments it only answers the question of the survival of the fittest. This still makes evolution quite useful, but its usefulness as a theory is not as was originally promoted.

Evolution has multiple meanings:

  • The meaning of just change over time is uncontroversial.
  • The meaning of change in species is also regularly observed.
  • The meaning of transmutation between species has very limited modern evidence.
  • The Darwinian meaning of the grand origin of all species from abiogenesis is not observed today and only seems is evidenced by historical data (based on certain metaphysical assumptions).

Mostly this last meaning is discussed here.

Darwin said his theory could explain how all species originated. The modern neo-Darwinian synthesis attributes this process to the area of genetics. Mutation and natural selection together move genetic code through the information gain that results in new body plans.

However, a big problem with this Darwinian theory of evolution is that it is usually not predictive. note Using evolution, we cannot predict what a species will change to. We can only identify its changes after they have happened and a new species has emerged.

A theory can have value for reasons other than its predictive power.

  • It may offer superior explanatory power.
    • However, the continually added accommodations to Darwinian evolution being needed to match evidence only reduces the theory's explanatory power.
  • It may have value over another theory because it is more parsimonious (simpler) than the other.
    • However, Darwinian evolution looks to me like a theory that is becoming increasingly less simple.

Without these it won't have alternative scientific value.

First, this is because unobservable deep time is required for the theory to function. Human lifespan is short compared to evolution’s context, so this evolutionary theory is primarily useful for creating a story about the past. This means that the theory isn’t very powerful scientifically. Science is about developing testable models, and since we are humans, this very slow model isn’t testable in our own brief future, so it isn’t even usable. note

Of course the predictability of future events is not the only important kind predictions. Aspects of the past may also be predicted. We could predict that we will find historical evidence of a certain type. These predictions in evolution include:

  • All evolutionary evidence pointing to a singular common tree of inheritance (because history only happened one way, therefore evidence understood must agree).
  • Evidence pointing to gradualistic changes between species (because mutation and natural selection can only supply gradual changes, and saltations are lethal).
  • Enough time for new species to emerge (because the gradual process takes time).

These are fundamental predictions and they all fail:

So Darwinian evolution also fails to predict essential and critical evidence of past events.

Secondarily, the interaction between the environment and populations of organisms results in fitness functions applied to the organisms. note A lot of the issue of predictability would depend on accurately understanding the functions. To the degree we can formally quantify these active fitness functions, then we can make predictions. To the degree we don’t have them captured as quantified forces, all we can do is (possibly) identify them after the outcome is known. Mostly, the situation is the latter because we just don't understand the functions deeply. This problem therefore makes evolution almost indistinguishable from just-so stories.

The problem with fitness functions is that tautologically they only answer a question of preservation. It is painfully obvious from what they are that they have no creative ability.

Creativity in natural processes, if it exists, must come from somewhere else. Conventionally this is attributed to mutations. However, mutations in the vast majority are detrimental. Even so-called neutral mutations are detrimental. If neutrals do not supply a fitness advantage, then they must supply baggage with an incremental cost. Since there is a large amount of redundancy in biological systems, the cost could be hidden for a long time. However, cumulatively the costs will outweigh the benefits.

Some small fraction of mutations will be beneficial, but vastly most of them are detrimental. We can think of this as a ratio of benefits over detriments. Because the quantity of degradations is much more than improvements, this raw ratio is always substantially less than 1. Then, natural selection is like a compensating factor greater than 1. The two are multiplied for the outcome.

However, natural selection has limited efficiency. The existence of “vestigial organs”, “junk DNA” and dis-utilitarian “designs” is obvious evidence of natural selection’s extended failure. If natural selection was effective, it should have killed off these problems and they should not exist now to be detected. Natural selection does the best that it can under the circumstances, but that first ratio of the two directions is very very bad.

So, the output of the mutation and natural selection process is insufficient to assure a cumulative ratio greater than 1. This means that the process as a whole cannot account for beneficial gains as seen in biological systems. Therefore this model completely fails.

These fitness functions (natural selection) are the most predictable part of evolution. Mutation, the other key part, is not predictable at all. For both of these, there are a whole collection of methods and effects to cover all the possible cases that could be in play. Developers of evolutionary theory continue to add to the collection. These additions do not improve evolutionary theory; each addition is an adjustment to the theory that weakens it. And as the linked article says this means we have the wrong fundamental model (Darwinian evolution).

 

The article argues that for economics the existing non-predictive model is fundamentally wrong. Like geocentrism it needs to be replaced. The same is true for the Darwinian evolutionary model. It is non-predictive, therefore it is fundamentally wrong. It needs to be replaced with a different model. It needs to be replaced with a non-evolutionary model.