Category: Science, Physics, and Philosophy of Science

Q&A: PREDICTION OR EXPLANATORY POWER IN TRUTH STATEMENTS?

—“Curt, Is it an important or necessary quality of a scientific theory to be predictive?”—

Not necessary – and that’s not what empiricist claimed, either. It must provide explanatory power and survive falsification (survive continuously). In other words, prediction is a form of justification. It’s survival from criticism(falsification), not confirmation that determines the truth content of a theory.

Prediction is just one way. But there is a difference between predicting a trend (aggregates) and predicting an individual actions(identities).

THE EPISTEMIC SEQUENCE

Free association -> hypothesis -> TEST (observation -> criticism) -> theory -> extended ‘social’ criticism -> law -> falsehood (increased parsimony)

Testimonialism is a higher standard of truth candidacy than that of ‘science’. It tells us that we must test our hypotheses for:

1 – categorical consistency (identity)

2 – internal consistency (logical)

3 – external consistency (correspondence)

4 – existential consistency (existential possibility)

5 – full accounting (account for externalities)

6 – parsimony and limits (limits)

7 – cooperative consistency (morality)

THE PROBLEM OF PARSIMONY (PRECISION)

Now lets start with the problem of parsimony and limits: predictive and actionable, and descriptive and non-actionable, are two different criteria.

In physical sciences, we test the determinism (regularity), limits (scope), and parsimony (precision) of a theory, by its predictive(forward) or descriptive(backward) power (external correspondence).

THE PROBLEM OF INFORMATION

Prediction is a test that we have not erred in our description of deterministic systems. And the physical universe is deterministic – because it cannot choose (it cannot predict itself). Moreover, as we scale (the amount of mass we are attempting to develop a theory of), it takes ever greater information (energy) to alter the deterministic course of the universe, even a little bit.

Just as in human beings, as we scale, individuals require ever greater amounts of information to alter their behavior – hence why prices are so important to us, and laws that create regularity (predictability in our risk taking) so important to us. And hence why macroeconomic manipulation using money and therefore prices causes changes in human behavior.

We can predict the orbit of large objects in our solar system. We cannot predict events (information) outside of the deterministic behavior of the objects – we cannot know the unknown externalities.

Every time we cross the galactic plane it seems to invite a great extinction. Can we predict it? Not precisely and therefore not actionably. We can only know that such events tend to happen with regularity. But our precision (parsimony) is very limited.

Can we predict when an asteroid will come free of the belt and tumble toward earth? We can only state that there is no reason one will not. That’s not very helpful. It is however, predictive. It is just not actionable. But the problem is not the theory, it is our lack of information given the externalities.

But humans can think, observe, and change their behavior by means of information, or anticipatory information. Human existence – memory, intuition, thought, and reason – evolved precisely to outwit the deterministic course of regular events and to capture some of the difference for our sustenance.

We can predict that gasses will expand to fill a volume in a vacuum, but not how the individual molecules will be arranged. That would require so much information, that the measurement itself would change the outcome.

So humans – or any sentient creature – can change the universe by his actions a little bit, using only information. (Just as we suspect the subatomic universe transmits information and reacts to equilibrate – somehow. )

We can predict by sympathetic testing (“empathy”), with fair accuracy, how an individual will act when subject to certain incentives, when isolated from many externalities. If we couldn’t then cooperation would be impossible. So by definition human behavior is at least marginally predictable.

But like molecules of gas in a volume in a vacuum, the amount of information necessary to predict the behavior of any molecule is such that measurement sufficient for that determination would affect the outcome. The same applies for humans. Attempts at measurement that the human is aware of change the human’s behavior. So we create institutions that assist us in creating regular behavior: myths, rituals, traditions, norms, rules, laws, governments, and war. Otherwise we ourselves could not predict much outside of our local family.

Humans are relatively predictable at macro-economic levels. Gas is predictable at macro levels. The local physical universe is predictable at macro-levels. But that’s not very parsimonious. It’s not very precise. It’s not actionable.

Predictability in the physical universe is a good test because we cannot empathize with the physical universe, and the standard of predictability is fairly low, and variables can often by isolated from random information.

Humans take very little energy – mere visual information and memory – to change their course. Moving space time using gravity just a little, little bit, takes vast amounts of energy (mass).

Or put another way, it takes great energy (information) to bend space time, and it takes great information to move populations form one behavior to another. Organization is a costly endeavor. And just as the universe will seek to equilibrate the energy transfer (information), so will humans see to seize opportunities generated (information created) by the transition from one state to another.

So predictability is determined by the number and density of variables, and the information necessary to for the object of our consideration to change state.

EXPLANATORY POWER

A theory must provide explanatory power over recorded raw data – utility for the purpose intended. Whether that same theory is actionable or not is a product of the transmission of information within the system, and the energy required to alter its course.

For humans we must record data that captures demonstrated preferences. Money is a good measure of humans because outside of interference by the state, consumption is a demonstrated preference.

A theory must provide explanatory power, and survive criticism. Prediction is just a method of criticism, not a confirmation. Hence falsification is superior to prediction. So first we create a confirmatory test in order to construct an observation. But we then criticize our observation to determine it’s truth or falsehood.

Prediction is low standard of test for the physical universe where lots of information (energy) is required to change state and information (change) is rare.

And prediction is a high standard of test in the human universe where trivial information is necessary to produce a change in state, and information is ever-present.

Prediction in highly deterministic systems is fairly easy and important since the variability is low. Prediction in lightly deterministic systems is not easy nor important for testing since the variability is high.

The question we are always trying to answer is ‘actionability’. Theories must be actionable given the information necessary to maintain or change state, and given the cost of obtaining or imparting that information.

I should probably write something more thorough on this in order to continue to kill off the rothbardian and Misesian pseudoscientific nonsense. Hoppe persists in using this straw man argument to positivism. But it’s a straw man.