To what extent do you agree with the claim “all models

are wrong, but some are useful” (attributed to George

A model is a simplified representation of reality that can be used to make predictions.

George Box stated, “All models are wrong, but some are useful.” So, how can we

know if a model is wrong or useful? Or how wrong does it need to be to no longer be

useful?

Truth and wrongness are difficult concepts to grasp because there is no absolute

truth or wrongness when discussing knowledge. The writer and scientist Isaac

Asimov identified different degrees of wrongness in his book Relativity of

Wrongness. He argues that more accurate answers exist without reaching the

absolute truth, but how can we evaluate how right, or in this case, how wrong a

model is?

In mathematics, the truest concepts can be believed to be true due to their validity,

but as we will see later, these ideas can be questioned. Validity is the logic behind a

statement. In the natural sciences, which delves into the real physical world, we can

not be completely certain that a claim or a model is true due to the complexity of

reality, so we focus on how verisimilar it is; how close to reality it is.

The historian, philosopher, and author Yuval Noah Harari claims that “Scientists

generally agree that no theory is 100 per cent correct. Thus, the real test of

knowledge is not truth, but utility. Science gives us power. The more useful that
power, the better the science.”. Utility is how accurate the model is in predicting an

outcome. Since every model is a simplification of reality, we can not expect it to be

undoubtedly true. However, the usefulness determines if a model is less wrong than

others. Therefore, in this essay, I will explore different perspectives varying in

wrongness and usefulness and evaluate them to see how much they agree or

disagree with the original statement, to then determine how much I agree with the

conclusion.

In mathematics, the introduction to irrational numbers always comes from pi. The

most important characteristic of irrational numbers is the infinite number of decimals,

pi is no exception. The never-ending list of decimals has wondered mathematicians,

and some try to find the maximum number of decimals, through complex algorithms.

Pi can be considered a model since it can predict outcomes. The uses of pi are not

limited to the circumference of circles, but every aspect of reality that involves

circles, spheres, or curvatures will include this number, including the complexity of

the cosmos and DNA. Dr. Dorina Mitrea said, “As soon as you start working on

problems, immediately you have to resort to Pi, that’s why it will continue to be

relevant. You can’t do anything without this special number.” She emphasises the

major usefulness of pi in mathematics. Since pi is infinite, it is impossible for us to

learn all the digits. Therefore, making our knowledge of pi wrong, or not completely

accurate. However, is it necessary to discover all the decimals of pi? The most used

number for pi, 3.14, is not as accurate as maybe 30 decimal places or 30 million, but

it is still the number almost everyone uses. So, our knowledge of pi is restricted,

hence wrong, but it is not essential to be true, since some decimals are useful

enough to describe reality.

work on calculus, but his work on motion, gravity, and forces was the most

significant. In modern society, they are used in our everyday lives because they are

useful enough to predict the motion of objects. Despite this, Newtonian mechanics is

not completely useful, since it does not work accurately when speeds reach light

speed. Therefore, Einstein thought of the theory of relativity. Newtonian mechanics

described the motion of objects at slow speeds, where the effects of relativity are

minimal. Whereas, relativistic mechanics analyses the behaviour of objects near the

speed of light. Our understanding of space, time, and motion was impacted by this

new knowledge. This new knowledge could override Newtonian mechanics, so why

did it not? As I said, Newtonian mechanics are used in our everyday life to predict

the movement of objects at our speed, slow speed. This makes it much more useful

than relativistic mechanics since we can give it so much more uses. It is wrong, and

the better option is still Einstein's theory of relativity, but its utility keeps it relevant. It

is so useful that it took us to the moon. Furthermore, the laws of motion are being

taught at schools, which suggests that they are useful enough to serve as a basis to

then learn about relativity. Newtonian mechanics show how usefulness, in this case,

daily use, overthrows the concept of wrongness.

In mathematics, the truest statements are axioms. They serve as the basis of

mathematics due to their self-evident truth. However, they are nothing more than

unprovable assumptions, and assumptions can affect the predictability of a model

since it does not recreate reality. Since axioms are assumptions, when we use

several to prove a theory, it is more likely that they might come into conflict.
Mathematics has two main branches, pure maths and applied maths. Since pure

maths goes into abstract concepts, it is more difficult to prove models wrong since

the hypothetical concepts are based on idealised axioms that can be considered

true. Take as an example the axioms of plane geometry, more specifically the fifth

Euclidean postulate that states, “If two straight lines in a plane are met by another

line, and if the sum of the internal angles on one side is less than two right angles,

then the straight lines will meet if extended sufficiently on the side on which the sum

of the angles is less than two right angles.”, referring to parallel lines. This claim can

be used to predict the behaviour of different lines, therefore it acts as a model. Since

the axiom assumes a perfect plane and perfect parallel lines, it will only be true in an

idealised context. Therefore, is it truly useful? Mathematics as an area of knowledge

investigates relationships and patterns, therefore this axiom, even though it is not

completely true, can still be used in many aspects thus being identified as useful.

Brikena Djepaxhija, a Norwegian mathematician, said, “Making an assumption

means proposing that a statement is temporally true as a productive basis for

subsequent activities”. Axioms are unprovable assumptions that are presumed to be

true due to their validity; the major impact it has on subsequent theorems and other

concepts makes them the most useful knowledge in mathematics.

From the second century AD until 1,500 years later, the Ptolemaic model was the

model that described our solar system. Astronomers and mathematicians believed

this model to be true due to the useful predictions it gave for that time. Other models,

like the Copernican heliocentric model, were rejected because they did not fit the

observational evidence given for the Ptolemaic model, even though, it was

completely wrong. This illustrates how, in the natural sciences, the utility of the
evidence a model is capable of producing, determines the decision of a model to be

considered true enough to be used. When a new model produces more useful

predictions, it replaces the previous model in use. In that time, the evidence provided

for the establishment of the Ptolemaic model was based on eye observations,

because of the lack of technology to expand on those observations. 1,400 years

later, Galileo Galilei invented the telescope, which was able to observe space with

more precision than the naked eye. Despite the initial scepticism, the heliocentric

model that Copernicus, Galileo and other scientists supported became the new

model to describe reality, due to the better predictions it gave, in this case, due to the

technology advancements. The Ptolemaic model shows how a model can provide

useful predictions even though it is completely wrong due to the lack of

verisimilitude. However, this may raise the question, of how wrong a model must be,

to be considered not useful. Is it useful if it is completely incorrect? Arguably, the

heliocentric model proposed by Copernicus is also wrong, because the orbits of the

planets are not perfectly circular, however, it is still less wrong than the previous

model, which makes it more useful.

In conclusion, I agree with the claim that “all models are wrong, but some are useful”,

however, the claim can be expanded to “All models are inherently imperfect

representations of reality, however, a model that lacks practical applicability is, in

essence, even more flawed.”. Pi focused on how our knowledge is restricted, but if

our knowledge is sufficient to describe reality, then truthfulness is not needed.

Newtonian mechanics showed how the simplicity of a model, even if not true, can be

more useful in everyday life than more sophisticated and more realistic models.

Axioms and axiomatic systems highlighted the necessity of a truth, even if assumed
or idealised, for the rest of an AOK to be useful. The Ptolemaic model proposed that

the model with more practical applicability is perceived as more true. Both

mathematics and natural sciences require models to simulate reality, yet, truth in

mathematics is based on validity and truth in the natural sciences is based on

verisimilitude. Models will be used forever because it is physically impossible to

recreate reality with perfection, so we should always strive for maximum practical

applicability for us as a society.