Isn't every theory or model wrong?

Question

I'm currently in class 12 and I was about different models of atoms in my school chemistry book and there were like 3 or 4 atomic models Rutherford's model, Thompson's model , Bohr's model then finally Quantum Mechanical model.

After reading every model it has limitations and just after some years someone introduced new model patching the previous limitations with giving different theories and equations.

*So technically every model was wrong. I was just wondering that isn't every model we have had so far wrong, and as technology progresses, we would be getting more models which are better than others and will have some limitations, which will later be explained by some other model. So at the end we will no be able to find any correct model of atoms because we are pretty much certain that the present model will be wrong and, after new discovery or technology is discovered, that model will also fail.

What I'm trying to say here is that any no matter how many models we get in future we will not be able to correctly determine what is the structure of atom or how electrons will revolve because every model will have some limitations*

What do you think? Am I right or partially right to see models as wrong?

Answer 1

All models are wrong, and many of them are useful.

Newton's model of gravity is wrong in the close vicinity of a black hole, but perfectly useful to predict the paths of rocket launches from earth.

Newton's laws of motion are wrong when applied to objects going 99% of the speed of light but perfectly useful to predict the behavior of a bowling ball rolling down an alley.

Maxwell's laws of electrodynamics are wrong when applied to objects the size of an atomic nucleus but perfectly useful to predict the behavior of an electric motor in a hybrid car.

As an aside, we know what the structure of atoms are well enough to base the design of atomic bombs, jet airliners, junk food and transistors on them. They are no longer a mystery, in the same way that we no longer refer to cerebrovascular accidents as "strokes of God".

(Note also that Einstein's laws of relativity are wrong when applied to interpersonal relationships but perfectly useful for describing gravitational waves, black holes, stellar creation and evolution, and so on.)

The purpose of developing new models is to take proper account of new information and to understand not just how but also why the universe behaves as it does. This is the job of science and is ongoing in all its fields.

Answer 2

Before we can say that a model is wrong, we must first ask what exactly we would mean by that. In The Relativity of Wrong, Isaac Asimov discusses the degrees of wrongness of various models of the shape of the Earth. He notes that, while it is true that the notions of a flat Earth and a spherical Earth are both wrong, in the usual sense, by simply pointing out just that, you are missing the more important point that the flat Earth is more wrong than the spherical Earth. From this, we can see that it is not possible to cleanly divide models into categories of right and wrong. Moreover, the process seems a bit redundant if we are ultimately going to categorise them all as wrong, anyway. Instead, I suggest that we ask not if a model is wrong, but rather under what circumstances (if any) it is right. With this view, we can see that the flat Earth and the spherical are both right (in the sense of making accurate predictions) under certain sets of circumstances, with the former being a proper subset of the latter. For another example, I would recommend reading this paper by Carlo Rovelli on the correctness of Aristotelian physics. To answer your question more explicitly, you could be considered right to label all models as wrong, but you'd be missing the point of modelling.

Answer 3

A model is never "wrong", because a model is an abstraction of the territory, not the territory itself. By expecting a "right" model, you expect an absolutely identical territory, which is impossible.

A "right" model just represents an abstraction of the object. A plastic sphere representing the earth could be "right" for some and "wrong" for others, depending of the goal of modeling.

Consider this two facts:

1. A model is an abstraction, which implies simplification. Architectural home planes are just lines, but a real house is a not made by lines. This aspect is particularly complex when trying to model an atom: we need to model it in a state, with defined properties, with a simple shape, having a size, etc., which are facts that are not proper to the atomic nature. Yes, the model is not the atom.

2. Models follow a goal. That's why there are multiple types of maps. A political map does not represent soil types, because that's not the goal of a political map. In the same way, an atom can be modeled in multiple forms: figures, equations, theories, etc. But none of them is the atom itself.

Take knowledge, for example.

Knowledge of the world is a model of the world in your head. But not one atom of such world is properly represented in your head. However, you can't just think that knowledge is wrong. Knowledge of the world is not identical to the world itself, and since we are still alive, it can be stated that the model, serving to the goal of survival, is at least useful. Never "wrong".

Answer 4

In the philosophy of science, theories and models are recognized as being useful but limited. One famous philosophical slogan to epitomize that is "A map is not the territory.":

The map–territory relation is the relationship between an object and a representation of that object, as in the relation between a geographical territory and a map of it. Polish-American scientist and philosopher Alfred Korzybski remarked that "the map is not the territory" and that "the word is not the thing", encapsulating his view that an abstraction derived from something, or a reaction to it, is not the thing itself. Korzybski held that many people do confuse maps with territories, that is, confuse conceptual models of reality with reality itself. These ideas are crucial to general semantics, a system Korzybski originated.

So, it might be more helpful to consider the question isn't one of right-or-wrong, but instead is to what degree is it useful. Scientific theories are models that grow more useful and more reliable as they become more sophisticated through testing and debate. And some philosophical thinkers think the same things about worldviews. I'd recommend you read the SEP's article Models in Science for more information.

Answer 5

Physics attempts to explain how the real world works using models.

These don’t attempt or claim to show the actual underlying mechanism of how things work, but they provide something that’s (a) reasonably accurate within certain limits, and (b) easy to understand. For example, if you’re building a house then Newtonian physics is perfectly good (and you can assume that the Earth is flat).

So the idea that all models are wrong has some validity; if a model isn’t absolutely precise under all conditions or more strictly if it doesn’t explain the underlying mechanism (which is arguably impossible) then it’s imperfect and if you’re expecting otherwise then it’s wrong.

Answer 6

None of the physics models are wrong. You simply shouldn't take a model as saying "Reality is this". Statements like this are completely hopeless, because it is impossible to define reality.

Even before we had discovered quantum mechanics, it was reasonable to say that Newtonian mechanics didn't capture reality. What a physics model does is to map certain fuzzy aspects of reality into concrete objects of mathematics. All physics models are correct in the sense that they never claim to capture reality, but only claim to capture the approximate behavior of certain aspects of reality, by mapping them to concrete mathematical objects.

There is no denying that the notion of a point particle under the influence of a force truly captures the behavior of the fuzzy notion of position of objects. It is a verifiable truth.

Answer 7

Capturing the essential
To add to the other answers and comments: the claim that All models are wrong but some are useful is actually an understatement. The value of any model is precisely in capturing the essential rather than dutifully representing the whole. This point is beautifully represented in Borges' famous On exactitude in science (see the text here):

...In that Empire, the Art of Cartography attained such Perfection that the map of a single Province occupied the entirety of a City, and the map of the Empire, the entirety of a Province. In time, those Unconscionable Maps no longer satisfied, and the Cartographers Guilds struck a Map of the Empire whose size was that of the Empire, and which coincided point for point with it. The following Generations, who were not so fond of the Study of Cartography as their Forebears had been, saw that that vast Map was Useless, and not without some Pitilessness was it, that they delivered it up to the Inclemencies of Sun and Winters. In the Deserts of the West, still today, there are Tattered Ruins of that Map, inhabited by Animals and Beggars; in all the Land there is no other Relic of the Disciplines of Geography.

—Suarez Miranda,Viajes de varones prudentes, Libro IV,Cap. XLV, Lerida, 1658

Generalization
Another aspect of models, particularly important in the modern context of machine learning, is their ability to generalize a model built on a set of examples to other examples. Let us take, e.g., the algorithms that automatically tag faces in photons on social nets, read handwritten text, or recognize our voices in automatic phone calls - the models behind this algorithms have been built (trained in ML language) on thousands of examples of photos, handwriting samples, voices. If they were created to assure the exactitude of representing those training data, they would be of no use to us (unless our photos/handwriting/voice are part of the training data) - in this case in ML one would talk about overfitting data. It is because these models captured only the features that are general to the training data and the rest of the mankind, that they are capable of tagging/recognizing photons, handwriting and voices of millions of people, while being based on a small subset of the users. (One also speaks of underfitting when the model captures too few features to be useful.)

Answer 8

I am confused by you putting theories and models on the same footing. They are different, and also I don't know what exactly you mean by a theory/model being wrong. If Bohr's model is wrong, then it is not clear why it was taken so seriously for decades, and is still taught in grade 8 science class.

However, theories/models do replace one another, and based on my understanding you want to know why. In philosophy of science I think the best answer to this question was given by Thomas Kuhn (interesting fact that he interviewed Niels Bohr one day before the great genius of physics died). According to Kuhn, theories get bundled into a systematic worldview called a paradigm. Paradigms are totalitarian and all-embracing. There are always some inaccuracies and discrepancies in any given paradigm, but those challenges are usually just swept under the rug. However, as they keep piling up, and eventually a lonely genius comes out of nowhere and gives us a new paradigm, that explains all those discrepancies that the old one could not. This revolution is called a paradigm shift.

Classical physics worked just fine for centuries. In the late 19th century things like the black body radiation anomaly and some others lead to Max Planck producing a paradigm shift that was later called the quantum revolution.

There are also interesting sociological things relating to paradigm shifts that I don't want to get into.

It's also interesting that Kuhn's account is in close agreement with Heidegger's regarding science.

Answer 9

Saying that a model is wrong could be counterfactual, because the whole idea behind modelling is to create something that explains or give us a tool to understand and predict a phenomenon. And by phenomenon I mean one that is not fully understood (yet), or one where we don't fully know the impact of a group of variables that affect it. This could apply to meteorological predictions or biological systems.

Also models are used, in most cases, to simplify stuff that is known to be complex. In this case, the model itself is known to be a "lighter" version of the phenomenon. For example: treating the Earth as a sphere, or other abstractions, such as saying a system is linear or second order when we know it's not. The good part is that this allows us to "look into" the phenomenon with tools we can use, instead of sitting and saying we can't study something because it's out of our scope.

So, with these ideas we can say that a model is never wrong because we never should assume that it is 100% right.

However, there are other ideas we can consider. A model is always done with a purpose. If the model works for that purpose, then the model works, even if it has its limitations (and we know them). While we don't expect a model to be 100% right, we do expect it to be as right as possible, or at least try to be as close to 100% right as possible.

So if I create a model of something, I can do it in different ways, and some of them may be better than others. Treating the Earth as a sphere or as a 2D map is something we are used to do because it serves its purposes. But what if I say: let's treat the Earth as a pyramid?

Common sense would tell us that a pyramid approach would always be "more wrong" than a spherical one, and therefore, such a model would be inferior to it. But what if we find a specific application where the pyramid works better and serves a purpose? Should be weird, yes, but not entirely impossible.

However there may be models that are useless and have no bases. Such as saying that the ambient temperature increases in a linear way since 00:00 until 23:59. That is just wrong, although a deffender of such a model could say that it works right within the 7 AM - 9 AM time interval.

Answer 10

The real argument here is epistemology vs ontology. What do we know through our flawed senses vs what is actually real.

People have been arguing this question for thousands of years. Plato with his cave and his clouds was dealing with this exact issue. All we see are shadows, but the real thing is often elusive.

It's not an easy question to solve, as you might expect. There is always going to be a disconnect between perception and whatever the inaccessible-to-us objective reality is. This has nothing to do with reality, and only our imperfect perception.

So, sure, you're going to get a lot of different models. And they're not all going to be complete, unless we fully understand something, which is uncommon. But generally they're all useful, they all come at the essential, ontological truth from a different angle and expose something valuable, some subtle distinction that makes the thing that is so far eluding us, a little less mysterious.

Answer 11

In your question you are focused on different models of atoms but models are used in all sorts of analysis of reality, not only physics. I don't know if this was an accidental omission and you are asking about physical models specifically or if your question is about all models in general.

If the question is about all models, I can give you an example of a model that is 100% accurate from my discipline - computer science. We use models to describe behavior of different elements of a software systems interacting with each other. The software is often written alongside the model or the model comes first. Assuming the software has no bugs, which is at least theoretically possible, if you use the model to predict behavior of software, you will be right 100% of the times.

One counterargument to my answer that can be made is that the software model doesn't take into account things like electrical interference, bad configuration, hardware limitations or someone tripping over the power cable and disconnecting the machine. To that critique I'll say that expected conditions are parts of the model. The model explains the behavior of software under specified conditions and says nothing about, for example what happens when a solar flare hits the server.