An Ode to the Spherical Cow
How Imperfect Models Drive Scientific Discovery
Stop me if you have heard this one before. A farmer asks a theoretical physicist for advice on improving milk production. The physicist examines the situation and announces that he has a solution. But it only works, he says, for spherical cows in vacuum.
It’s a joke on how physicists’ reliance on wildly improbable idealizations. From zero friction surfaces to gas molecules that pass through each other, the best known physics models resemble the real world as much as a 5 year old’s crayon painting resembles the Mona Lisa. But wacky though they may sound, idealized models are the lifeblood of physics.
The Model Zoo
Reality is messy. Cows are lumpy, the air is turbulent and quantum fields buzz with incredible complexity. A model is not meant to be a perfect picture of reality—it’s a way to strip the problem down to something tractable.
In this post, we take a walk through some of physicist’s most beloved models and see what makes them so powerful.
The point particle
The first physics model students encounter takes ‘stripping the problem down’ to its limit, modelling objects as a point of the same weight. Taken at face value, the point particle is an absurdity—such an entity would be infinitely dense.
But when it comes to situations when the dimensions of the object are irrelevant, the point particle provides a powerful approximation. Consider, for example, the problem: how long does it take for the earth to complete one turn around the Sun? The Earth, a giant lumpy ball hurtling through space, is pretty far from a point. But approximating it as one returns the time period of 365.26 days. Boom! Ignoring mountains, oceans, and cows (spherical or not) gets us within 0.001% of the right answer. Mind slightly blown?
Of course, one has to know the limits of idealizations. The point particle becomes less and less effective the more and more the shape and size of the object begins to matter. For example, when an object is falling through the air, the air drag depends on its surface area, shape and flow conditions. A point particle would be a terrible approximation for someone falling with a parachute.
The ideal gas
Molecules imagined as non-interacting point particles. Computationally, it’s baby's first gas model. But the payoff is massive: many gases under ordinary pressures and temperatures follow “the ideal gas law” (PV=nRT). From speed of sound to hot air balloons, the ideal gas model explains all.
It begins to fray in situations where interactions and size of molecules becomes important, such as low temperatures and high pressures, or during phase transitions. Even in the places where it fails, ideal gas serves as a reference point: deviations highlight the importance of intermolecular forces. This shows another function of models: to serve as benchmarks.
The Black Body
Every object absorbs some of the light and heat that it receives and radiates back some of it. The black body is an impossible perfect absorber and emitter of heat—it absorbs and emits back all radiation that comes its way. Think of it as a vacuum cleaner for radiation.
The black body accurately describes thermal radiation across a wide range of systems — from incandescent filaments to the cosmic microwave background. Assuming the Sun to be a black body gets the temperature of the Sun to a 0.4% error. That’s the power of a good lie.
Oppenheimer-Snyder model for black hole formation
Models are not just good as first approximations, they also illuminate the essence of processes by cropping away inessential details.
A great example is Oppenheimer-Snyder’s model for black hole formation. They considered the collapse of a perfectly spherical star with no internal forces or pressure. This reduced a complex problem to one that could be solved exactly. In the solution, you could see the story of the star collapsing into a black hole unfold.
The alternate, adding realistic details, would have resulted in complicated equations that could only be worked out numerically. If one did not know what one was looking for, the big picture would be lost in the haze of numerical simulations. But once we know the core narrative, it is much easier to interpret the results.
The Ising model
The GOAT of models. A minimalist grid of spins that can only point up or down, interacting solely with their nearest neighbours.
Born to explain ferromagnets, it ditches realism—no quantum weirdness, no long-distance interactions, no disorder. Yet this minimalist masterpiece nails phase transitions in magnets—even the exact numbers!1 But the Ising model has transcended its origin. Today, this grid of up/down arrows explains everything from stock market crashes to social media trends. GOAT for a reason.
The subtle art of modelling
Coming up with a good model then is as much art as it is science. Pick the wrong simplification? You just erased a hurricane. Modelling the atmosphere as an ideal gas works for many purposes—but ignore moisture content and you miss the latent heat driving weather systems. Treating electrons in a metal as a free gas gives you good conductivity predictions—but without considering electron–phonon interactions, you can’t explain superconductivity.
The tradeoff between realism and insight is a tightrope walk. Too much reality? You are lost in the weeds. Too little? Your model is useless. A good model builder needs to have a keen instinct for which aspects are essential for a specific problem and which can be thrown away. The above models are essential to physics because they tread the path perfectly.
As George Box famously said: all models are wrong, but some are useful. The useful ones are worth their weight in gold. So the next time you hear about a physicist's spherical cow, tip your hat a little. She is not just a joke at physicist’s expense, she is our mascot!
This incredible feature is a result of universality, a topic that deserves its own post.
Psst. You can support my work by buying me a coffee