A tale of two problem solvers | Average cube shadow area

TL;DR

For vertical, infinitely distant light, the projected area of a tilted square face equals its face area times |cos θ|, where θ is the angle between the face normal and the vertical.

Briefing Cornell Notes

Briefing

The average shadow of a cube—when light comes from directly above and the cube is tossed into every possible orientation—turns out to depend only on the cube’s surface area, not on its detailed geometry. The final result is that the mean shadow area equals one quarter of the cube’s total surface area, giving 1/4 · 6s² = 3s² for a cube of side length s. The striking part isn’t just the number; it’s the route to it, which links a concrete geometry problem to a broad principle about convex solids.

Two contrasting problem-solving styles drive the solution. Bob focuses on computation. He first reduces the shadow of a tilted face to a trigonometric scaling: for a flat face with area s², the projected shadow area equals s² · |cos θ|, where θ is the angle between the face’s outward normal and the vertical direction of the light. Averaging over all orientations then becomes an average of |cos θ| over a uniform distribution of normal directions on a sphere. Carrying out the integral yields a clean constant: the average shadow area of a randomly oriented square face is exactly one half the face’s area.

Alice takes a higher-level route that avoids most of that calculus. She begins with a geometric counting idea: for any fixed orientation of a convex solid, every point in the shadow corresponds to exactly two “face crossings” of a light ray through the solid—entering and exiting. Convexity guarantees this two-to-one relationship (except for edge cases that contribute zero area). From that, she concludes that the shadow area of the whole cube equals one half the sum of the shadow areas of its six faces.

The next step is a clever rearrangement of averaging. Instead of averaging the cube’s shadow over orientations directly, she swaps the order of summation and averaging: the average of the sum of face shadows equals the sum of the average face shadows. Because each face is just a rotated copy, each face has the same universal average projection factor. Once Bob’s constant for a square face is known (1/2 of the face area), Alice’s framework turns it into the cube result: the cube’s average shadow area is (1/2) · 6 · (1/2)s² = 3s², i.e., one quarter of the cube’s surface area.

The argument doesn’t stop at cubes. Alice’s reasoning extends to any convex solid: the average shadow area is proportional to surface area with the same proportionality constant. She pins down that constant using the sphere, whose shadow is always a circle of area πr² while its surface area is 4πr², making the ratio 1/4. For solids with curved surfaces, she justifies the extension by approximating them with polyhedra whose faces hug the sphere more tightly.

Finally, the discussion turns meta: the real unresolved subtlety is what “random orientation” means. Bob effectively assumes a uniform distribution of face normals over the sphere, while Alice’s averaging relies on a specific distribution choice that’s easy to miss—an issue reminiscent of Bertrand’s paradox in probability. The cube-shadow puzzle thus becomes a lesson in both geometry and the hidden assumptions behind “averages.”

Cornell Notes

A cube tossed into all orientations casts shadows whose average area depends only on surface area. For a cube of side s, the mean shadow area is 1/4 of its total surface area: 1/4 · 6s² = 3s². Bob computes the key constant by projecting a tilted square face: a face’s shadow area scales as s²|cos θ|, and averaging |cos θ| over uniformly distributed face normals on a sphere gives an average of (1/2)s². Alice then uses convexity to relate the cube’s shadow to the sum of its face shadows (cube shadow = 1/2 of the total face-shadow sum) and swaps averaging with summation so the same constant applies to every face. The same proportionality principle extends to all convex solids, with the constant fixed using the sphere’s shadow.

Why does the shadow area of a tilted square face scale like |cos θ| times the face area?

Let θ be the angle between the face’s outward normal and the vertical light direction. When θ = 0°, the face is parallel to the ground, so its shadow matches the face area. When θ = 90°, the face is perpendicular to the ground, so the shadow collapses to a line and has area 0. Geometrically, the projection “squishes” the face in one direction by a factor of cos θ (adjacent over hypotenuse in a right-triangle cross-section), while the perpendicular direction is unchanged. That yields projected area = (face area)·cos θ, and the absolute value fixes the sign issue when θ > 90°.

How does convexity produce the relation “cube shadow = one half the sum of face shadows” for a fixed orientation?

For a convex solid, any light ray that hits a point in the shadow enters the solid once and exits once. That means the ray intersects exactly two faces (counting entry and exit), except for rays that pass exactly through edges, which contribute zero area and can be ignored for area calculations. Therefore, if one counts face-shadow coverage by summing the areas of the six face shadows, each point in the cube’s shadow gets counted twice, so the cube’s shadow area equals half that sum.

What does it mean to “swap” averaging and summation in Alice’s method, and why does it help?

Instead of averaging the cube’s shadow area over orientations directly, Alice writes the cube’s shadow area (for each orientation) as a sum over its faces. Averaging over many orientations then turns into an average of that sum. Because the average of a sum equals the sum of the averages (under the same orientation sampling), the cube’s average shadow becomes (1/2) times the sum of the average shadow areas of the individual faces. Since all faces are congruent, each face shares the same universal average projection factor.

How does Bob’s integral lead to the constant that the average shadow area of a square face is half the face area?

Bob treats the face’s normal vector as uniformly distributed over the unit sphere. The shadow area depends only on the angle θ between the normal and the vertical, giving projected area proportional to |cos θ|. Uniformity means the probability of landing in a band of normals between θ and θ + dθ is proportional to the band’s surface area, which scales like sin θ. The average becomes an integral of |cos θ| weighted by sin θ over θ from 0 to π. Evaluating it yields the clean result: average shadow area = (1/2)·(face area).

Why does the proportionality constant for convex solids match the sphere’s ratio 1/4?

Alice’s framework implies that for any convex solid, the average shadow area is proportional to surface area with the same constant. The sphere fixes that constant because its shadow is always a circle: area πr², while its surface area is 4πr², so the ratio is 1/4. For curved surfaces, the argument is made delicate by approximating the sphere with polyhedra whose faces increasingly approximate the sphere; the ratio stays constant at each approximation step, so it persists in the limit.

Review Questions

For a single face, what geometric quantity determines the projected shadow area under vertical, infinitely distant light: the face’s tilt angle, the normal vector, or something else? Explain the role of |cos θ|.
In Alice’s method, where exactly does convexity enter, and what would fail for a non-convex shape like a donut?
What hidden assumption about “random orientation” is needed for Bob’s uniform distribution on the sphere of normals, and why might different choices change the average?

Key Points

1
For vertical, infinitely distant light, the projected area of a tilted square face equals its face area times |cos θ|, where θ is the angle between the face normal and the vertical.
2
Averaging over all orientations can be reduced to averaging |cos θ| over uniformly distributed face normals on the unit sphere.
3
Bob’s calculation gives a universal constant: the average shadow area of a square face is exactly one half the face’s area.
4
Convexity ensures a two-to-one relationship between cube shadow points and face crossings of light rays, implying cube shadow area equals half the sum of face shadow areas for any fixed orientation.
5
Alice’s averaging step swaps “average of a sum” into “sum of averages,” letting the same universal face constant determine the cube’s mean shadow.
6
Combining the face constant with the convexity relation yields the cube’s mean shadow area: 1/4 of total surface area, i.e., 3s² for a cube of side s.
7
The proportionality principle extends from cubes to all convex solids, with the constant fixed by the sphere’s shadow-to-surface-area ratio of 1/4.

Highlights

The cube’s average shadow area is 1/4 of its surface area—so the answer is 3s² for a cube of side s.

Bob’s key constant comes from integrating |cos θ| over a sphere of face normals, producing an unexpectedly simple 1/2 factor for a square face.

Alice’s most powerful move is convexity: every shadow point corresponds to exactly two face crossings, making the cube shadow half the sum of face shadows.

The final generalization is bigger than cubes: every convex solid’s average shadow area scales with surface area using the same constant, pinned down by the sphere.

A subtle probability issue remains: “random orientation” depends on how orientations are distributed, and different choices can change the average. 

Topics

Average Shadow Area
Convexity
Orthogonal Projection
Geometric Probability
Linear Algebra