物理 · Physics

Motion, Force & Determinism

Day 2 · 2026 · Phase A — the Skeleton of Classical Mechanics
A force is not what keeps a thing moving — it is what changes the motion. Getting that backwards cost humanity two thousand years.
Aristotle said: to keep a thing moving, keep pushing it. It matches daily life — and it was wrong for two thousand years. Galileo and Newton's fix changed one word: a force changes not motion but the change of motion. That one edit stitched "the heavens" and "the earth" into a single set of laws — a falling apple and an orbiting Moon become the same thing. Pushed to its limit, it conjures a ghost: if everything is fixed by forces and initial states, is the future already written? Four cards: inertia, the real meaning of F=ma, universal gravitation, and Laplace's demon.

Inertia: Uniform Motion Needs No Force The First Law

Classical mechanics · 1638 / 1687
Intuition Push a box, let go, it stops — so "motion needs a force to sustain it," right? Wrong. It stops not because "motion wants to stop" but because friction quietly drags it. Remove the friction: a curling stone glides far across ice; a spacecraft with its engine off keeps flying for decades. With no force, motion doesn't stop — it just keeps doing what it was doing.
Mechanism This is Newton's first law (the law of inertia): with no net external force, a body stays at rest or in uniform straight-line motion. The key word is "net" — many forces may act at once; as long as they cancel (net force zero), the state of motion does not change. So "at rest" and "moving uniformly" are physically the same state — they differ only by which frame you stand in.
No friction net force zero → uniform (equal spacing) speed unchanged Friction friction backward → slowing (spacing shrinks) friction
Same launch: with no friction each step covers the same distance (uniform); with friction each step shortens until it stops — stopping is the work of a force, not motion "wanting" to stop.
The counterintuitive point Daily experience fooled us for two thousand years: because Earth is full of friction and air drag, we have never seen force-free motion, so we mistook "stops when you don't push" for the nature of motion. Newton's genius was subtraction — imagine friction gone entirely, and what remains is the pure law. Physical truth often hides in the "ideal case," not on the messy surface of the everyday.
Cross-domain · Cognition / AI This is a textbook naive-physics illusion: the human brain ships with a "to move, keep exerting" prior that contradicts the real law. It's a warning for AI too — intuitions trained on mountains of everyday data can systematically diverge from the underlying rule; surface correlation (let go and it stops) is not mechanism (friction stops it). To see the mechanism, both physics and models must learn to strip the noise and subtract.
In one line: motion doesn't need a force — a change in motion does.
Think: if "at rest" and "uniform motion" are physically the same, does "absolute rest" even mean anything?
No. No experiment can distinguish “absolute rest” from “uniform motion” (Galilean relativity), so “absolute rest” has no physical content — only rest relative to some frame. Einstein pushed this intuition to its limit, making it the starting point of special relativity.

The Meaning of F=ma Newton's Second Law

Second law · 1687
Intuition If a force changes the "change of motion," how exactly? The answer is startlingly plain: the bigger the force, the faster it changes; the heavier the thing, the harder to change. Push a shopping cart and a truck with the same force — the cart shoots off, the truck barely stirs. The whole difference is mass. Mass is a body's stubbornness against being changed.
Mechanism Newton's second law nails this down as an equation:
F = ma
F is the net force (the combined effect of all forces), m is mass (the measure of inertia — larger means harder to push), and a is acceleration — note, acceleration, not velocity. Acceleration means "how fast velocity changes": hitting the gas, braking, turning are all acceleration (the size or direction of velocity is changing). Rearranged, a = F/m: same force, double the mass, half the acceleration.
When the net force is zero, a = 0 (the number 0, zero), velocity holds — which is exactly the first law again, revealing it as a special case of the second.
m force F large a (light → easy) 2m same force F small a (heavy → hard)
The same force F: double the mass, half the acceleration. Mass is the measure of "resistance to being changed."
The counterintuitive point Many memorize F=ma as "force = mass × acceleration" and miss what it actually says: force is proportional to acceleration, and velocity does not enter the equation. A car cruising at 100 km/h feels a net force of zero — even though it's fast. Speed is irrelevant; whether velocity is changing is what involves force. F=ma is half definition (it defines how we measure force and mass) and half empirical claim (the real world happens to obey the proportion) — only together do they make a law.
Cross-domain · Control / Optimization "Output proportional to what you apply, dragged by an inertia factor" is a wildly general template. In control theory, a system's response is set by its mass/inertia, which engineers exploit to design brakes and stabilizers. In machine learning, gradient descent updates parameters ≈ learning rate × gradient, and momentum methods borrow physical inertia outright — letting updates "carry momentum" through noise and shallow pits. Inertia is both the stubbornness of physics and the smoother of optimization.
In one line: force governs acceleration, not velocity; mass is the measure of "hard to change."
Think: in an elevator you feel heavier the instant it starts, but nothing while it rises steadily — why? Which phase has acceleration?
Only the instant of starting has acceleration (your speed is growing from zero). There the floor must push you a bit harder than your weight to accelerate you along — so you feel “heavier.” During constant-speed rise, acceleration and net force are zero, indistinguishable from standing still. What you feel tracks acceleration, not speed.

Universal Gravitation One Law for Heaven and Earth

Newton · 1687
Intuition An apple falls, yet the Moon does not — so the ancients naturally figured "the heavens" and "the earth" were two worlds with separate rules. Newton asked a crazy question: what if the Moon is falling too? It plunges toward Earth continually, but its huge sideways velocity makes it "miss" Earth every time — so it falls around forever, and that is an orbit. To circle is to keep falling and keep missing.
Mechanism Newton claimed: any two masses attract each other, with a force of magnitude
F = G m1 m2r2
m1, m2 are the two masses, r is the distance between them, and G is the gravitational constant (a tiny fixed number, read "big G"). The r2 in the denominator means: double the distance and gravity drops to one quarter — the inverse-square law. The same equation gives both the apple's fall and the Moon's orbit.
Earth Moon sideways velocity v gravity F no gravity → straight line
The Moon's sideways velocity would fly it off in a straight line; gravity keeps bending it back — the two in balance are the orbit. Orbit = perpetual falling + perpetual missing.
Why it matters This was physics's first great unification: "motion in the heavens" and "falling on the earth" merged into one law. Its power isn't just explanation — it's prediction: astronomers later used gravity-computed discrepancies to "calculate" the planet Neptune at the predicted spot — a pen finding a planet before a telescope did. Unify → predict → verify became the engine of all physics thereafter — electromagnetism, the electroweak, even the reunderstanding of gravity itself all replay it.
Cross-domain · Scientific method / AI "One clean rule explaining a whole swath of seemingly unrelated phenomena" is the power of unification, and the gold standard of a good theory: the wider the coverage and the fewer the assumptions, the likelier you've touched a real mechanism. It's isomorphic to machine learning's aim — a model that generalizes across many tasks beats patching each task separately; generalization is to a model what unification is to physics. Yet be honest: Newtonian gravity is superbly accurate, but cracks under very strong gravity and is ultimately replaced by the picture of curved spacetime (the "gravity = curved spacetime" issue works this out). Even a triumphant unification may be only the approximation of a deeper one.
In one line: an orbit isn't "not falling" — it's "always falling, always missing."
Think: astronauts float "weightless" on the space station — is that because there's no gravity up there? (Hint: they're falling with the station.)
No. Gravity there is still ~90% of its ground value. They “float” because they and the station are in free fall together (always falling, always missing), with no squeeze between them. Weightlessness = free fall, not the absence of gravity — the everyday face of the equivalence principle.

Laplace's Demon Determinism

Determinism · 1814
Intuition Newton's laws have a startling property: know each particle's position and velocity right now, plus the rules of force, and you can compute the next instant, and the next… all the way down. In 1814 Laplace took this to the limit, imagining an all-knowing intellect (later dubbed Laplace's demon): if it knew the state of every atom in the universe right now, then the entire past and the whole future would be as plain to it as the present.
Mechanism This is determinism. Its mathematical root: Newton's laws are a set of differential equations (rules for how a state changes over time), and given the initial conditions (this moment's positions and velocities), the equations' solution is uniquely fixed — the future has no second option. The world is like a wound-up clock: initial state + laws = one predetermined trajectory.
The counterintuitive point (an honest distinction) "Determined" is not "predictable" — the point most easily mystified, so keep it straight:
  • Chaos: many systems (weather, the three-body problem) are extremely sensitive to initial conditions — a hair's difference at the start, wildly different endings. The equations are still deterministic, but you can never measure "infinitely precise" initial values, so it is fixed in principle, uncomputable in practice (the "chaos & nonlinearity" issue is devoted to this).
  • Quantum mechanics: deeper down, the microscopic laws are probabilistic — the same initial conditions give only probabilities of various outcomes, not a single future. The demon's classical dream is shaken at the root (the quantum issues handle this head-on).
So don't take determinism as science's endorsement of fatalism: even if the world is deterministic, it is neither predictable nor does it dissolve questions like "why did I choose thus."
Cross-domain · Computation / AI Laplace's demon is essentially a perfect simulator: feed in the initial state, run the laws, output the future. That's the very program of numerical simulation (weather forecasting, molecular dynamics, cosmology) — except a real "demon" hits two walls: initial values can't be measured exactly (chaos) and compute is finite (there is no computer larger than the universe to simulate the universe). It also mirrors AI's limits: how accurately a model can extrapolate is locked by data precision and the system's degree of chaos — deterministic isn't computable, and computable isn't predictable.
In one line: fixed ≠ predictable; Newton's clock is precise, yet chaos and the quantum still block it.
Think: if simulating the universe needs a computer bigger than the universe, does "predictable in principle" still mean anything in practice?
Very limited meaning. If “prediction” needs a computer bigger than the universe, “predictable in principle” is empty talk, indistinguishable in practice from unpredictable. It warns us to separate two things: determinism is an ontological claim (is the world fixed?), predictability is an epistemic/compute question (can we actually compute it?).

Going Deeper

Aren't the first and second laws the same statement twice? Since F=ma gives uniform motion at F=0, why list the first law separately?
Mathematically the first law is indeed the special case of the second at F=0 — but it is not logically redundant. Its real job is to assert the existence of inertial frames: only in the class of frames where a force-free body moves uniformly does F=ma hold. In an accelerating or rotating frame (say a sharply braking bus) you'll "see" things lurch forward for no reason, as if forced — but that's the frame deceiving you. So the first law isn't filler; it sets the stage for the second: fix "where you're watching from" first, and only then does the second law mean anything.
Inertial mass (how hard to push) and gravitational mass (how much gravity you feel) — why are they the same m? Might they just coincide numerically?
A question deep enough to stump Newton himself. The m in F=ma measures "stubbornness against change" (inertial mass); the m in the gravity law measures "how strongly you feel gravity" (gravitational mass) — conceptually unrelated, yet experimentally exactly equal (verified to extreme precision). Galileo's "heavy and light land together" is a direct consequence of this equivalence. Newton used it as a coincidence; Einstein elevated the "coincidence" into the equivalence principle — gravity and acceleration are indistinguishable — and rebuilt the whole theory of gravity from it (the "gravity = curved spacetime" issue starts right here). An overlooked coincidence is often the crack where the next revolution begins.
If an orbit is "always falling," why don't planets actually spiral into the Sun, instead staying in orbit for billions of years?
Because "falling in" and "rushing sideways" are in dynamic balance: gravity pulls the planet inward, the sideways velocity flings it outward, and the two trade off along the ellipse while their totals stay conserved (both energy and angular momentum are conserved — the "energy & momentum" issue covers why that conservation runs so deep). As long as nothing steadily steals energy, this balance holds on astronomical timescales. To make it fall in, you need dissipation — friction with dense gas, or gravitational waves bleeding energy away ever so slowly. So a stable orbit isn't "not falling"; it's the tendency to fall and the tendency to escape locked into a loop by conservation laws.
Since Newtonian gravity was superseded by relativity, is it still "right"? Are we learning an outdated error?
Newtonian gravity wasn't "overthrown" — it was located. It's an excellent approximation of a deeper theory across the vast regime of "weak gravity, low speed," with errors small enough to launch probes and predict tides with room to spare (exactly the point of "effective theory": each scale has its good-enough law). Relativity shows a difference only in extremes (Mercury's tiny perihelion precession, GPS timing, near a black hole). So learning Newton is not learning an error — it's learning a truth with clearly drawn boundaries. Knowing where a law holds and where it fails matters more than the law itself.

Further Reading