You're staring at a graph. A straight line climbs steadily upward. A parabola curves, steepens, flattens, then steepens again. Both represent change — but they don't change the same way.
That's the whole point of rates of change. And if you've ever sat through a math class wondering why we care about the difference between a line's slope and a curve's "average rate of change over an interval," you're not alone. Most textbooks make this feel abstract. It's not.
Let's talk about what's actually going on.
What Is Rate of Change, Really?
Strip away the notation. Rate of change answers one question: how much does the output shift when the input moves?
That's it. Also, you put in x, you get out y. Change x by some amount. How much does y respond?
In a linear function, the answer is boringly consistent. Here's the thing — predictable. If you walk at 3 miles per hour, every hour adds 3 miles. Every step in x produces the same step in y. Day to day, it doesn't care what time it is. Now, the rate of change is 3. Rise over run. That's the slope. Constant. You learned it as m in y = mx + b. It doesn't care how far you've already gone Most people skip this — try not to..
Not the most exciting part, but easily the most useful.
Quadratic functions? Different story.
The Curve Changes Its Mind
A quadratic — f(x) = ax² + bx + c — doesn't have a single rate of change. It has infinitely many. At every point on that parabola, the steepness is different. The rate of change itself changes.
This is where students start to glaze over. On the flip side, at the moment it leaves your hand, it's moving fast — high rate of change. Its height is quadratic in time. That's why throw a ball upward. Then it accelerates downward. But think about it physically. Zero. In practice, for an instant, it's not rising or falling. At the peak? The rate of change goes negative and keeps growing in magnitude That alone is useful..
The function is smooth. Plus, the rate of change is not constant. Worth adding: it's linear, actually — the derivative of a quadratic is linear. But we're getting ahead of ourselves.
Why This Distinction Actually Matters
Here's what most introductions skip: the difference between constant and variable rates of change isn't academic. It changes how you model the world.
Linear models assume the future looks like the past scaled up. On top of that, quadratic models assume acceleration — or deceleration. That's a fundamentally different assumption about reality.
When Linear Thinking Fails
Say a company's revenue grows $100K per year. Practically speaking, linear model: in 10 years, add $1M. Simple That's the part that actually makes a difference..
But what if growth accelerates? Year 1: +$100K. Practically speaking, that's not linear. Each year, the increase gets bigger. Also, year 2: +$120K. Consider this: that's quadratic-ish (actually exponential, but quadratics approximate acceleration over short windows). Year 3: +$144K. If you project linearly, you'll underestimate badly The details matter here..
Conversely: diminishing returns. Next 10 boost it less. Consider this: the rate of change decreases. A factory adds workers. First 10 boost output a lot. Linear projection overestimates.
The shape of the rate of change is the shape of your assumption about how the world works.
How to Calculate Rate of Change (Both Kinds)
Linear: It's Just Slope
You know this. But let's be precise.
Given f(x) = mx + b, the rate of change between any two points x₁ and x₂ is:
(f(x₂) - f(x₁)) / (x₂ - x₁) = m
Always m. Pick x₁ = 2, x₂ = 5. Pick x₁ = -100, x₂ = 100. Same answer. The secant line is the function. There's no "average" versus "instantaneous" distinction — they're identical.
Quadratic: Average Rate of Change Over an Interval
Now it gets interesting. For f(x) = ax² + bx + c, the rate of change between x₁ and x₂ is:
(f(x₂) - f(x₁)) / (x₂ - x₁)
Plug in the quadratic:
[a(x₂² - x₁²) + b(x₂ - x₁)] / (x₂ - x₁)
= a(x₂ + x₁) + b
Stop. Look at that result.
The average rate of change over [x₁, x₂] equals a(x₁ + x₂) + b. Day to day, it depends on both endpoints. Not just the width of the interval — the location matters Not complicated — just consistent..
This is huge. For a linear function, sliding the interval along the x-axis does nothing to the rate of change. For a quadratic, sliding the interval changes the answer.
Example That Makes It Click
f(x) = x². Simple. a = 1, b = 0 Most people skip this — try not to..
Interval [1, 3]: average rate = 1(1+3) + 0 = 4
Interval [2, 4]: average rate = 1(2+4) + 0 = 6
Interval [10, 12]: average rate = 1(10+12) + 0 = 22
Same width (2 units). Worth adding: completely different rates. The parabola gets steeper as you move right.
The Midpoint Shortcut (This Is Cool)
For any quadratic, the average rate of change over [x₁, x₂] equals the instantaneous rate of change at the midpoint (x₁ + x₂)/2.
Check: f(x) = x², interval [1, 3]. Midpoint = 2. Day to day, derivative f'(x) = 2x. At x = 2, f'(2) = 4. Matches the average rate we computed It's one of those things that adds up. Simple as that..
This isn't a coincidence. That said, it's a property of quadratics — the secant line over any interval is parallel to the tangent line at the midpoint. You can prove it with the algebra above. The average rate formula a(x₁ + x₂) + b evaluated at the midpoint x = (x₁ + x₂)/2 gives 2a((x₁+x₂)/2) + b = a(x₁+x₂) + b. Same thing.
This shortcut saves time. It also hints at the deeper connection between average and instantaneous rates — the Mean Value Theorem, if you're heading toward calculus.
From Average to Instantaneous: The Limit Idea
Here's where the story goes from "algebra with intervals" to "calculus."
Average rate of change uses two points. Instantaneous rate of change asks: what happens when those two points collapse into one?
You shrink the interval [x, x+h] until h → 0. The average rate becomes:
lim(h→0) [f(x+h) - f(x)] / h
That's the derivative. For f(x) = ax² + bx + c:
f(x+h) = a(x+h)² + b(x+h) + c
= ax² + 2axh + ah² + bx