Mathematics · Foundations

Logarithms are the name for a question you can already answer

You almost certainly got 5, because 2 multiplied by itself five times is 32. Hold onto how that felt, because you just performed a logarithm. You didn't need the word, and that's the whole point: a logarithm is not a new kind of math, it's a name for the question you just answered.

The question was: "2 raised to what power gives 32?" The name for that question is log₂(32), and the answer is 5. Read it out loud as "log base 2 of 32," but hear it as "2 to the what makes 32."

Start from what you know: powers of two

You already carry this table in your head. The logarithm is simply reading it from right to left.

Going right (exponent → result) is exponentiation. Going left (result → exponent) is the logarithm. They are the same staircase walked in opposite directions, which is why a log is called the inverse of an exponential.

See the inverse

Because they undo each other, their graphs are mirror images across the diagonal line y = x. The exponential rockets upward; the logarithm is that same curve flipped, rising fast and then flattening.

Work one, then finish one

Worked: log₂(8) asks "2 to the what makes 8?" Since 2³ = 8, the answer is 3.

Your turn: log₁₀(1000) asks "10 to the what makes 1000?" Fill the blank: 10 to the ___ is 1000. (Answer: 3, because 10³ = 1000. Notice base 10 is just a different staircase.)

Why this earns a place in your toolkit

Logs turn multiplication into addition and compress huge ranges into readable ones, which is why they run quietly underneath the things you care about. Decibels, earthquake magnitudes, and pH are all log scales. In machine learning specifically, log-loss and log-likelihood are everywhere, precisely because logs convert products of probabilities into sums that a computer can add without underflowing to zero. When you meet those terms later, they are this same idea wearing a work uniform.