Now the measuring stick. This is the only code in the course you are handed finished. Everything else you write yourself.

You get it for two reasons: counting is not the interesting part of any algorithm, and every lesson has to measure with the same instrument or the numbers cannot be compared.

Create metrics/counter.go and main.go (see the panel).

Why we count operations instead of seconds

Run your search over a thousand items and time it with the clock. Here is what you actually get (Windows, real run):

n = 1000, searching for a title that is not there
  run 1:  wall 0s           comparisons 1000
  run 2:  wall 0s           comparisons 1000
  run 3:  wall 0s           comparisons 1000
  run 4:  wall 0s           comparisons 1000
  run 5:  wall 0s           comparisons 1000

The clock says 0 s. Not "fast" — zero. The work happened: a thousand comparisons, five times running, the same number each time. The operating system's clock is simply too coarse to see it.

This is not a made-up example. It is a bug that was found and fixed in the transit app this lesson's figures come from: sub-millisecond runs timed as "0s", "which would have made the fastest implementation look untimed rather than fast".

A comparison count has none of those problems:

• it is exact — 1000, not "about a thousand";
• it is deterministic — run it a hundred times, get the same answer;
• it is machine-independent — your number matches your classmate's, even if their laptop is twice as fast.

That last property is what makes every later lesson's test possible: a test can demand "no more than this many comparisons" and mean the same thing for everyone.