Your fingerprint, measured

The traits a script can measure

Here are the main traits fingerprinting scripts read, in plain words (checked October 2026; exact details vary by browser):

• Canvas: the page secretly draws some text and shapes, then reads back the pixels. Your graphics chip, drivers and fonts make tiny differences in the result, so the picture becomes a code such as 7d2a.
• Fonts: the page measures how wide a word is in many font names. If the width changes, that font is installed. A designer with 300 fonts looks very different from a fresh laptop.
• Screen: width, height, colour depth and pixel density.
• Time zone and language: what your computer is set to, for example Europe/London and en-GB.
• GPU: the graphics chip. Through a drawing feature called WebGL, many browsers tell pages a model name.
• Audio: the page processes a short sound without playing it. Small numeric differences in the result depend on your hardware and software.

Intl.DateTimeFormat().resolvedOptions().timeZone
navigator.language
screen.width + "x" + screen.height
navigator.hardwareConcurrency

New to code? Read this first

This lab uses a little Python. You only need four ideas, and the code is mostly written for you:

• A dict is a set of labelled values, like a form: {"timezone": "Europe/London", "screen": "1920x1080"}. person["timezone"] reads the value labelled timezone.
• A list is a row of things in square brackets. Our crowd is a list of dicts, one per visitor.
• A for loop, for person in crowd:, runs the indented lines once for each visitor.
• if runs its indented lines only when a test is true. == means "is equal to" and != means "is not equal to".

Counting things in Python looks like this. It is the pattern you will use:

count = 0
for person in crowd:
    if person["screen"] == "1366x768":
        count = count + 1
print(count)

What you will see

Each trait on its own is shared by several people: half the crowd is in London. But each extra trait removes everyone who differs on it. By the fourth trait, only you remain. In a real crowd of millions the same thing happens, it just takes more traits.

function countMatching(crowd, visitor, traits) {
  let count = 0;
  for (const person of crowd) {
    if (traits.every(t => person[t] === visitor[t])) count++;
  }
  return count;
}

func countMatching(crowd []map[string]string, visitor map[string]string, traits []string) int {
    count := 0
    for _, person := range crowd {
        same := true
        for _, t := range traits {
            if person[t] != visitor[t] {
                same = false
            }
        }
        if same {
            count++
        }
    }
    return count
}