Environment Signals Check

doc

Review device environment signals such as fonts, audio context, storage estimates, and runtime capability exposure.

Environment checks look at the browser as a machine, not as a network. What is the screen, what fonts are installed, how many cores are available, is there a touch interface. Individually these are trivia. Together they describe a configuration closely enough to be a tracking handle, and that is the point of measuring them.

The properties in play

PropertyWhat it narrowsHow stable
Screen resolution and device pixel ratioPanel type and OS scaling settingsVery stable; changes on monitor or zoom change
Available fontsOS and installed softwareHigh; changes on install or OS update
Hardware concurrencyCPU core countFixed per device
Device memoryRAM tier, coarsely roundedFixed per device
Touch supportInput classFixed, except on hybrid devices
Plugins and MIME typesLegacy configuration; now mostly emptyLow value after plugin removal

Font enumeration is the strongest of these, and the least discussed. A default Windows install, a design workstation and a developer laptop have three different font sets, and each set is readable without any permission prompt.

Why a combination matters more than any field

Resolution alone matches millions of screens. Core count alone matches millions of machines. The intersection — a specific resolution at a specific pixel ratio, with a specific font set, a specific core count and a specific memory tier — matches far fewer. This is the arithmetic that makes environment checks worth running: entropy comes from the join, not from the parts.

The corollary is also true and easy to forget: changing one field to a common value does not reduce identifiability much, and changing it to an unusual value increases it. Randomising the parts is a weak strategy against a join.

What the check reports

  1. Display configuration. Resolution, pixel ratio, colour depth, and whether the layout viewport differs from the physical screen — the last tells you about zoom and scaling.
  2. Font availability. Which of a reference set are present. The set matters, not the count.
  3. Compute tier. Core count and memory bucket. Both are deliberately coarse in modern browsers.
  4. Input class. Touch points, pointer type, and whether a keyboard-like device is claimed.
  5. Legacy surfaces. Plugins, MIME types, and older APIs. Usually empty, which is itself informative about the browser's age.

How to read it

  • A distinctive font list is the loudest signal. If the set is unusual, everything else matters less.
  • Pixel ratio mismatches are common and benign. A laptop at 150 percent scaling reports a ratio that does not match the panel's native density. This is normal, not suspicious.
  • Empty legacy surfaces are expected. An absent plugin list means a modern browser, not a stripped one.
  • Hybrid devices legitimately report both touch and a fine pointer. A convertible laptop is not a contradiction.

What cannot be fixed

You cannot hide your screen size without breaking layout. You cannot hide installed fonts without breaking text rendering for the sites that need them. You can reduce what is reported in some cases — but the underlying configuration is what makes the browser usable, and the properties derived from it are what make it recognisable. The realistic goal is awareness, not concealment.

Common misreads

  • "My resolution is common, so I am anonymous." Common fields contribute little entropy. The join is what narrows.
  • "The check found unusual fonts, so something is wrong." Unusual fonts mean unusual software. Not a fault.
  • "Changing zoom will help." It changes one reported value and usually creates a mismatch with the physical panel.
  • "Empty plugin list means the check failed." It means the browser no longer supports the plugin model, which is the norm.

FAQ

Does environment data identify a person?

It identifies a configuration. The same model with the same software produces the same profile, which is why it is a class, not an individual.

Why does my pixel ratio not match my screen?

OS scaling and browser zoom change the reported ratio independently of the panel. A mismatch is expected on scaled displays.

Can I reduce what I expose?

Partially, through browser settings and reduced API surface. The configuration itself remains visible in aggregate through derived properties.

Is font enumeration still possible?

In reduced form. Browsers have restricted direct enumeration, but presence tests against a reference set still work, and the resulting set is often distinctive.