Choose the right input category

Mouse vs touchpad scroll events

A mouse wheel and a touchpad can both scroll a page, but the browser does not receive a universal physical-input count from either. Declare the input type instead of asking the trace to guess, then compare only runs made with the same setup.

The four declared categories

Detented mouse

A wheel used in its stepped or ratcheted mode. Browser event count still must not be treated as a physical detent count.

Free-spin mouse

A wheel used in an unlocked or momentum-like mode. One spin can continue producing input without repeated finger actions.

Touchpad

A pad gesture translated into wheel events by the operating system or browser, often with acceleration and momentum.

Unknown

Use when the device or wheel mode cannot be confirmed. Unknown runs remain useful observations but should not anchor device-specific comparisons.

How the event patterns can differ

ObservationDetented mouseFree-spin mouseTouchpad
Physical actionStepped wheel rotationContinuous wheel spinFinger gesture on a surface
MomentumUsually limited by the wheel mechanismCan continue after releaseOften added by the OS
Typical trace shapeMay look grouped or discreteMay form dense burstsMay be high-resolution and tapered
Reliable inferenceNone of these shapes proves the hardware category

These are possible patterns, not classification rules. Drivers, OS preferences, browsers, applications, and device firmware can make traces overlap.

Event count is not a notch count

A physical detent does not have a required one-to-one mapping to a web WheelEvent. The input stack may split, combine, accelerate, suppress, or schedule information before JavaScript receives it. Counting received events is deterministic for a captured trace, but describing that number as physical notches would be an unsupported conversion.

The challenge therefore reports browser events per second. A free-spin mouse may generate a very different event stream from a detented wheel, and a touchpad has no wheel notches at all. ScrollTest provides no universal quality band or cross-device percentile for those values.

Delta magnitude needs its unit

Each event reports deltaX and deltaY plus a delta mode: pixels, lines, or pages. Pixel deltas are browser/CSS movement units rather than a physical distance. Line and page values depend on settings and document context. A larger value is not automatically faster, better, or more accurate.

ScrollTest keeps unlike modes separate and displays their labels. The full measurement methodology defines the stored fields and comparison scope.

Direction and momentum

ScrollTest classifies a nonzero vertical delta as up or down. When vertical delta is zero, it reports a nonzero horizontal delta as horizontal-only and no movement on either axis as true zero. System direction settings may change how a vertical sign relates to finger motion. Touchpad momentum and a spinning wheel can continue emitting input after release, so changing physical direction before momentum ends may create an opposite-direction observation without a hardware fault.

During the controlled diagnostic, wait for a complete pause between down and up phases. If reversals persist across deliberate repeated runs, follow the reverse-scroll checklist and compare environments one variable at a time.

Rules for useful comparisons

  • Compare detented only with detented, free-spin with free-spin, and touchpad with touchpad.
  • Keep the device, connection, browser, OS settings, test intent, duration, and metric version fixed.
  • Compare pixel totals only with pixel totals, line with line, and page with page.
  • Use at least three valid same-scope protocols; exclude visibility-interrupted or materially delayed samples.
  • Read event rate and delta movement as separate measurements.
  • Treat an unknown-category result as an observation, not a device benchmark.

What a browser test can answer

ScrollTest can show whether the browser received events, which units and directions it reported, how the delivery intervals varied, and whether the same declared setup produced a repeatable trace. It can flag a pattern worth retesting and produce a local report for support.

It cannot prove that every physical action generated an event, diagnose an internal encoder, identify a device model, establish universal accuracy, or make different input categories directly comparable. For environment-specific limits, see browser compatibility.