METHODS / RESULTS / LIMITS

How MouseCheck results work.

Use this reference to choose a test, decode its named result and repeat it under a controlled condition.

MouseCheck reports a signal delivered to this browser, a context you declare, or a calculation from values you enter. None of those is automatic access to mouse firmware, an internal switch or the USB report stream.

START WITH THE QUESTION

Nine routes. Nine different outputs.

A broader number is not a better answer. Choose the page whose input and endpoint match the decision you need to make.

01

Signals in this browser

02

Measurements and settings

03

Troubleshooting after a result

READ THE EVIDENCE AT ITS LEVEL

A result cannot silently climb this ladder.

  1. 1

    OBSERVED

    Browser signal

    An event, timestamp, delta, unit or count delivered to this page.
  2. 2

    DECLARED

    Your physical action and context

    The button, direction, input type, app, connection or setting you report.
  3. 3

    CALCULATED

    Rule-based result

    A named status, median, rate estimate, formula output or deterministic next-check path.
  4. ?

    NOT IDENTIFIED

    Internal hardware cause

    Requires a separate physical, electrical, USB or display endpoint. The browser result does not supply it.

Example: one opposite wheel event is observed. “I rolled down only” is declared. “Unexpected direction change” is calculated. “Worn encoder” is still not identified.

RESULT DICTIONARY

Find the completion rule before trusting the label.

These are MouseCheck’s current product rules. Site-defined minimums and comparison bands are disclosed; they are not presented as universal hardware standards.

Result familyComplete whenAllowed conclusionNext action
Basic input: Normal / Does not meet / Not checkedMovement, main, secondary and both vertical wheel signs are resolved in one round.Required signals did or did not reach this browser check under the named condition.Return to the complete test or choose the symptom-specific page.
Unwanted repeat / Missing input / No repeat observedThe declared separate-press plan is stopped cleanly and downs/completions can be compared.The selected button produced matching, extra or incomplete browser actions in that round.Choose the double-click next check.
Scroll consistent / Reversal / Same-sign phases / Not enough inputDown and up phases are stopped; each needs eight vertical events for a complete consistency comparison.Browser event signs stayed consistent, changed inside a phase, failed to separate or were insufficient.Choose the scroll next check.
Approximate browser movement sample rate / No estimateFive seconds finish with at least 20 usable short gaps whose duration totals at least 3,000 ms.An approximate browser-observed sample rate and closest standard comparison band, or why no estimate exists.Repeat the same motion after changing one condition; use hardware tools for USB report rate.
Browser click-response roundTen trusted main-button samples finish in the fixed target without focus interruption.Median/P95 browser event age and event-to-first-rAF callback timing for that tab and round.Use the guide’s physical/USB/display protocol for a mouse-only or click-to-light question.
Configured DPI / Documented stage / Estimate / Not DPI / UnknownThe source, value and active-stage evidence are declared where applicable.Whether a value is usable as configured DPI, only an estimate, an OS setting or still unknown.Use a qualified number in the converter, or keep it unknown.
Sensitivity starting value / Measured correctionPositive source/target values and a supported/reference or custom linear yaw are provided.A calculated horizontal hip-fire starting setting under shown assumptions.Test one slow full turn in the target mode and correct from measured distance if needed.
Troubleshooting next-check planRequired observation/context fields are selected.A deterministic comparison path based only on those declarations.Perform the three actions in order; internal cause remains unidentified.

EXACT METHODS BY FAMILY

Inputs, completion rules, outputs and boundaries.

Open the family that matches your result. The compact field strip gives the contract; the notes explain the calculation or event rule.

01 / BASIC INPUT

Complete browser mouse check

Which basic mouse signals reached one test area?

Open test ↗
INPUTTrusted movement, buttons and wheel
COMPLETE RULEFive required signals resolved
OUTPUTNormal / Does not meet / Incomplete
BOUNDARYEvent presence, not mouse certification
  • Required: mouse movement, main-button down, secondary-button down, positive vertical wheel delta and negative vertical wheel delta.
  • “Tried, no signal” records the visitor’s attempt and prevents a Normal result while no matching event arrives.
  • Middle, drag, browser `dblclick`, horizontal wheel and side buttons are optional evidence. Back/Forward zero is inconclusive because a browser may navigate before the page receives an event.
  • Release changes held state; a completed drag needs at least 24 CSS pixels and release inside the area. Ctrl-modified zoom-style wheel input is excluded.

A Normal result means this round met the published five-signal browser rule. It does not rule out an intermittent problem or inspect physical parts.

02 / BUTTON REPEAT & RECOGNITION

Separate unintended repeat from intentional double-click recognition

Two tasks share one page but keep independent conclusions.

Open test ↗
INPUTDeclared separate presses or intentional primary pair
COMPLETE RULEClean Stop against a 5–100 press plan
OUTPUTMatched / Repeat / Missing / Recognition
BOUNDARYNo switch-wear or universal timing verdict

Unwanted-repeat round

  • Default plan: 20 separate physical presses; visitor may choose 5–100.
  • Main completion is native `click`; right/middle completion is a matching in-area release.
  • More selected-button downs than declared, or a native main `dblclick` during the single-press protocol, is repeat evidence. Fewer downs/completions is missing or incomplete input.
  • Right/middle completed pairs with down gaps ≤500 ms are a non-overlapping pace note only; they do not decide the verdict.

Intentional recognition

  • The separate module listens only for the browser’s native primary-button `dblclick`.
  • It shows the latest down-event gap and names recognized, not recognized or insufficient input.
  • Operating-system double-click speed can affect recognition; it is not the same as accidental-repeat evidence.

MDN `dblclick` semantics ↗

03 / WHEEL DIRECTION & EVENT RATE

Declare physical direction; observe browser direction

Signal consistency and event density are separate outputs.

Open test ↗
INPUTDown/away phase, then up/toward
COMPLETE RULE8 vertical events per stopped phase
OUTPUTConsistent / Reversal / Same sign / Insufficient
BOUNDARYBrowser deltas, not physical detents

Direction protocol

  • Each phase’s first nonzero vertical `deltaY` becomes its reference. Later events are same or opposite to that phase reference.
  • A complete consistent result needs zero within-phase reversals and opposite reference signs across the two declared physical phases.
  • Zero or 1–7 vertical events remain unverified. Eight is MouseCheck’s sampling rule. Each phase keeps at most 200 events; units remain pixel, line or page.

Five-second event-rate sample

The clock begins with the first nonzero vertical event. The page reports vertical events ÷ 5, sign counts and reported units. It does not grade the number, call it polling frequency or add unlike units into distance.

MDN wheel-event boundary ↗

04 / MOVEMENT FREQUENCY

Approximate browser-observed movement samples per second

A controlled sample, not a direct USB report-rate reader.

Open test ↗
INPUTFive seconds of continuous mouse movement
COMPLETE RULE≥20 usable gaps and ≥3,000 ms usable time
OUTPUTApproximate rate + comparison band
BOUNDARYBrowser scheduling and coalescing apply

Collection

  • Trusted mouse or unclassified mouse-compatible Pointer Events are accepted; identified touch/pen input is excluded.
  • Coalesced children replace their parent when available. A `mousemove` fallback is accepted only when no Pointer Event arrived in the previous 100 ms.
  • Only increasing positive gaps ≤100 ms are usable. Longer gaps are pauses. Focus loss discards the round.

approximate samples/s

1,000 ÷ median usable gap in msSynthetic example: 8 ms median → about 125 samples/s.

The nearest 125/250/500/1000/2000/4000/8000 Hz band is a MouseCheck comparison label. Same/different selected band is not a quality grade or USB certification. W3C coalesced events ↗

05 / BROWSER CLICK RESPONSE

Event age and first animation-frame callback

The endpoint begins after the operating system and browser expose an event.

Open browser check ↗
INPUT10 trusted main-button presses
COMPLETE RULEFixed target, active tab, no interruption
OUTPUTMedian, nearest-rank P95, spread and samples
BOUNDARYNo switch-to-USB or click-to-light endpoint

Browser-clock endpoints

  • Event age at the handler = performance.now() − normalized event.timeStamp.
  • The second value runs from normalized event timestamp to the first requestAnimationFrame callback after target state changes.
  • Epoch-style timestamps are normalized against performance.timeOrigin; incompatible values fall back to the handler clock and are labelled.

Interpretation

A requestAnimationFrame callback occurs before a paint opportunity; it does not prove completed pixels or photons. Mouse-only latency needs controlled actuation plus USB reports. Click-to-display latency needs synchronized input and visible-light endpoints.

Choose the correct latency method ↗

06 / DPI SOURCE QUALIFICATION

The source decides whether the number is usable

Configured, documented and estimated values are not interchangeable.

Open guide ↗
INPUTSource, value and active-stage evidence
COMPLETE RULESource-specific required fields resolved
OUTPUTConfigured / Stage / Estimate / Not DPI / Unknown
BOUNDARYNo sensor or firmware inspection
STRONGEST FOR CONFIGURATIONActive value shown in supported device software
CONDITIONALExact-model stage table + proof that stage is active
ESTIMATE ONLYRepeated browser movement units ÷ physical inches
NOT DPImacOS Tracking speed or Windows Mouse pointer speed

A stable browser ruler estimate still does not read the active firmware profile. If stage or source cannot be proved, the correct result remains unknown.

07 / SENSITIVITY CONVERSION

Calculated horizontal hip-fire turn-distance match

A transparent starting value that must be checked in the target mode.

Open converter ↗
INPUTProfiles/yaw, sensitivity and source/target DPI
COMPLETE RULEPositive values + linear yaw model
OUTPUTTarget setting, cm/360, counts, rounding and correction
BOUNDARYNo ADS/FOV/acceleration or identical-feel claim

Target settingsource sensitivity × source DPI × source yaw ÷ (target DPI × target yaw)

cm/360914.4 ÷ (DPI × sensitivity × yaw)

Counts/360360 ÷ (sensitivity × yaw)

Measured correctiontested target setting × observed cm/360 ÷ desired cm/360

CS2/Apex 0.022° and VALORANT 0.07° are exposed as reference profiles, not publisher/current-build certified constants. Use Custom only with a documented linear yaw. Verify one slow physical full turn and keep ADS/scopes, acceleration and perceived speed outside this result.

CONTROLLED A/B

A comparison is useful only when the rule stays fixed.

  1. 1

    Freeze

    Name the device/input, browser/app, connection, power, profile and intended action.

  2. 2

    Complete

    Meet the page’s full stop/sample rule. Do not save an interrupted or insufficient round as normal.

  3. 3

    Preserve

    Keep the named result and its evidence, units, sample count and optional note.

  4. 4

    Change one

    Alter one app, setting, profile, port, receiver, power condition, device or computer.

  5. 5

    Repeat

    Use the same physical action and completion rule under the new condition.

  6. 6

    Stop at the boundary

    Report where the pattern follows; do not turn that scope into an unidentified internal cause.

CURRENT PAGE DATA FLOW

What stays in the page, what copies and what remains unknown.

This is the behavior of the current local source. It is not a claim about a future host, edge log, analytics product or provider retention.

Page familyCurrent input in the open pageClear / retain behaviorUser-triggered output
Home, double, scroll, movement rate, latencyBounded event/timestamp lists, counters, declared conditions and up to two saved A/B summaries where offered.Reset clears the live round; Clear A/B clears saved summaries; reload clears page memory.Copy or `.txt` download only after its button is pressed.
Sensitivity converterEntered profiles/settings plus calculated result, rounding and optional measured correction.Editing an input clears the stale result; reload clears the page.Copy only after pressing Copy result.
DPI and troubleshooting guidesSelected declarations, optional short note where offered and the current deterministic result.Editing, Reset or reload clears the result. No saved A/B.No copy/download control on these planners.
Production hostingServer, CDN, edge, access logs, analytics and provider processing.Not established by this local source review.UNAVAILABLE until the real host/data flow is reviewed.

VERIFICATION STATUS

Local evidence is not production or hardware proof.

Each current page family has a pre-code research/specification record, a synchronized local build and focused browser acceptance. That establishes the implemented browser behavior under those tests.

LOCAL PASSRule/code alignmentCurrent pages, scripts, formulas, named outcomes and local browser paths.

NOT TESTEDPhysical device matrixActual model/firmware, free-spin, trackpad, Windows/macOS combinations and assistive technology.

UNAVAILABLEProduction and fieldHost logs/data flow, deployment parity, indexing, real-user comprehension and field performance.

REFERENCE BOUNDARY

Use the result within the endpoint that produced it.

These methods do not inspect switches, wheel encoders, mouse firmware, raw USB/wireless reports or the instant a display changes light. A browser movement value is not an automatic reading of mouse DPI. A calculated game setting is not a publisher-certified promise of identical feel. Keep the named condition, completion rule and uncertainty with every result.