REACTION LAB
A plain reaction-time test with no 3-D scene, no aiming and no network, timed by the same stopwatch the duel uses. Run it on the same device as another test, under the same conditions, and the two numbers can be compared directly.
| # | Reaction (ms) | Frame wait (ms) | Foreperiod (ms) | Input |
|---|
WHAT IS BEING TIMED
- Start: the frame that first draws the cue. The browser takes the timestamp just before that frame is painted, not when the program decided to show it (on a busy device those differ by up to a frame or more; the frame wait column shows by how much, and none of it is counted against you).
- Stop: the press's own timestamp, which the browser records when the input arrives from the device rather than when the page's code gets round to it.
- Not visible to any web page: how long the screen takes to show a frame once it has been handed over (typically 10–40 ms, more on TVs and some laptops) and how long the mouse, keyboard or touchscreen takes to report a press (about 1–8 ms for a wired mouse, 20–60 ms for many touchscreens). These make every browser test read a little slower than the true reaction, by roughly the same amount on every trial on the same device.
- Sound cue: timed from when the browser says the sound reaches the speakers (its reported output latency is added). Bluetooth headphones add 100–250 ms that browsers often under-report: use wired audio or the screen cue.
- A press within 100 ms of the cue counts as a guess (as in the duel) and the trial is run again.
HOW TO COMPARE WITH ANOTHER TEST
- Same device, same input (mouse, key or finger), same hand, same sitting. Wired audio if using sound.
- Match the cue: compare the screen cue with another visual test, the sound cue with an auditory one. People react to sound roughly 30–50 ms faster than to light, and to both together faster still.
- Run 20 or more trials on each and compare medians; single trials vary by ±20–40 ms in everyone.
- A steady difference of 10–40 ms is expected if the other test is calibrated for its hardware and this one is not (see above). A difference that wanders from trial to trial would point at a fault.
CAMERA CHECK (THE GOLD STANDARD)
To measure the device's hidden delays directly: tick Camera check, then film the screen and your hand together with a phone in slow motion (240 frames per second; each video frame is about 4.2 ms). The square in the corner turns white with the cue. Step through the video and count frames from the square turning white to the button going down (or the finger touching the glass), multiply by 4.17, and compare with the number shown on screen. The difference is this device's display-plus-input delay; it should be about the same every trial.
HOW THE DUEL DIFFERS FROM THIS TEST
- The duel's cue is a bell (sound) and a gold flash with the word DRAW! together, after a random wait of 2.4–6.2 seconds.
- With aiming on (the default), the duel time includes finding the head with the mouse or finger: it is a draw-and-aim time, not a reaction time, and has its own records board. Turn aiming off in the lobby for a pure reaction duel.
- Online, each player is timed on their own device exactly like this. The server only steps in if a device claims a time faster than its own connection makes possible; the result card says when that happens.