Story, art and structure all matter, and we get to them. But a game is a machine you operate, and if the operating is wrong nothing downstream survives it. So we start with the controls and we measure them.
The order is fixed, because the order is what makes one review comparable to the next.
End-to-end delay from button press to visible reaction, filmed with a high-speed camera rather than reported by the game itself. Four pads, two keyboards, 100 samples each.
How long an action commits you. We time animation locks, turn rates, recovery windows and the exact number of frames in which a decision cannot be reversed.
Whether the game tells you what it just did. We record failures and check, one by one, whether the reason was visible on screen at the moment it happened.
Only now do we discuss writing, structure, art and pacing — with the mechanical facts already on the table, so praise and complaint both have something to stand on.
The phrase appears in almost every positive review of an action game and it carries no information whatsoever. Two games described as tight can differ by forty milliseconds of response and by half a second of animation commitment, which is the difference between a game that feels like an extension of your hand and one that feels like sending instructions by post.
We think the honest thing is to publish the numbers and then say what they mean. A 90 millisecond response is fine in a strategy game and disqualifying in a fighting game, and the only way to make that judgement usefully is to know the figure first.
Each entry lists the response figure we recorded, the hours played and the final score.
Platformer
9.2
A 41 millisecond response, a jump arc you can predict from the first minute, and a coyote-time window of exactly six frames. It is the rare precision platformer where every death is legibly your own.
Parkour
7.4
The movement model is genuinely excellent and the game repeatedly removes your control of the view to show you something. We counted 214 forced camera events in a fourteen-hour campaign.
Fighting
8.6
Frame data is in the game rather than on a wiki, the training mode shows recovery visually, and the online delay is disclosed on screen. The roster is small and it barely matters.
Shooter
8.9
An option to display the exact collision area sounds like a concession to beginners. In practice it turns an opaque bullet-pattern game into a readable one without making it easier.
Racing
8.1
The tyre simulation is not the most accurate we have driven, but the wheel weight tells you what the surface is doing. We measured steering response across three wheels and one pad.
Mech action
7.8
Deliberate sluggishness is a legitimate design choice when the recovery windows are honest. Three of the eight weapons break that contract, and the review names all three.
Feel is roughly half the score here. That is unusual, deliberate, and stated on every review page so nobody has to guess.
"I play on a pad and half of all reviews are written by people on a keyboard who never mention it. Publishing the figure for each controller is the single most useful thing on this site."
"The readability section talked me out of a game I was certain I wanted. It turned out my objection to the last three games I bounced off was exactly the thing you measure."
"You are the only place I have seen treat animation commitment as a separate axis from response time. Those are two different problems and everybody else lumps them together."
