The screen throws a hand and gives an order: beat it, or lose to it. Rounds marked 😐 mean exactly that; rounds marked 🐸 mean the opposite, so «beat it + 🐸» is a round you must lose. The difficulty is not the rule but the hand — your eye lands on the picture first and your finger starts moving before the badge has been read. The clock starts at two seconds and loses 45 milliseconds per correct answer until it stops at 0.85 seconds. One wrong hand or one timeout ends the run, and your best score stays in this browser only.
Exactly one hand is correct in every round. «Draw» is never used as an order — inverted, it would allow two different hands and make scoring ambiguous.
The frog appears in about 55% of rounds. Anything rarer and «just follow the order» becomes a habit, and the badge stops being read.
The clock stops shrinking at 0.85 seconds. Below that, the outcome is decided by human reaction limits rather than by judgement, so it would stop measuring skill.
Best scores are stored in this browser (localStorage) and never sent to a server.
A round is three things: the hand on screen (✊ ✌️ ✋), the order (beat it or lose to it) and the badge (😐 or 🐸). Those three fix the answer. 😐 keeps the order as written; 🐸 turns it inside out.
There are only four combinations. Beat + 😐 gives the winning hand, lose + 😐 the losing hand, beat + 🐸 the losing hand, lose + 🐸 the winning hand. Written out it is trivial; walking that table inside nine tenths of a second is the game.
Exactly one hand is ever correct, so no round is lost to ambiguity. Every miss traces back to the order in which you read the screen.
Rock-paper-scissors is overlearned. Seeing ✊ surfaces «✋ beats it» before any deliberate thought. That is an automatic response, and on frog rounds it is exactly the wrong answer, so an extra step appears: cancel the answer that arrived and choose its opposite.
This is the structure of the interference tasks used in laboratories — Stroop, flanker, go/no-go. When the automatic response conflicts with the instruction, people slow by roughly 0.15 to 0.3 seconds and make more errors. It is also why this clock never drops below 0.85 seconds: below that, the inhibition cannot finish in time at all.
So this game does not reward fast people; it rewards people who can stop an answer already on its way. The two abilities are different, and high scorers on a plain reaction test are often ordinary here.
The first round allows two seconds, enough to walk the table while learning it. Each correct answer takes 45 milliseconds off: 1.55 seconds at ten, 1.1 at twenty, and the 0.85-second floor from round twenty-six onward.
0.85 seconds is just enough for four steps — read the badge, read the hand, suppress the automatic answer, move the finger. Visual recognition takes around 0.2 seconds, the suppression 0.2 to 0.3, the movement another 0.15. There is almost no slack left.
The floor exists because anything faster is decided by biology rather than judgement. With difficulty fixed from round twenty-six, a score past that point measures how long you can hold up under constant pressure.
If frog rounds were rare — say one in five — people would build the habit of obeying and stop reading the badge. The occasional frog would then be a surprise rather than a decision, and the score would be mostly luck.
If the frog were near-constant, the opposite habit would form: always invert, never read. Either extreme removes the reading step that the game is built around.
Setting it just above half means neither answer is ever the safe default, so the badge has to be read every round. That reading is the ability the score is meant to capture.
Pass the phone around, one run each, lowest score takes the forfeit. A run usually lasts twenty to forty seconds, so ten people take under five minutes.
Ties can be settled with a single extra run, which starts fast enough to separate people quickly. Only one best score is kept per browser, so write the order down if you want to keep it.
Nothing about it requires drinks; what the forfeit is stays a decision for the group.
The screen needs four words: beat, lose, as-told and reversed. Everything else is a hand drawing, so the game is identical in every language and only the name changes — Japanese television fixed it as あとだしジャンケン, while other languages simply call it rock-paper-scissors in reverse.
Clock, frog ratio and scoring are the same everywhere, and your best score continues in the same browser.
You must lose. The frog inverts the whole order, so beat becomes lose and lose becomes beat. If the screen shows ✊ rock with «beat it + 🐸», the right answer is ✌️ scissors — the hand that loses to rock.
So that each round has a single correct answer. If «draw» were an order, inverting it would mean «do not draw», and two different hands would satisfy that. Keeping the orders to win and lose keeps the scoring unambiguous.
The picture reaches your eye first, and rock-paper-scissors is overlearned, so «the hand that beats this» surfaces before you decide anything. On frog rounds that automatic answer is wrong, and cancelling it costs roughly two tenths of a second. Psychologists call this an inhibition task; being slowed by it is the normal result, not a failure of attention.
Read the badge before the hand. Going hand-first means forming an answer and then undoing it — two jobs. Badge-first turns it into one job: find the hand that loses. Once the clock reaches 0.85 seconds, that ordering is most of the score.
No. It runs in the browser with no account, records stay on this device, and adding the page to your home screen makes it work offline.