The task: three bubbles appear, each holding one arithmetic expression. Click them in ascending order of the result — smallest first. Not "find the largest": sorting by computed value is a different skill and this is the one the exam tests.
Target: 20 of 25. That is the only corroborated number in this whole round — one selected candidate advised aiming for at least 20, and a second independently reported scoring exactly 20 and progressing. Treat it as the floor.
| Detail | Status | Where it comes from |
|---|---|---|
| Select in ascending order of the result | REPORTED | REV-001 verbatim, + 8 online sources |
| Exactly 3 bubbles, one binary expression each | REPORTED | Observed on a coaching walkthrough screen: 56/2, 32-10, 5*5 |
| 15 seconds per question, 25 questions | REPORTED | REV-001 + justnk.in; a third source says 14 s |
| Mixed + − × ÷, and the sums are easy | REPORTED | REV-001: "simple arithmetic". The clock is the difficulty, not the algebra |
| Decimals appear | REPORTED | REV-002: "some questions came from decimal also so practice that as well" |
| The three results sit in a narrow band | REPORTED | The observed screen's values were 28, 22, 25. Widely-spaced values could be ranked by eye; close ones force the arithmetic |
| Wrong order scores 0 for that question | RECONSTRUCTED | No source describes partial credit either way |
| Timeout scores 0 and auto-advances | RECONSTRUCTED | No source describes what expiry does. Chosen as the harshest consistent reading |
| No negative marking | RECONSTRUCTED | One prep site says so; no candidate has ever confirmed it. Toggle it on in Practice to rehearse the worse case |
| Fractions | RECONSTRUCTED | No source reports fractions. Off in Exam mode |
| Keyboard keys 1 / 2 / 3 | RECONSTRUCTED | The real test is pointer-only. Keys are here so the mouse is not your bottleneck while drilling |
| 0.7s answer reveal between questions | RECONSTRUCTED | A deliberate deviation. The exam almost certainly does not show you the values. It is here because seeing why you were wrong immediately is the fastest way to improve. It adds about 18 seconds to a 6m15s session; the clock itself is unaffected because it stops during the reveal |
| Practice timer will not go below 3 seconds | RECONSTRUCTED | A floor, so a mis-typed option cannot make the game unplayable |
Full sourcing: research/01-mechanics-spec.md and data/reviews-online.csv.
Exam mode locks the reported real parameters: 25 questions, 15 seconds each with a hard reset, 3 bubbles, decimals on, fractions off, no negative marking, difficulty ramping across the session.
The pooled clock gives you the whole budget at once (questions × seconds) instead of resetting every question. Two sources say the real clock is per-question and one clone says the session is pooled — this is here so you can rehearse the alternative, not because it is the likely one.
Click the bubbles from the smallest result to the largest. Keys 1 / 2 / 3 work too.