Comparing two approaches in a missing object task means testing a preview-and-recall strategy against a candidate-elimination strategy on the same five fixed scenes, where every correct round awards exactly 200 points and a perfect run totals exactly 1,000 points. The preview-and-recall approach relies on committing the original scene to memory during the 2.4-second preview window, then matching what you saw to one of the four named candidates that appears below the changed scene. The candidate-elimination approach skips memory commitment and treats the four labeled candidates as a closed logic puzzle, narrowing them down by code, position, or visual mark until one option is left. Because the score, timing, and answer options are fixed, the only variable between approaches is your reasoning path, which makes What's Missing Game a fair test bed for either method. Both approaches route through the same pure candidate reducer for keyboard, touch, and pointer input, so you can switch methods between rounds without losing access to any candidate.
The Two Approaches You Can Compare
Each round of What's Missing Game presents two visual states: a complete original with five labeled item cards, and a changed version with one position visibly empty and four explicit candidates below it. That single structure creates two genuinely different reasoning paths, and the comparison is the entire point of the exercise.
The preview-and-recall approach treats the 2.4-second preview as the source of truth. During the preview, you encode the five items into a mental snapshot using the geometric mark, the two-letter visual code, and the plain English label that each card displays. When the changed scene appears, you compare your snapshot against the four candidates and pick the one that matches the empty position. This approach rewards strong visual encoding and is sensitive to preview timing, because anything you fail to encode during the window cannot be recovered later.
The candidate-elimination approach treats the four candidates as a closed set. Instead of memorizing the original scene, you inspect the changed scene, identify what is missing using the marked empty position, and then narrow the candidates using code, label, or layout clues until only one viable answer remains. This approach relies less on the preview and more on the consistency between the changed scene and the candidate set, which makes it useful when the preview feels too short to absorb five distinct items.
Both approaches produce the same scoring, the same error handling, and the same terminal states. Restart returns you to the Morning Desk cover with zero points and zero recorded errors, so each comparison begins from the same deterministic fixture rather than a resampled board.
How to Compare the Two Approaches in a Run
Run a fair A/B comparison on the same five fixed scenes using these steps.
- Open the Morning Desk cover and commit to one approach for the entire run, so score and error counts stay attributable to a single method.
- Press Start preview and study the five original item cards; encode them by mark, two-letter code, and label, or note only the codes if you are running the elimination approach.
- Wait for the preview boundary; the same layout returns with one position marked as gone and four explicit answer candidates below it.
- Apply your chosen method: match your preview memory against the candidates, or eliminate three candidates using code, label, and empty-position cues until one remains.
- Navigate the two-by-two candidate grid with the arrow keys, then press Enter or Space to confirm. Touch or pointer clicks call the same pure reducer.
- After the run, note whether any first-distinct wrong candidate produced a repairable error, and whether the run deadlocked on a second distinct wrong signature.
- Press Restart to return to the Morning Desk cover and rerun the identical five scenes using the other approach, then compare the two result lines.
The two runs give you comparable inputs because the original scenes are named fixtures (Morning Desk, Picnic Cloth, Workshop Bench, Travel Shelf, and Rainy Hall) and every changed scene is derived by removing exactly one identified item. Nothing else in the layout shifts, label changes, or code reorders, so any score difference between runs reflects your reasoning path rather than content drift between attempts.
Why the Game Makes the Comparison Fair
Several design choices in What's Missing Game remove the variables that would otherwise distort an approach comparison. The score formula is fixed at 200 points per correct round and 1,000 points for a perfect run, with no speed bonus, no streak modifier, no preview-skipping bonus, and no random multiplier. That means the preview-and-recall approach cannot win by being faster, and the candidate-elimination approach cannot win by avoiding the preview.
Pointer clicks, touch taps, arrow-key movement, and Enter or Space confirmation all route through the same pure candidate reducer. Input method therefore cannot change the score, the error identity, or the correct answer, which would be a hidden confound in any real comparison. Modifier shortcuts are ignored, and held activation keys are ignored after the initial keydown, so accidental native repeat cannot quietly submit a choice you did not intend.
The original-to-changed relationship is also tested independently. The specific missing item for each round is kept inside active gameplay rather than being published in the page description or rendered into hidden data attributes, so future-round answers never leak into the current DOM. Independent tests validate every original-to-changed relationship and candidate order without placing the answers where a determined player could read them out of the source.
Scoring, Errors, and Visibility Cancellation
A correct comparison immediately awards exactly 200 points and prepares the next covered scene, so five correct rounds produce exactly 1,000 points. The error model is deliberately recoverable: a first distinct wrong candidate records one error but leaves the changed scene and the candidates visible, so you can compare again and repair the round without losing the rest of the run.
Choosing that identical wrong candidate again does not consume another error because the reducer stores a stable round-and-candidate signature and treats a duplicate as an atomic no-op. Only a second different wrong candidate closes the run in deadlock, at which point the terminal result receives programmatic focus so keyboard and screen-reader users are brought directly to the outcome.
Preview timing is protected against background-page exploits. If the document becomes hidden during the original-scene preview, that preview is canceled and the round returns safely to Ready without recording an error. Returning to the tab requires starting the complete preview again, so the changed scene cannot silently appear after time expires in a hidden tab. This matters for a comparison because tab-switching during the preview window would otherwise hand an unfair advantage to the candidate-elimination approach.
Every surface, border, foreground text, accent state, and error state uses the site's existing design tokens, with no remote images, hardcoded color palette, or canvas dependency, so the two approaches see the same visible affordances at every step.
When Each Approach Has the Edge
The two approaches behave differently depending on the kind of visual comparison you are running. The table below summarizes where each one is strongest; the exact per-round figures come from your runs on What's Missing Game.
| Scenario | Preview-and-recall approach | Candidate-elimination approach |
|---|---|---|
| Preview feels long enough to encode five distinct items | Strong: one clean memory match per round | Workable but underuses the preview |
| Preview feels too short to absorb five items | Weak: memory carries too little detail | Strong: code and label cues narrow candidates |
| Mark, code, and label are visually similar across the five items | Weak: memory confuses near-duplicates | Strong: elimination isolates the visible difference |
| One candidate shares the empty position with an item you recognize instantly | Strong: recall pins the answer in one step | Weak: still has to rule out three distractors |
| You want to compare two runs with identical inputs | Both approaches benefit because scenes, candidates, and timing are deterministic |
If you want a stricter benchmark, run the same five fixed scenes once with each approach and record the final score, the number of first-distinct errors, and whether either run deadlocked on a second distinct wrong signature. Avoid Common Mistakes in Missing Object Games lists the failure patterns that skew such a comparison, including tab-hiding during preview, treating the duplicate wrong candidate as a fresh error, and restarting from a non-Morning-Desk cover.