A cube net puzzle round is a short, repeatable spatial exercise where you classify flat patterns of six edge-connected squares as either foldable into a cube or not. Out of every way to join six squares edge to edge, mathematicians have shown that exactly 35 free hexominoes exist, and only 11 of those 35 survive a rigorous fold test. Every other six-square arrangement either leaves a face missing, fails to form a cube, or assigns two cells to the same cube side. The Cube Net Puzzle turns that combinatorial result into a fixed eight-question round so you can practice the classification with reproducible scoring, deterministic pattern order, and the same eight classifications across browsers and devices. The round is short enough for a coffee break, the math behind it is sourced from published references, and every answer is verified by a local hinge-propagation algorithm rather than a visual guess. Because the questions, the score formula, and the answer key are all fixed, two players running the round on different devices will see identical questions and identical scores for identical reasoning.
What Counts as a Valid Cube Net
A cube net is a flat arrangement of six squares joined edge to edge that can fold along its shared edges into a closed cube. That requirement is stricter than it looks. Just connecting six squares is not enough: the shape also has to wrap the cube without putting two squares on the same face, leaving a face uncovered, or forcing contradictory orientations at a loop.
When translations, rotations, and reflections are treated as equivalent, there are 35 free hexominoes — the complete set of shapes you can build by joining six unit squares edge to edge. Of those 35, exactly 11 fold into a cube. The result is well established in the mathematical literature. Wolfram MathWorld's entry on polyhedron nets describes the complete enumeration, and an American Mathematical Society feature on unfolding polyhedra catalogs all eleven cube nets and shows them together in its Figure 9. The Cube Net Puzzle records those 11 nets as canonical coordinate fixtures and pairs them with a local fold test that propagates a real three-dimensional orientation across every shared edge.
How the Fixed Round Is Built
The round is intentionally short and intentionally deterministic. There are eight fixed questions drawn from the full 35-shape set, selected as four valid cube nets interleaved with four non-nets in a stable order. That choice matters for two reasons.
First, the mix prevents you from leaning on a single heuristic. Pure guesswork gives you a 50/50 expected score at best, and a thoughtful reasoner still has to apply the 11-net knowledge to each specific shape. Second, the order is fixed, so the round is the same on every device, every session, and every browser. There is no random seed, no server-side feed, no changing answer key, and no hidden rotation. If you want to verify your logic against the previous run, the previous run is identical.
The fixtures are tiny coordinate tables checked against the sourcing tables in the local test suite. Each candidate pattern is normalized under all eight planar symmetries, the folding algorithm walks the cells with breadth-first search, and the round accepts a shape only when six connected cells produce six unique cube normals — the positive and negative x, y, and z directions. Anything that fails any of those conditions is treated as a non-net. The independent enumeration check grows polyominoes one edge at a time, collapses rotated and reflected copies to one canonical key, and confirms that the process yields 35 distinct free hexominoes, that the folding algorithm accepts exactly 11, and that those 11 keys match the sourced fixture table. The same harness also rejects duplicate cells, disconnected six-cell inputs, and a two-by-three rectangle whose faces would overlap when folded.
| Round element | What the fixed puzzle provides |
|---|---|
| Number of patterns | 8 fixed questions |
| Pattern mix | 4 valid cube nets + 4 non-nets |
| Pattern order | Stable across devices, sessions, and browsers |
| Source of valid shapes | Canonical coordinate fixtures matching the 11-net reference table |
| Validation method | BFS propagation of a real 3D orientation across every shared edge |
| Acceptance condition | Six connected cells with six unique axis normals |
Classify 8 Patterns in One Sitting
- Open the Cube Net Puzzle in any modern browser on phone, tablet, or desktop.
- Read the six edge-connected squares on the board and imagine each shared edge as a hinge.
- Mentally rotate the pattern around its hinges and check whether any two faces would land on the same cube side.
- Press Y (or tap Yes) if the six squares fold into a cube; press N (or tap No) if they do not.
- Read the score and mistake counter on the status line above the diagram.
- Repeat for all eight fixed patterns until the round ends or the run deadlocks.
- Press Restart at any time to restore question 1, zero the score, clear the mistakes, and unfreeze the board.
If you want a deeper walkthrough of the eight specific patterns, the guide Cube Net Puzzle: Classify 8 Six-Square Patterns covers the classification logic in detail. For touch-only or keyboard-only play, the companion guide Play Cube Net Puzzle on a Phone or Without a Mouse walks through the same controls without a mouse.
Reading the Fold Without a Model
The fastest mental shortcut is to count the hinges that would meet at a cube corner. A valid cube net always produces three squares around any folded corner, and each of those three is connected to the other two only through the folded cube — never through a direct edge in the flat shape. If two squares in the flat pattern share an edge and would also end up adjacent on the cube, the shape is at high risk of forcing a face overlap or a contradictory fold.
A second check is whether any two squares are positioned so that folding them around their shared hinge would push them onto the same cube direction. When that happens, the pattern is automatically invalid regardless of how the rest of the shape folds. The local fold test catches both cases because it propagates a real three-dimensional basis across every shared edge and rejects the shape the moment two cells produce the same axis normal.
A third useful cue is the silhouette of the pattern. Many non-nets look like an extended row of squares with extra cells that have nowhere to go. If you can find a single square whose only neighbors in the flat pattern would all fold to the same side of the cube, the shape cannot be a net. The puzzle deliberately interleaves these trap shapes among the four real cube nets so that pattern recognition alone produces roughly the same score as a coin flip.
Scoring, Mistakes, and Restart
Every correct classification adds 125 points. Every wrong classification leaves the same pattern on screen so you can reconsider its hinges, increments the mistake counter by one, and skips the 125 points for that round. Two mistakes end the run until you press Restart, which restores question 1, zeroes the score, clears the mistake counter, and unfreezes the board.
| Action | Result on the fixed board |
|---|---|
| Correct classification | +125 points, advance to next pattern |
| Wrong classification | Pattern stays, mistake counter +1, no points |
| Second wrong classification | Run deadlocks, board frozen until Restart |
| Finish all 8 correctly | Final score exactly 1,000 |
| Restart | Question 1, 0 points, 0 mistakes, board unfrozen |
| Keyboard shortcuts | Y votes foldable, N votes impossible |
| Boss key | Double-Escape from the shared game shell |
The full 1,000-point finish is reachable only by classifying all eight patterns correctly. Because the round is deterministic, the same eight answers across two runs must always produce the same intermediate states and the same final score. That makes the round useful for release verification, for practicing under known conditions, and for comparing reasoning across devices without worrying about changing patterns.
Why the Round Is Deterministic
Determinism is the property that makes the eight questions reproducible. There is no random seed selecting the patterns, no remote feed shipping the questions, and no server answer call returning the correct response. The shape set, the mix of four valid and four invalid patterns, the interleaving order, and the per-question classification are all hard-coded fixtures. Local and production browser tests can submit the same answer sequence, observe the same intermediate states, and finish at the same 1,000-point score every single time.
The same property also helps the player. You can rerun the round to test a new reasoning strategy against the exact same eight patterns, share a particular question with a friend by describing its shape, or compare two heuristics on identical inputs. Restart always resets to question 1 with a zero score and zero mistakes, so each attempt starts from the same baseline. Score persistence from the shared game shell records the run in your browser only — no account, no cloud sync, and no analytics upload tied to your answers.
What This Round Does and Does Not Measure
The Cube Net Puzzle is a recreational spatial exercise grounded in a real mathematical result. It is not a spatial-ability test, an IQ screen, an educational placement exam, a cognitive assessment, a medical screening, or an employment measure. Familiarity with the 11-net result, repeated play, input device, and guessing all change the outcome. The interface states this limitation plainly and the round should not be used to diagnose, rank, or label anyone.
The local fold test is also intentionally narrow. It validates six unit squares folding into a cube and nothing else. It does not check arbitrary polyhedron nets, tabs for physical papercraft, material thickness, cut tolerances, printable scale, or whether a real sheet can pass through itself during folding. For the stated six-square classification, the finite search, sourced 11-net table, and orientation-based golden tests give a reproducible answer rather than an intuition-only animation.
Playing Across Devices
The round works the same way on a phone, a laptop, or a desktop. Touch players use the large Yes and No buttons. Mouse players click the same buttons. Keyboard players press Y to vote foldable and N to vote impossible. The diagram is rendered as an accessible SVG with a descriptive label, the status line reports the current question and mistake count, and every interactive target stays comfortably sized on a narrow screen.
The shared game shell also supplies score persistence, Restart, and a double-Escape boss key so the round can be paused at any moment. Score persistence lives in your browser only — no accounts, no remote answer calls, no sensor permissions, and no package dependencies. The puzzle runs entirely in the browser as a React client island and uploads nothing during play. The coordinate tables are tiny integer pairs, and the enumeration plus fold checks operate on small sets and maps so their cost is bounded and negligible on ordinary phones and laptops. Styling uses only the existing color, border, type, and surface tokens so both light and dark themes stay readable without introducing a separate visual system.