Poly Puzzle: a strategy guide
Poly Puzzle is the daily game for people who think with their hands. Each day a simple low-poly object — a heart, a leaf, an arrow, a house — is sliced into a handful of irregular polygons and scattered into a tray. Your job is to drag, rotate and slot each fragment back into the outline.
It is descended from the tangram, but the pieces are not a fixed set. Because the silhouette is cut fresh every day, the fragments are always trapezoids, wedges and slivers you have never seen before, and the intended arrangement is never the one you remember from yesterday.
The rules, restated
- The dotted outline on the board is the target silhouette.
- Drag a fragment from the tray onto the board. Pieces stay where you drop them — nothing snaps back to the tray.
- Rotate the selected fragment with the rotate buttons below the tray, in either direction.
- On touch devices you can also twist with two fingers, or double-tap a piece to rotate it a step.
- The arrow cluster nudges the selected piece in small increments for fine alignment; hold an arrow to repeat.
- A fragment locks into place when it is close enough in both position and angle — there is a small tolerance, so you do not need pixel precision.
- The puzzle is solved when every fragment is locked and the silhouette is complete.
A worked example
Take a heart silhouette cut into six pieces. Two of them are large mirrored curves that form the lobes, one is a long triangle that becomes the point at the bottom, and three are small filler wedges.
The efficient route is to place the long triangle first. It is the only piece with a sharp, unmistakable apex, so there is exactly one place it can go, and once it is down it fixes the scale and orientation of everything else.
Next come the two lobes, which are easy to identify because they are the only pieces with a convex curve made of several short edges. With those three placed, the remaining wedges have almost no ambiguity left — each one has a unique combination of edge lengths that matches a single remaining gap.
Trying it in the opposite order, small pieces first, is much slower: a small wedge often fits three different gaps at three different angles, and you can spend a minute placing it in the wrong one.
How to read a fragment
Every fragment carries three pieces of information: its longest edge, its sharpest angle, and its overall area. Experienced players read all three before they start dragging. The longest edge tells you which boundary of the silhouette it can sit against. The sharpest angle tells you whether it belongs in a point or in the body. The area tells you roughly how much of the remaining gap it will consume.
This is faster than visual matching because the eye is bad at comparing rotated shapes. A trapezoid rotated 140 degrees genuinely does not look like the same trapezoid, which is why players so often skip over the correct piece several times before finding it. Edge lengths do not rotate.
A practical habit: when you pick up a piece, immediately rotate it until its longest edge is horizontal. Now compare it to the outline's horizontal edges. If none match, rotate the outline in your head rather than the piece — you will usually spot the target in a second or two.
Assembly order
Think of the silhouette as a boundary problem before it is an area problem. Pieces that touch the outer edge are constrained on at least one side; interior pieces are constrained on all sides but only once their neighbours exist. So work the perimeter first, then fill the middle.
Within the perimeter, go from the most distinctive feature outwards. A heart's point, an arrow's head, a house's roof apex — these are single-candidate positions. Placing one gives you a fixed reference for scale and rotation that every subsequent judgement can lean on.
Save the small symmetric wedges for last on purpose. They are the pieces with the most valid-looking wrong homes, and by the time everything else is down there is only one gap left for each of them.
Common mistakes
Filling the middle first. It feels productive because pieces go down quickly, but interior placements are unconstrained until the perimeter exists, so half of them turn out to be a few degrees off and have to be redone.
Fighting the snap. If a piece refuses to lock after two attempts, it is in the wrong gap — the tolerance is wide enough that a correct placement almost never needs a third try.
Dragging with the fingertip covering the piece. On phones, drag from the edge of a fragment rather than its centre so you can see the live highlight telling you where it will land.
Quick reference
Place the biggest piece first — Large fragments have the fewest possible homes. Every one you place removes ambiguity from all the smaller ones.
Match edges, not shapes — Look at the length of the longest edge on a piece and find the outline edge of the same length. That pairing decides the rotation for you.
Work the corners inwards — Sharp corners of the silhouette can only be filled by a piece with a matching sharp angle, and there are usually only one or two of those.
Rotate before you drag — Getting the angle roughly right in the tray area means the snap tolerance does the rest when you drop it. Dragging first and rotating on the board fights the placement.
Use the nudge arrows for the last millimetre — If a piece is clearly in the right gap but will not lock, it is usually a couple of degrees off. Nudge rather than re-drag.
Frequently asked questions
- Is there a new shape every day?
- Yes. Both the silhouette and the way it is sliced are generated from the day's seed, so everyone gets the same puzzle and it never repeats the previous day's cut.
- How precise do I have to be?
- Not very. Pieces snap when they are within a few degrees and a short distance of their home, and there is a live highlight showing where the current piece would land.
- Are the touch controls different from mouse controls?
- The touch build lifts the dragged piece above your fingertip so it stays visible, adds two-finger twist, and enlarges the control buttons. The logic is identical.
- Is there a hint system?
- No. The outline itself is the hint, and the snap tolerance is generous enough that patient trial and error always gets there.