Building 1D3P: When Smaller Is Too Good

The little snake can curl its four body units into a single cell and pass through spaces it cannot reach while stretched out.

That sounds like a useful ability for a puzzle game. In the earliest prototype of this mechanic, it was too useful.

Once the material curled up, there was usually little reason to unfold it again until the goal required the four-cell form. There were two forms in the rules, but much of the time only one was worth using.

The problem was not too few levels. The two forms did not yet offer enough of a trade-off.

This was one of the design problems behind Homeward, Little Snake, the first game from 1D3P. 1D3P is my independent browser-game project: not a growing pile of minigames, but small, complete puzzles built around clear ideas.

Making Both Forms Useful

The mechanic began as a model of four units of material, not the snake it looks like today.

Expanded, it occupied four consecutive cells. Compact, it occupied one. With ordinary walls that only checked occupied cells, the smaller form had a clear movement advantage: where the strip fitted, the single cell fitted too. Narrow spaces that admitted the single cell might block the strip.

Different wall layouts could produce different maps while leaving the reasoning much the same:

Find a curl mechanism, cross the maze, find an unfold mechanism, then enter the goal.

To change that, the expanded form needed a reason to exist beyond filling the goal.

The prototype added low-clearance cells. The expanded form could pass through them; the compact form could not.

Each form now had places the other could not go:

• Narrow spaces called for curling up.
• Low-clearance spaces called for stretching out.

But this was not enough. The second experiment’s notes still called out a pattern of curling at narrow passages and unfolding at low-clearance passages.

There were more rules, but the player could still choose a form just by reading the symbol directly ahead.

The next change needed to affect more than the spaces each form could enter. Choosing a form needed to change the options available along the rest of the route.

Unfolding Is a Choice, Not a Reset

The next prototype replaced fixed unfolding with a choice of four directions.

Movement here is different from classic Snake. The expanded body slides as one four-cell strip. An ordinary move does not bend it or change which way it extends. To choose a different unfolding direction, it must curl up first, then reach an unfold mechanism.

The end that stays fixed during unfolding is the anchor. A mechanism triggers when the anchor reaches it; covering it with another part of the body does not trigger it.

These rules give the same position different futures. A direction that fits now may not suit the move that comes next.

Level 10, Anchor Relay, contains a useful example.

The compact snake is on the unfold mechanism at C3. Its goal is to cover all four goal cells in column F. Here is the same position before unfolding, after unfolding right, and after unfolding down:

Orange marks the body; @ marks its anchor. C is a curl mechanism, S an unfold mechanism, G a goal cell, L a low-clearance cell, and # a wall. Mechanisms and goals covered by the body remain in place.

Right and down are both legal unfolding choices. Neither moves the anchor from C3; they change where the rest of the body extends.

Right looks direct. The body already reaches the goal column and covers one of its goal cells.

But moving the anchor right onto the curl mechanism at D3 would require the whole body to occupy D3, E3, F3, and G3. G3 is outside the board.

That move is blocked.

The game checks whether the whole body can move legally before triggering a mechanism. The curl mechanism at D3 cannot rescue this move: the body would leave the board before the anchor could legally reach it.

Unfolding down leaves a five-move route:

1. Unfold down, occupying C3 through C6.
2. Move right to D3. The whole strip fits, and the curl mechanism then compacts it.
3. Move right to E3, reaching the unfold mechanism.
4. Unfold down again, occupying E3 through E6.
5. Move right to F3, covering every goal cell and completing the level.

Unfolding right is not a dead end. The snake can return left to A3, curl up, come back to C3, and choose a different unfolding direction.

Starting from the position in the diagram, the shortest routes to completion without undo are:

The player can also undo the choice. These counts assume they continue without doing so.

The difference appears after unfolding:

Both directions fit now, but they leave different options for the next move.

Stretching toward the goal does not necessarily mean fewer moves to finish.

More Rules Do Not Automatically Mean More Choices

At C3, unfolding direction is more than a visual change. It determines whether the snake can reach the next mechanism directly or must first take a detour.

The early experiments were asking increasingly specific questions.

Initially, the question was whether changing form could support a basic gameplay loop.

With low-clearance cells, it became whether each form had a distinct use.

With four-direction unfolding, the next question was how a choice affected which cells the body could occupy, where the anchor could move, and when the snake could change form.

These are different questions.

If one form is consistently more useful, adding maps may not solve the problem. If a player can always decide by reading the symbol immediately ahead, adding mechanisms may not deepen the reasoning either.

Sometimes the thing worth revising is not the amount of content, but the relationship between the options already there.

Where My Judgment Comes In

AI handles the hands-on production of the game; I review the work and steer its direction.

The questions here go beyond whether a level has a solution. They also concern whether different actions in the same position have different consequences. Those questions need concrete board states, not just working code or a solvable level.

For this first game, the story is not how many tools were used. It is how a simple mechanic was revised to make room for different routes and costs.

You can play Homeward, Little Snake in your browser.

The next time you reach an unfold mechanism, look a few moves ahead:

Does this direction merely fit now, or does it also leave room for the next transformation?