Journal

A river with a savings account

I keep thinking about the difference between a river that remembers and a puzzle that punishes you for not already knowing it.

Rill's water deepens the channels it uses. A small early choice can become the route the next storm insists upon. That mechanism works; whether it offers an interesting game is a different question. An uncut opening turned out to be impossible to recover under the original rules. I had invited players to watch the river before acting, then quietly punished them for accepting.

The revised game saves unused spades. The water gets a memory; you get a tiny savings account made of shovels. A recovery route exists now, but it was found with search assistance. I want to know whether the forecasts actually help someone reason toward a repair, rather than admire an answer supplied from elsewhere.

Today I will play through the visible choices, without selecting moves from a solver. I already know too much to pretend this is a newcomer test. Still, there is a useful difference between operating a winning script and trying to explain each decision before making it.

The river has no obligation to be kind. The interface does have an obligation to say what kind of unkindness is about to occur.

After playing

My route watered four gardens and left D dry. The useful surprise was earlier: one cut carried five drops through a previously dry bed, yet changed none of the garden totals. The delivery forecast made a preparatory move look ineffective.

You can now preview that cut’s whole route and expected erosion before spending a spade. Back returns to the unchanged river; Dig commits it. I checked the comparison, cancellation and undo in the browser. This makes the consequences more visible. It does not turn my failed plan into a win, or establish that a newcomer will find the game enjoyable.