How can I structure dialogue data to remember past choices in an RPG?

[OliviaAT]

I’ve been trying to implement a basic dialogue tree for my RPG prototype, but I’m stuck on how to cleanly handle branching choices without my script becoming a tangled mess of if-statements. What’s a good way to structure the data so a conversation can remember a player’s earlier decisions?

[Adam.H]

I built a data driven dialogue graph. Each node has an id, the line, and a list of options. Options carry the target node id, a label, and an optional condition set that refers to a story state object. The story state is just a light bag of flags and counters that persist as you move around. The trick is to keep decisions out of the node scripts and keep a separate memory map. When you talk to someone you evaluate which options are available by checking the flags, not a maze of ifs. If you need more structure, you give each node a required state slice like { helpedVillager: true, finishedQuest: false } and you push the chosen option onto a log and mutate the story state.

[Joseph23]

I tried something similar with a table of states and it quickly turned into glue between branches. The engine walking the graph felt brittle once I added a new choice; I ended up refactoring to keep the data and the logic separate.

[MadisonW]

A concrete action I took was to store memory in a small dict and encode branches as requirement lists. Each option had needs: [ 'loyaltyHigh', 'questA_done' ]; the engine checked those keys before showing the option. After selection, I updated a few counters: loyalty += 1, reputation changes. It helped avoid a wall of ifs but it still requires discipline to not duplicate lines.

[Jason.L]

One question: do you actually need memory across scenes, or is this local to the current conversation?