arlequim project pt.2: resuming the project

I'm posting again on the blog to give a brief update about a project I finally resumed after about a year. I restarted the procedural interactive fiction project, that for now I'm still calling Arlequim (Harlequin in portuguese), since it's still a prototype.

I decided to start the code from scratch, since I improved a lot as a C# coder in the mean time and also took some better decisions regarding pipeline and planning. Instead of immediatly starting the visual interface, even with temporary graphics, I'm still designing the scenes independently of each other in an editor window that allows me to simulate the whole scene. One of the great advantages of the game structure being segmented in generic scenes with flexible roles is that it becomes easy to test the arc and the internal dynamics of a scene in a self contained manner, without having to play or simulate big portions of the game.

My objective for the rest of the month is either developing the visual interface for testing the interaction with the scene, dragging the characters inside and out of it, or investing more time on designing scene models, developing a bit of a bigger set and making their designs more interesting before making a playable visual prototype of the system. As I mentioned before, it's possible to test if the internal dynamic of a scene is working just by the simulation on the editor window.

At the moment I'm iterating on the design of a verbal argument scene between two parties of characters, with one possible neutral side trying to mediate the situation, attempting to get the weights and the algorithm for act selection right, to achieve an interesting arc.

The scene simulation window with argument scene and the current act that is happening, there's already a sliver of functional text generation to express the events.

 

As I mentioned in the previous post, months ago (I think I did), the chance of an act being drawn and expressed depends on the aspects that make up the current scene's context, an act of conflict has a better chance of happening if the scene has a big degree of conflict, this act then increases the degree of conflict in the scene leading to this feedback loop. I'm now trying to find the weights and functions that will create an explicit arc, like a scene becoming increasingly more conflictual, while not preventing many different acts from occuring.

The ideal is to balance the identity and legibility of a scene with a reasonable swat of possibilities. What I mean by that is, during a scene I want the player to be able to see that it is threading certain specific paths, like conflict or introspection, for it to be easy to make a reading of the situation and for the scene to gain a specific identity among its other possibilities (with different characters, for instance), but I also don't want for a single path to dominate the whole scene.

If all the acts are conflict, it's obvious where the scene is going and there is no tension, being too boring for the scene to happen in such a predictable way. The ideal would be to have two or three possible threads, so the player can have  expectation of how the scene ends. 

To test this I started to work on an automatic testing process that runs the whole scene at once and emmits me a report of what role was assigned to which character, what was the sequence of acts, with what aspects the scene ended with, how many acts it took to finish, what was the ending and what changes on the world state happened throughout.

My goal here while testing the scenes is for them to consistently end with a few dominant aspects, giving identity to each scene instance, with a series of acts that correspond to this aspects, especially the one that ends the scene, the most salient one. It doesn't matter if the aspects grow in an interesting way if it's not possible to notice this having na impact on the sequence of acts.

This is all too abstract yet, I'm gonna write another post through this month detailing how I develop the model of a scene, how the game instances it and how I'm planning to design each one, to try and give a good little dynamic to each of them. I wish I had written more, but I'm really tired at the moment.

For now I just wanted to tell that I resumed the project and what part of it I'm developing right now.