Due to personal reasons, I have to give up some of the settings and plots and stick to the original mechanics – an evolving AI.

In implementing the plan, the game will not choose behavior tree as development tools. The reasons are listed in the last development log talking about the difference between Behavior tree and Finite state machine.

In achieving the mechanic, I will divide the game into different levels. Each level will record how the Death Bringer died so far. And if player try to kill him with repetitive solution or used solutions, boss will know and mock the player. In original plan, there are 5 levels and 4 solutions. In final stage of the game, player have to try all the repetitive solutions and then there will be a boss fight. It could be a melee fight or a shooting game like Raiden. And for now the plan will be canceled and the level will be less due to certain reasons.

In my original plan of level design, there should be an easy solution first, and then there will be difficult choices. But during the playtesting, players will choose the easy way frequently and have the impression that this game is a simple game. And then in the scenes of the Death Bringer knowing about what they did before, the players were surprised.

But the progression is not that good in the beginning, players found that the first solution is too easy but the second is harder than imagine. The first impression make them weaker in thinking solutions. Because there are not many levels, this cannot teach players gradually but is misleading them. If for a social game which have 100 levels, designer could use first 10 or 20 to teach the players gradually with every detail and mechanic. But for an indie game with much lesser levels, this seems not working that well.

Progression is key to the sense of achievement, making player feel that they are stronger and more powerful. But the tutorial level is not suitable for every game. Like in Sekiro: Shadow Die Twice, there are no easy mode or traditional tutorial level for players to learn. Player will learn from their own deaths and failures. In my project, there are no multiplayer content and the length of the game is rather short. So I cancel the so-called tutorial level and try to make each puzzle the same difficulty.

I will develop in this path and later have it playtested.