It'll take a couple of runs to try and get as much 'exposure/experience testing' done as possible. Especially with so many potential outcomes on a run. My current hypothesis is that there is always a possibility of failure, certain factors/variables can tip the scale to improve the odds, and this also applies in the opposite direction. I've noticed that I may have highly reliable positive outcome for an early-game event/decision combination with next to no repercussions, but there seems to always be a lingering 'that worked...for now'. So the experience becomes a game of chicken of 1. Have I tipped the scales in my favour, or have I done something to worsen my odds? and 2. Am I just going to get whacked because this game is secretly a social experience to test me sanity? Having played another round or 2 since my og comment, whether this is intentional or not, I like that the risk of repetition makes the hostile entities feel more sentient and alive. The fact they're not outwardly aggressive the moment they see you, are they playing a game with me, are they just curious, and how significant or insignificant are my decisions that game over is inevitable.
Since this is further down in a comment thread, I will say this. The Bucket and Sing Song combination is so funny and one of the reason the Bucket is currently my No# 1 priority item amongst other genuine uses.