This is why I’d rather just give a review without a star rating, to be honest. An explanation of someone’s thoughts on the game, its specific joys and sticking points, and why the reviewer did/didn’t like it is more useful to me than a numerical quantifier, especially when everyone’s working off a slightly different subjective scale. I don’t want to look at a rating and wonder “is this reflecting the work’s execution or the rater’s personal preference?” I definitely don’t want to consign someone to algorithmic irrelevance based on my own preferences. (This is also why I would just not play the games in scenario 4. I’m clearly not the target audience, and that’s fine! I don’t have to be!)
In general, I only rate games in the context of actual comp judging, and may or may not get around to transferring those ratings to IFDB. I might give a rating to something that really stands out as worthy of attention (or avoidance). I would like to get more comfortable with reviewing, though.