would recommend adding more choice points throughout - even if they don't necessarily do a ton.
the fundamental underlying problem with renpy (and VNs in general, really) is that most people don't read in 1-3 sentence chunks - which is what the vast majority of renpy VNs force. it works fine when you're doing something like call and response, where every couple of vn 'pages' you have a response to make, but it is *really* tedious reading through a dozen or more pages of text in 3-line increments.
i can't speak for everyone, obviously, but when i get into segments like that i just end up skipping everything.
there's a half dozen or so segments in the game right now like this - and 'a dozen or more pages' is probably an understatement for most of those.
it is trite to say 'don't tell the reader, show them' - although it's still good advice. it's just one of those pieces of advice that's incredibly easy to give and incredibly hard to actually do. so, my concrete suggestion is to add more choice points where player input is needed. simply breaking up the vast exposition sequences by turning them into a hub and spoke conversation model (like when you talk to ginny at the bar or on the bed, asking a question, getting a response, and landing back at the same 'hub' of questions) would definitely be an improvement - although you need to be careful still to make the individual answers short enough that they don't recreate the same issue.
i also think you could move a lot of the exposition to other places, so it doesn't all get flung at the player in one chunk. for instance, one of the questions you can ask Ginny at the bar leads to a discussion about the cops being corrupt/oppressive (or something like that - as I say, it is very tedious reading long answers in 3-line chunks). You could *show* the player that, rather than telling them, with (as an example) random events that happen when travelling from place to place showing the cops extorting people or whatever. I could see this being problematic since it might require renders, and thus take render time away from other stuff - but I'd say it would be better even if the renders were just placeholder images, since at least then you're putting everything together right. No renders + de-chunked exposition is better than No renders + chunked exposition, imo.
