Something I've often looked at is the nature of a scene and how different lengths of scenes and passages affect the fluidity and reading of a work. I'll admit I draw a lot of influence from Randy Ingermanson's method of scene structure. To save people from having to parse that whole page, I'll summarize his basic ideas:
Ingermanson says that scenes have a large- and small-scale structure. On a large scale, a scene requires that a character in the scene have a goal, that they face conflict in trying to reach that goal, and that ultimately, some sort of disaster occur to ultimately make that goal moot---whether it becomes impossible and requires a new goal be achieved first, whether it be accomplished and allow for the next goal to ensue. Whatever the reason, the disaster kicks off a "sequel" to the scene, in which the character experiences an emotional reaction, faces a dilemma about what courses of action to take, and finally decides on the next goal.
I tried this structure for a time in a strict sort of way, plotting out every scene and sequel scene with the three fundamental components. Ultimately, I found it very slow doing. One of the conceptual problems I had was that I couldn't really figure out what scale these concepts work on. It's easy to see how this works for, say, a platoon of soldiers trying to take an enemy position. It's harder to think how that may apply as cleanly or easily to a romance or another type of story.
That said, I definitely ascribe to the idea that characters have goals and that the turning points in stories occur when those goals change: they're accomplished, failed, or rendered moot. Those are the fates of goals, and it's not bad writing to suggest that most goals fail because then, the character must replace it with another one or given in to defeat. Goals need not be material; they can be subjective. Escaping pain can be a goal---it's just not one that a character is likely to be conscious of sometimes.
It's also important to realize that the driving goal of a scene may not be the POV character's goals. The opening scene of "Before and After" occurs with Gendo questioning Rei about her behavior. It's Gendo who wants to understand something. It's Gendo who faces conflict in Rei's inability or even refusal to understand. But viewed oppositely, maybe you can say it's Rei who's trying to say the right thing, to say what Gendo wants her to say, and that the rest of the chapter is her further effort to comprehend what Gendo meant by questioning her---and whether he's right.
I speak of "Before and After" in particular because it represents a bit of a contrast for me compared to third-person pieces like Identity or Echoes. One of the things I decided to do early on in "Before and After" was to avoid scene breaks at all, instead using narration to smooth the gaps. It's not that I don't think scene breaks are bad for first-person; I just considered it as much of a challenge as something I thought Rei would do. It's long been something I wanted to work on: the fluidity between scenes. I've often thought that frequent scenes under a thousand words or so to be too choppy on a regular basis. I don't strictly monitor scene length anymore. Act 4.7 of Identity is a good example: there are scenes that change POV without breaks (which Ingermanson considered a big no-no), there are short scenes, there's a long battle scene and a short one. With 4.7 I really tried to introduce actions in a sequence that made sense, using the idea that X happens and this is how Ranma and company react. Hence, I jumped POV's a lot mid-scene because a break would separate the action from the reaction.
Point of view is another big topic. A while back I read "Simplicity," by Tender Falling Rain--on the review I left I complained about big paragraphs, not because big paragraphs are bad, but I felt it might help to start small and ease people into the story, but that's another topic for another day--and the author there did exactly what I was talking about earlier: they changed point of view rather abruptly, I thought. It's something I read a lot in fanfiction, often to accomplish one thing: to show two people's internal reactions to the same scene. As always, I won't say that this shouldn't be done (again, I've done it myself), but I often ask myself when I see this technique done: was it necessary? Are these reactions that absolutely can't be portrayed with a single camera, so to speak? Do we need to jump from set of eyes to set of eyes to really comprehend the scene? Don't get me wrong; I think the versatility of an omniscient POV is a great thing, but I also think the personal nature of a limited POV (and here I use the terms loosely--I think narrative mode can temporarily be either and overall be neither) is really powerful. When you're unwilling for whatever reason to do a first-person narrative, sticking with the thoughts and feelings of a single character through the completion of a scene can really hit home what they see and how they see things...or how they misunderstand things, how what they see isn't reality.
That's a big point where I must disagree with Ingermanson. He speaks later on about the small-scale structure of scenes, about "motivation-reaction units" or MRUs. To summarize, motivation is what characters outside the POV character do and say, and reaction is what the POV character feels, thinks, says, and does to react to them. Ingermanson says that the motivation is "necessarily objective," and this is a big thing where I've felt I must diverge. I like the idea that what people see and hear is subjective and colored by emotions. I deal a lot in the notions of hallucinations and visions. What we experience is subjective. When I describe a sunny day as bright and warm, it carries a different connotation from hot and blinding. But then, maybe blinding is a bad word, and I should simply say it's bright before I say that Paul shaded his eyes, squinting, and reached for a pair of shades. As always, we balance show versus tell in writing.
Perhaps I've rambled enough. What did I mean to say? That a scene is done when it's done? That you can use as many or as few breaks as you like and it's fine? That goals can be as lofty as world peace and as petty as buying a bag of coffee grounds? That POV can be as nose-in-the-dirt or watching from orbit as needed? Yeah, something like that. Writing's all about appropriate place and time. There are few hard and fast rules, just a lot of choices. What conveys the meaning stylistically? Do you want to call attention to a scene with a break or keep the flow going? Do you want people to feel immediately a character's heartache or joy, or do you need to cover a hundred years of backstory in a few short paragraphs? It's all about finding the right tool for the job, I guess, and it's best, as a rule, to not rule out any of them.
Most of the time, anyway.