Woah, interesting idea! 
Agreed, I have long held the same belief.
Strongly disagree. And the examples you provide aren’t really applicable or at least comprehensive, IMO. If you’re referring to the “manual”, clearly you’re talking specifically about a how-to context. In that context then yes, shorter, to the point, but still accurate and helpful content is preferred. But that is just one of, well, 100s of possible contexts/subject areas. There are many that fully justify “deeper” content and in fact very much benefit from it vs. “shallow” content. Here on Meta I would say there should be a bias toward “shallow” in a way, but there is also benefit in deeper content at times, like for example this exact discussion! 
Short form content is ascendant because it is the natural extension of shortening the dopamine loop in content consumption. It’s a byproduct of human evolution and biology that platforms are taking advantage of. It has little to do with the “quality” of the content. It’s also closely tied to the combination of preference learning needs of platforms to drive their ad revenue, and the again shorter loop of testing understanding of preferences. This is a big part of why TikTok’s recommendation algorithm is generally better than many other platforms: they can test and eliminate bad recommendation options quickly, and focus in on what users really respond to. Longer content makes that loop longer.
Facebook could easily surface higher “quality” content, in my view. But they have little incentive to do so. Their growth is plateauing likely due to increased competition and their bad privacy practices.