I think the focus on short-form video (or short-form content in general) misses the point. I don't know that short-form media is inherently worse than long-form, but it does lend itself more readily to algorithmic manipulation. Essentially, the shorter the time between starting a piece of media and starting the next piece (and the more quickly users upvote / like / heart), the more finely-grained the data that gets fed into the preference engine. It's simply more profitable to ~~make~~ distribute content that's as simple and atomic as possible.