Our systems (even on relatively longread-friendly platforms like Medium) disincentivize nuance. They don’t need to.
Consider Medium’s ranking process, by which it sorts content for the dashboard/feed, recommendations, and various lists. While I’m unaware of a writeup of how it actually works, it’s clear that ranking is not based purely on recency. Instead, it appears to be based upon a combination of recency & interaction (recommendations and comments) weighed by social network distance. A minor tweak that would increase the weight of full reads (and increase that weight based on estimated read time, so that an article that takes an hour to read that got five reads by people you’re following will be ranked higher than an article that takes five minutes to read but got twenty-five reads) could make short clickbait articles mostly disappear from our feeds. Instead, recommendations clearly count for much more than reads or length.