At some point, marketing collectively got very good at gaming systems and very bad at talking to people. And the worst part is it happened so gradually that most teams didn't notice when the shift occurred.
You can see it everywhere now. Blog posts structured entirely around what ranks rather than what's useful. Hooks written specifically to spike the retention metric in the first three seconds. Video formats chosen not because they serve the idea but because the platform is actively pushing that format this month and the reach is good. Captions reverse-engineered from what performed last week. All of it optimised for what the algorithm rewards, barely any of it optimised for what a real person actually wants to encounter.
And it works, in a narrow sense, for a while. The numbers look okay. The reach is decent. But then the algorithm updates and everything built around its preferences has to be rebuilt from scratch, because none of it was built on anything real. The audience didn't actually connect with it. The algorithm just happened to be surfacing it.
The Deeper Problem
The deeper problem is that this approach trains you to think about your audience as a secondary concern. You're writing for a system and hoping the people it delivers you to will also care. That's backwards. The system, at least in principle, is trying to show people things they genuinely want to see. So the most sustainable way to do well algorithmically is to make something a real person actually wants to read, watch, or save. But that's a harder brief than "hook in the first three seconds" so the shortcut is tempting.
The content that outlasts algorithm changes, the stuff people bookmark, come back to, share months after it was posted, was almost never made with platform mechanics as the primary brief. It was made for a person. Specific, real, with actual things on their mind, in an actual moment in their life or work. The algorithm caught up because humans caught up first. People engaged with it genuinely, and the system noticed.
What Actually Useful Platform Knowledge Looks Like
There's a version of knowing the algorithm that's actually useful: understanding the formats, the timing, the distribution mechanics of each platform. That knowledge is worth having. But it should be in service of getting good work in front of the right people, not a substitute for making good work in the first place.
Write for the person. The algorithm is just trying to find them for you.