Everything is fine, give it some time.
Here's a formula you can try :
Title : "I made a AI powered insert tool use-case"
Article: "Hey look what I did ! By hooking up insert LLM model to this insert random tool in a few hours of vibe coding I got it to do something that vaguely resembles something useful and made no effort to evaluate performance/compare to anything. Here's my toy demo example."
Should be front page material, going on the last few months.
The comparison of best-of-the best is not that useful in practice because you have to filter through so much noise to find it, so the average case is more useful in practice.
That is the random peak that results from lucky combinations in some corners of a space - it is not owing to how the channel works, it is not owing to the culture it promotes.