I dunno, even though the authors address its use, making the task Tower of Hanoi doesn't meet the excitement of the title.

Especially since it's a recursive problem so each step is naturally broken up into subtasks. And the algorithm of what subtasks to break it up in to is public. This makes it much easier for it to get down to a case that the LLM can reliable solve.

And yet the reverse paper was posted ad nauseam, covered by every news slop site, and overblown with really negative takes.