CapnZapp's characterization paint a negative light, but I don't think the points themselves are inaccurate. It would honestly be a red flag if Paizo's development team didn't have design discussions & internal rulings on these issues.
Going back to the statement you quoted, we can examine what it means for a system to be "easier to learn." This kind of phrasing is not really meaningful. I could make a statement that the goal was for the system to be "more fun," but it should be obvious that "fun" varies from player to player. Similarly, "easier to learn" is actually carrying a lot of weight.
For someone who is logically & mathematically minded there is nothing inherently complex or overwhelming about 3E/PF1 rulesets (The complexity comes from sheer quantity of rules and edge-case interactions). You can generally ascertain how effective your character at a task is by the number of bonuses you get to certain rolls. However, if you put this type of person into a FATE game, they will probably struggle to conceptualize how aspects of their character might translate into chance of "success" or "failure."
PF2 largely removed the ability to optimize along the most-easily-communicated metric, the raw numerical axis. This by itself isn't bad, but the game retains the complexity that was ultimately the source of the problem. Players have more decision points than ever. The quantity and edge-case interactions remain. PF2 demands optimization moreso than ever, with a high default-difficulty. It's just that this optimization eschews the obvious d20 +Modifier and is in harder-to-quantify aspects such as breadth of options, action efficiency, and other avenues that require deeper understanding of the system(s) than simply maximizing a number.