PMO is NOT a solution

In terms of just logical stuff, I like to think about it in terms of reward hacking. Reward hacking is a concept borrowed from AI research. Basically, to get an AI to complete a task, you give it a score based on how well it completed the task, and it learns over time to increase the score. That score is called the reward function of the AI. If the AI ever has a reward function that doesn't accurately measure what you want it to do, then the AI will end up doing stuff that you don't want because it maximizes the reward function, or the AI's "score". When the AI does stuff that you don't want but still gets a really good score, it's called reward hacking.
This boat maximizing the score in a racing game is an example:


Think about your brain as similar to an AI. You have goals that you're trying to achieve. Think about pleasure/happiness as your "reward function", or the "score" that your brain gets when it does stuff. PMO doesn't help you achieve any of your actual goals. But PMO has a huge short-term "reward" for your brain in the form of pleasure. So what that means is that PMO is the human version of reward hacking. Your brain gets pleasure even though it is actually working against your goals.

The conclusion is that you can't trust your reward function → you can't act based on what brings short-term pleasure. So the best decision is to not act on PMO urges, since you already know that they work against you achieving your goals.