Personal responsibility is not a policy that can’t be applied across a population. You might have a combination of will, mental health state, and life advantages that puts you in the the 80 or 90th percentile of being able to resist or avoid addictive activities but you are not common.
Willpower is finite and companies have thousands of lifetimes worth of resources.
I don't believe personal responsibility and willpower is a fixed quantity, even at the population level. In fact, I think they are negatively impacted by emphasizing the influence of external factors in personal behavior.
Telling an addict that they are powerless to regulate their consumption only erodes their ability to improve. External factors my have led them to their addiction, but they can't fix an additional without assuming agency taking control of their behavior. Yes, this is HARD, but no addict can recover without it.
The only durable solution that will help people is shifting the curve such that more people are able to resist or avoid addictive activities. Policy solutions are simply too ineffective and porous to have a durable effect.
Social media platforms are giant algorithmic feedback engines creating the most reliable way for their regular users to fulfill their neuro-chemical and social needs. That these people cannot find better information in their mediasphere means that we need to encourage people and remind them of their agency - yes - but we cannot leave it at "words of encouragement" when the other end is "massive highly personalized feedback loop of all the people you miss seeing"
I think much of the problem with duality dissolves once you eliminate the concept of the soul or free will. People tend to ignore the fact that the individual still exists as an information processing agent that takes actions, even if you remove free will.
If a program fails to provide the desired output when given an input, it is neither the input or that code that is to blame, but an incompatibility between the input and code that is at fault. You it is possible to get the desired outcome by modifying either, or more likely both.
I'm sure I am overlooking some implications or counterpoints, so I look forward to reading your suggestion.