We should measure our winning, somehow, and see whether reading LW increases it.

Sure, this answer just brings a new set of questions. Such as: what exactly should we measure? If we use something as an approximation, what if it becomes a lost purpose? If we change our method of measuring later, what if we are just rationalizing conveniently? (We can create an illusion of infinite growth just by measuring two complementary values X and Y, always focusing on the one which grows at the given moment.)

I would say that a person reading LW for longer time should be able to list specific improvements in their life. Improvements visible from outside; that is, what they do differently, not how they think or speak differently. That is the difference from the outside. If there is no such improvement, that would suggest it is time to stop reading; or at least stop reading the general discussion, and focus on stuff like Group Rationality Diary.

(My personal excuse is that reading LW reduces the time spent reading other websites. Debates on other websites suddenly feel silly. And the improvement is that reading other websites often made me angry, but reading LW does not mess with my emotions. -- I wish I could say something better, but even this is better than nothing. Of course it does not explain why reading LW would be better than abstaining from internet. Except that abstaining from internet seems unlikely; if I stopped reading LW, I would probably return to the websites I used to read previously.)