I don't think this approach necessarily helps with that distinction of good breaks and bad procrastination. I think it would be interesting to distinguish those, but I would look for another method if that were my goal. That also isn't to disregard the interest of the experiment above, just to say it doesn't help here.
