Another angle is that they might use it to scrape more content for internal use. That’s on top of the training data that the user interactions generate. Which I was your point, I believe.