> look which experiment shows inteference and which doesn't

This is also a kind of measurement, and one you can't meaningfully perform before the other. By the time you know the interference pattern, it's already too late to decide whether you want to measure the intermediate position.

You might try to measure without looking at the results, but that doesn't change anything about the fact that the interaction happened. (Superpositions do not collapse by being observed by a human, they collapse by interacting with the rest of the world.)