[logicallee]

Your comment is extremely bizarre. The paper seems exceptionally clear and well-written! It's rather short. It gives the necessary background and states its assumptions, it is very easy reading.

Can you quote a paragraph you didn't like? Is your main issue that the data descriptor is defined elsewhere (as opposed to a date range)? I don't think the descriptor itself is emotional at all.

By the way, I thought based on the abstract that the effect would be rather large. For anyone who downloads the PDF, go to the graphs - the effect is actually a very small difference even on search terms rated to get people in trouble with their governmetn, as well as on personal-privacy related search terms.

The effect is clearly there but rather small compared to what I was expecting.

Granted I only spent a few minutes with the PDF and could be misinterpreting.