It's good that these things are started to be studied with real data. However, it's way too soon to be drawing many conclusions.
1. As pointed out, they only measured a very small portion of public torrent traffic. They did not measure Demonoid and other sources.
2. They compared sales of certain ebooks after they were offered for free and after appearing on P2P.
3. They have a pretty small sample set: 16 titles, only 8 on P2P.
4. They showed that for the average ebook, sales increased modestly after it was offered for free, and after it was available on bittorrent, and the change in sales did not correlate with how popular a book was prior to being available for free. We're talking average increases of 6.5%. But the range varied on the P2P titles from 18.2% up to 33.1% down.
5. With such a small statistical sample from only one specific genre, it's hard to draw any broad conclusions. It does seem that offering certain titles for free as a promotion for a limited time does help sales in the short term, but the long-term effect of piracy on sales (which such free promotions might facilitate since they provide a package of easily redistributable files without DRM) on those and other books remains unknown.
6. They provide the same cautions as I have when studying this issue: correlation is not cuasation, larger samples may uncover a trend that's not yet visible, and what works today with ebooks may not be true tomorrow due to so many other changing variables.