In theory. There are many complex and subtle issues that must be thoroughly analyzed before passing judgment on this paper. For example, what if other miners, instead of joining the selfish pool, decide to create their own competing selfish pools? Also, as nullc points out elsewhere on this thread[1] and the Bitcoin Forum[2], peering between miners is already so extensive that temporarily keeping blocks secret could be much costlier than assumed by the authors of the paper.
The Bitcoin community must go through this paper in detail, but IF the authors' logic and math prove correct, this could be a real vulnerability.
--
[1] https://news.ycombinator.com/item?id=6669735
[2] https://bitcointalk.org/index.php?topic=324413.msg3476697#ms...
--
Edits: moved paragraphs around and added sentence about and footnotes pointing to nullc's thoughts.