This video and my own experiences in MC:PE have shown that, when there's a block in front of an observer and the observer detects a block update, the block in front of the observer's output doesn't activate anything but pistons and sticky pistons. However, as of 16w41a, observers don't strongly power anything, rather they only activate blocks directly in front of them.

If this inconsistency, like all the other ones, is resolved as favoring MC:PE's observers over the Java ones, then blocks in front of observers would once again be able to activate pistons and sticky pistons but the behavior wouldn't apply to any other redstone components.

This change would allow for much more potential with observers than what we have currently, and in my opinion, it would be a simple and sensible change. Please consider this when you decide on how this will be resolved.