This website offers an interesting new approach of publishing not stories but single observations:
ScienceMatters
Might reduce the "entry threshold" and opens a way to publish single observations that would never get into a paper finally.
An article about it in Science and Labtimes.
And they offer also an anonymous peer review for both sides, which is a good idea, too.
For me it's a great idea and hopefully it will establish itself. What do think of it?