The journal editors are apparently targeting three types of difficult to publish data: experiments that fail, experiments that are incomplete or inconclusive, and experiments that disagree with current mainstream scientific understanding. I am not sure I really see a value here, though maybe I am just getting old and crotchety.
Experiments that fail. When I was a student, I thought publishing failed experiments would be a very valuable resource. With more scientific experience, I realize that many groups will wonder whether it was the experiment or the execution that failed, especially for negative (or inconclusive) results that go against scientific intuition or prior experience on related systems. Publishing this sort of failed experiment MAY help, but may not.
Experiments that are incomplete or inconclusive. Personally, I don't think incomplete or inconclusive experiments really should be published. Yes, there is lots of data languishing in notebooks, but that is at least in part due to lack of motivation and/or time to write the (probably minor impact) results up. Incomplete experiments can be misleading, and inconclusive experiments either need rethinking, better design, better instrumentation, or more data. Both of these situations are likely caused by funding running out or someone who was working on the problem leaving the lab. This leads straight back into lack of time/motivation to write.
Experiments that disagree with current mainstream scientific understanding. I am not sure we need a new journal for this. Shouldn't current journals be eager to publish results that conflict with current understanding (and therefore advance science)? I know that truly new, truly innovative stuff sometimes has a hard time getting published, but still extraordinary claims should require extraordinary evidence, shouldn't they?
I totally agree that the literature is biased towards positive results. But many negative results are currently available (and more and more frequently searchable) in Masters and PhD theses. It will be a real shame when dissertations become introductory and concluding chapters wrapped around published papers (as is happening more and more). So maybe we do need something new? With the barrier to Web publication low, I certainly admire the editors for giving it a try!
Now the All Results journals may end up being an excellent resource for scientists moving into new areas or pushing the edges, but I have my doubts. The cynic in me thinks that it will mostly be a CV-padding tool for future academics.