As explained by Jean-Claude Bradley in this excellent interview: The Pursuit of Automation: Open Notebook Science:
The difference between Bradley’s idea and traditional open source projects is the clutter, i.e., all of the data collected from research. In a journal article, what the reader receives is a set of findings, published after peer review, along with the conclusions and basic data used to reach them. Bradley’s project allows the reader to see the process unfold, from hypothesis to conclusion, with all data, experiments and notes that are collected along the way. This inclusiveness allows other researchers to better assess their own ideas, because not only do they have the conclusion and data used to support it, but also they have access to every failed or incomplete subset of data to use in their equations.