Research has found that 54% of all studies cannot be reproduced, due to missing software, data or methods, even for highly cited papers from high impact journals. So, why can we do? We must try harder, was recently proposed in Nature, and the speaker agrees.
The problem is with the incentive: publish or perish. Publishing is what matters, not the quality of the research or the reproducability. This leads to the fact that research, that is in fact a never ending cycle of refinement, now is made very finite, papers are published in a static way, like a pdf, and is separated from the research findings. Science claims are uncoupled from science.
This is why the guys from Sage Bionetworks started clearScience: leveraging open and RESTful apis to reassemble an entire research environment in order to reconstruct studies and even build upon them or fork it. Another good idea to facilitate open science. Read more about clearScience.