Science has been for hundreds of years the most open and transparent institution. But I think that recently it has appeared a new contender that is even most transparent, open software development. I think science could use the same approach for research, from step zero.
One scientist has an idea, he publishes his hypothesis and intended metodology. Others can jump in and tell him, for example, that the hypothesis has been proved wrong in a recent paper, or suggest improvements in his methodology. Others can chip in and offer to replicate the experiment to increase the sample size. Mathematicians could observe and correct the statistical analysis before the conclusion are published. I think that will improve the quality of the results.
Of course, the bigger problem is that most papers would have dozens or hundreds of authors, diluting the individual contribution of each one. On the other hand, finished reasearch already would be peer reviewed and corrected.
The parallels are greater than even those you described. I'm helping a startup focused on bringing out an integrated research environment to help fix many of the issues presented in the comments here and the article. Happy to chat with anyone who would like to see a demo. Email Lane (at) MyIRE dot com.
You're definitely on to something; the parallels between science and open software are tantalizingly close.
From my experience a main problem is real estate; you can't do most science experiments without a proper lab. We could probably brainstorm some creative solutions (like operating an "open" contract research organization to be the hands for everyone's ideas), but at the moment this is too big of a barrier.
With real estate comes cost, then funding, then jockeying for funding, and finally journal impact factors as a measurement of worthiness.
It's also why one can't make a science-related "hackathon" without changing the concept.
One scientist has an idea, he publishes his hypothesis and intended metodology. Others can jump in and tell him, for example, that the hypothesis has been proved wrong in a recent paper, or suggest improvements in his methodology. Others can chip in and offer to replicate the experiment to increase the sample size. Mathematicians could observe and correct the statistical analysis before the conclusion are published. I think that will improve the quality of the results.
Of course, the bigger problem is that most papers would have dozens or hundreds of authors, diluting the individual contribution of each one. On the other hand, finished reasearch already would be peer reviewed and corrected.