Forgive the mindless ramblings (mine, that is) in this thread and the one on fanaticism.  I was reading about science and scientific method the other day and about Sir Robert Boyle's literary technology.  In the 17th century, Boyle was a strong proponent of using scientific instruments in experimentation.  No-brainer today, perhaps, but he ran into considerable opposition in his time.  At any rate, in addition to using an air-pump to conduct experiments, Boyle also crafted a method of presenting scientific experiments on the page in order to allow for a kind of "virtual witnessing" of the experiment.  No internet, no blogs, and no air travel to get natural philosophers (no "scientist" yet) to non-existent professional conferences to share their findings with their colleagues.  So they needed to report their experiments in a way that allowed for others to read and follow their methodologies.  Boyle also advocated that natural philosophers should share their failed experiments as a means of proving their trustworthiness.

And this is where I'm going with this: this final point about sharing failed experiments is not really publicly practiced in the scientific community these days.  Scholarly journals publish research findings, but rarely failed experiments and researchers looking for further funding rarely (and justifiably) feel inclined to share their blunders.  Anyway, I was also leafing through a recent issue of WIRED magazine, which contained an article on how we should be publishing failed experiments as a part of the scientific process, and I got to thinking about chess and opening theory.  It seems to stand out as a prime example of precisely that process.  Databases collect all games, and we see the empirical examination of theory.  Further, the databases, Informant and other sources don't just publish good opening innovations; we have access to the mistakes and problems.  So, maybe I don't have a question--other than: are there other viable corollaries--just an observation that I've been mulling over.  Do mistakes improve our broader understanding of something or enhance our production of knowledge?