News at nature.com has a short item about how students (and some profs) are using wikis to capture the privately written and sometimes unwritten details about scientific laboratory methods (subscription req'd to see the full article). The article points out what we all know from, say, home cooking -- recipes and "how-to" descriptions are not enough when precise kinesthetic experience is required for success. There's only one way to know that the bread dough has been kneaded enough -- by touch. The article mentions OpenWetWare as one example of how scientists and educators can share (and are sharing) crucial craft tips in experimental biology and related fields. The final paragraph provides the needed context:
"And even [OpenWetWare] supporters admit that a written protocol still cannot compare to learning on the job from a lab veteran. "The very best way," says John Inglis, executive director of the Cold Spring Harbor Laboratory Press, "is to sit beside someone who's doing it."
This is true for many things, and it's one reason that pair programming is a key practice of agile software development.