Canaries In A Data Mine

from the I-wanna-new-drug dept

One of the toughest things to deal with in science is that sometimes experiments don't work out the way they're supposed to. (And it usually happens around the 3rd year of grad school.) But short of faking results, scientists have to make the best of the data that is collected. Though there are plenty of PhD theses that are filled with experiments that didn't quite work out, drug companies are under a bit more pressure to produce products that work. So when their drug trials fail, they need to squeeze anything they can out of the research -- which is turning into analyzing the data for correlations and correlated correlations. But that's not the interesting part: science is supposed to look for these sort of clues to new discoveries. The interesting part is that data is being collected in ways that make it easier to do data mining for unexpected trends. (Perhaps some of the data mining could even be opened up to the public?) Some might suggest that scientists are "giving up" and submitting to combinatorial discovery, but whatever works, right?