I think it does sort of imply causation, but it's quite important to remember it can imply that a different variable is causing the correlation between the variables you're observing.
There was an interesting article in
Quorum magazine a couple of years ago describing the work of a
medical researcher who was using large amounts of "dirty" data to mine it for possible fruitful
avenues of research into new drugs and treatments. Most medical researchers want "very clean" data
(free of multiple factors like smoking, alcohol consumption, diet etc), but he has developed
statistical techniques to compensate for the fuzziness of the data.
One nice discovery was that there was a correlation between use of statins and the survival rate of
patients who had received certain organ transplants. It would not have been discovered anywhere as
quickly, if at all, if he had waited to get data where multiple factors were accounted for.
(Apparently, the best medical data in Europe at that time came from Belgium IIRC, but there was not
large amounts of it, and it took a long time to prepare, i.e "clean").
Of course, using "dirty data" requires very careful analysis. Matters like the "95% levels of
confidence" as measures of significance still plague some research fields. There are many who continue
to believe it automatically and magically provides statistical safety (and attracts grants and investors),
despite its complete arbitrariness.
