"This Simple Philosophical Puzzle Shows How Difficult It Is to Know Something"

"thinkers for thousands of years had more or less taken one definition for granted: Knowledge is “justified true belief.”...

The trick, Pritchard says, is first to notice that there are two distinct “master intuitions” about knowledge that seem to be two “faces” of a single intuition, but are not. These are the “anti-luck intuition” (your true belief, which Pritchard calls a “cognitive success,” can’t be lucky to be considered knowledge) and the “ability intuition” (your true belief has to be in some sense a product of your cognitive ability)...

So the way to have knowledge, Pritchard concludes, is have your relevant cognitive abilities produce a belief that’s not only true and creditable to your agency, but also safe. By “safe,” Pritchard means that your belief couldn’t have easily been false. Temp’s belief, for instance, is safe—there’s a hidden guy guaranteeing he’ll believe the correct temperature each time he checks"

http://nautil.us/blog/this-simple-philosophical-puzzle-shows-how-difficult-it-is-to-know-something

This's makes me think about techniques in biology research to "know" something - biological systems are super chaotic and so if you are trying to measure some behavior (say, how active a cell's metabolism is depending on the type of sugar you put in the cell media; or how long you will be able to tolerate the air freshener your roommate bought) at least some of how that behavior change is going to be purely due to randomness. 

So to "know" that A causes B, you have to be able to measure something non-lucky: something about B that will always indicate a change in B (in the above examples, you might measure the amount of CO2 produced by the cells, or the number of times per day that you sneeze). You have to measure with ability: with enough sensitivity that you can pick up the change (your CO2 meter shouldn't be the same one that the city uses to measure pollution, and you shouldn't use a sneeze-measurement-system that relies on your cat freaking out at the loud sound [unrelated side note, what is it with cats and acting like the world is ending whenever a human sneezes?]). And, lastly, it needs to be safe: the measurement system needs to be accurate and reproducible so that you can trust the data you are getting.

Breaking that down was fun for me and probably no one else...