Yeah, the problem is that proper science is incredibly hard to do, and incredibly time intensive and thus expensive.

And since only a single metric (amount of published content) is really rewarded, anything else (including the fail-safes necessary for proper science) falls by the wayside.