If you claim that X always increases when Y increases and someone shows you that X decreases when Y increases (during the period of the experiment) then either you're wrong (most likely), the experiment was poorly done (possible), or a miracle occured.
Doo doo doo, doo doo doo doo doo doo doo (cue swami music)
Correlation isn’t causation. Coincidental correlations CAN be turned into superstition.
Bad experimental design is obvious here. Insolation theorists make claims about the trend of the average over many centuries, not about glitches of a few decades. Why worry about the ripples in the pond when the whole pond is draining or filling?