Hypothesis: Endocannabinoids (ECs) are a general purpose "error status" notification system for animal body parts. Specifically, reduced concentrations constitute a chemical signal, generated in a democratic fashion by a large number of cells, that "something is likely to be dangerously wrong with this body part." This signal is then received by receptors (e.g. CB1 and CB2) mainly in neurons and/or immune system cells, stimulating them to try to take corrective action. That action may or may not be beneficial – and when it isn't that can be regarded as a health problem caused by malfunction or inappropriate action of your endocannabinoid system – but hopefully in net when all such events for some species are considered, it is. The same concept ("safe mode") is now near-universally used in, and has been beneficial for,many semi-autonomous spaceprobes. This gives a unified theoretical explanation (which makes numerous predictions) of the vast majority of ≈1000 otherwise random-looking isolated facts.

I artificially divided the confrontation with evidence into two levels: "simple" and "deeper look."At the simple level, the Hypothesis agreed with 100% of my ≥36 items of experimental evidence, providing confidence>99.9999999% versus the null hypothesis.Looking deeper, when we confront the Hypothesis with a wide spectrum of ≈100 items of experimental evidence it agrees with 90%, the exceptions being: (a) it only agrees with about 75% of bone evidence – and only with a specific sub-hypothesis about how it works in bone – and (b) it collapses, with prediction accuracy of about 50% or less, for evidence related to mammalian reproduction. Therefore I think the Hypothesis does not truly explain everything, just most things.Although no evidence item is very convincing by itself, the total produces at least eight "nines" worth of confidence. Perhaps a&b are because, e.g, mammalian reproduction evolved much later than the endocannabinoid system (at that time, neither vertebrates,nor animals with internal bones — both include all mammals – had yet appeared) and "overrode it";and to try to see the latter in its "pure form" we would need to investigate encocannabinoid↔reproduction effects in earlier nonmammal animals(an almost unstudied topic). Also complicating (b) is the fact that human pregnancy and birth is very atypical among mammals.

Along the way I show three mostly-new experimental facts: (A) Marijuana alone has very small or zero association withlung cancer, but (B) smoking both marijuana and tobacco causes lung cancer riskabout 2.4× the risk from smoking tobacco alone, which in turn is over 10× the risk faced by nonsmokers; (C) oral dose of 1 large drop of oil extract from marijuana cures a substantial subset (but not all) of achalasia patients for 60 hours.