Anthropic signature: strange anti-correlations

Imag­ine that the only way that civ­i­liza­tion could be de­stroyed was by a large pan­demic that oc­curred at the same time as a large re­ces­sion, so that gov­ern­ments and other or­gani­sa­tions were too weak­ened to ad­dress the pan­demic prop­erly.

Then if we looked at the past, as ob­servers in a non-de­stroyed civ­i­liza­tion, what would we ex­pect to see? We could see years with no pan­demics or no re­ces­sions; we could see mild pan­demics, mild re­ces­sions, or com­bi­na­tions of the two; we could see large pan­demics with no or mild re­ces­sions; or we could see large re­ces­sions with no or mild pan­demics. We wouldn’t see large pan­demics com­bined with large re­ces­sions, as that would have caused us to never come into ex­is­tence. Th­ese are the only things ruled out by an­thropic effects.

As­sume that pan­demics and re­ces­sions are in­de­pen­dent (at least, in any given year) in terms of “ob­jec­tive” (non-an­thropic) prob­a­bil­ities. Then what would we see? We would see that pan­demics and re­ces­sions ap­pear to be in­de­pen­dent when ei­ther of them are of small in­ten­sity. But as the in­ten­sity rose, they would start to be­come anti-cor­re­lated, with a large ver­sion of one com­pletely pre­clud­ing a large ver­sion of the other.

The effect is even clearer if we have a prob­a­bil­is­tic re­la­tion be­tween pan­demics, re­ces­sions and ex­tinc­tion (some­thing like: ex­tinc­tion risk pro­por­tional to product of re­ces­sion size times pan­demic size). Then we would see an anti-cor­re­la­tion ris­ing smoothly with in­ten­sity.

Thus one way of look­ing for an­thropic effects in hu­man­ity’s past is to look for differ­ent classes of in­ci­dents that are un­cor­re­lated at small mag­ni­tude, and anti-cor­re­lated at large mag­ni­tudes. More gen­er­ally, to look for differ­ent classes of in­ci­dents where the cor­re­la­tion changes at differ­ent mag­ni­tudes—with­out any ob­vi­ous rea­sons. Than might be the sig­na­ture of an an­thropic dis­aster we missed—or rather, that missed us.