Title:What does strong subadditivity tell us about black holes?

Abstract:It has been argued that small corrections to evolution arising from non-geometric effects can resolve the information paradox. We can get such effects, for example, from subleading saddle points in the Euclidean path integral. But an inequality derived in 2009 using strong sub-additivity showed that such corrections {\it cannot} solve the problem. As a result we sharpen the original Hawking puzzle: we must either have (A) new (nonlocal) physics or (B) construct hair at the horizon. We get correspondingly different approaches to resolving the AMPS puzzle. Traditional complementarity assumes (A); here we require that the AMPS experiment measures the correct vacuum entanglement of Hawking modes, and invoke nonlocal $A=R_B$ type effects to obtain unitarity of radiation. Fuzzball complementarity is in category (B); here the AMPS measurement is outside the validity of the approximation required to obtain the complementary description, and a effective regular horizon arises only for freely infalling observers with energies $E\gg T$.