Black Holes: Complementarity or Firewalls? Ahmed Almheiri, Donald Marolf, Joseph Polchinski, and James Sully Abstract: We argue that the following three statements cannot all be true: (i) Hawking radiation is in a pure state, (ii) the information carried by the radiation is emitted from the region near the horizon, with low energy effective field theory valid beyond some microscopic distance from the horizon, and (iii) the infalling observer encounters nothing unusual at the horizon. Perhaps the most conservative resolution is that the infalling observer burns up at the horizon. Alternatives would seem to require novel dynamics that nevertheless cause notable violations of semiclassical physics at macroscopic distances from the horizon. The black hole information paradox presents a sharp conflict between quantum theory and general relativity. Gauge/gravity duality has provided some insight, giving strong evidence that all information is carried away by the Hawking radiation. It is widely believed that an external observer sees this information emitted by complicated dynamics at or very near the horizon, while an observer falling through the horizon encounters nothing special there. These three properties have been incorporated into the axioms of black hole complementarity. Various thought experiments have been examined, and argued to show no inconsistency between the observations of the external and infalling observers. For example, when a bit is thrown into a black hole, then as long as there is a minimum time of order $ r_s \ln(r_s/l_P) $ before the bit thermalizes and can be reemitted with the Hawking radiation, no observer will see illegal quantum cloning. This time scale has an interesting resonance with ideas from quantum information theory and from Matrix theory, which suggest that it may actually be achieved. There would be an inconsistency if one were to consider a large Hilbert space that describes both observers at once. Such a Hilbert space appears when quantum gravity is treated as an effective field theory, but it cannot be part of the correct theory of quantum gravity if BHC holds. This is consistent with the idea of holography, wherein quantum gravity is to be constructed in terms of degrees of freedom that are highly nonlocal from the bulk point of view. We consider a thought experiment that is a small variation on that of Ref. [4], differing in that it uses the naturally produced Hawking pairs rather than introducing additional entangled ingoing bits. This leads us to a rather different conclusion, that the thermalization time does not protect us from an inconsistency of BHC. Rather, if the experience of the outside observer is as we have assumed, then the infalling observer must encounter high energy quanta at the horizon. Our first thought experiment requires these only in low partial waves. However, a second thought experiment, using a detector lowered through the potential barrier to the near-horizon region, allows us to probe higher partial waves and come to the same conclusion about these. Thus, the infalling observer either burnsBlack Holes: Complementarity or Firewalls? Ahmed Almheiri, Donald Marolf, Joseph Polchinski, and James Sully Abstract: We argue that the following three statements cannot all be true: (i) Hawking radiation is in a pure state, (ii) the information carried by the radiation is emitted from the region near the horizon, with low energy effective field theory valid beyond some microscopic distance from the horizon, and (iii) the infalling observer encounters nothing unusual at the horizon. Perhaps the most conservative resolution is that the infalling observer burns up at the horizon. Alternatives would seem to require novel dynamics that nevertheless cause notable violations of semiclassical physics at macroscopic distances from the horizon. The black hole information paradox presents a sharp conflict between quantum theory and general relativity. Gauge/gravity duality has provided some insight, giving strong evidence that all information is carried away by the Hawking radiation. It is widely believed that an external observer sees this information emitted by complicated dynamics at or very near the horizon, while an observer falling through the horizon encounters nothing special there. These three properties have been incorporated into the axioms of black hole complementarity. Various thought experiments have been examined, and argued to show no inconsistency between the observations of the external and infalling observers. For example, when a bit is thrown into a black hole, then as long as there is a minimum time of order $ r_s \ln(r_s/l_P) $ before the bit thermalizes and can be reemitted with the Hawking radiation, no observer will see illegal quantum cloning. This time scale has an interesting resonance with ideas from quantum information theory and from Matrix theory, which suggest that it may actually be achieved. There would be an inconsistency if one were to consider a large Hilbert space that describes both observers at once. Such a Hilbert space appears when quantum gravity is treated as an effective field theory, but it cannot be part of the correct theory of quantum gravity if BHC holds. This is consistent with the idea of holography, wherein quantum gravity is to be constructed in terms of degrees of freedom that are highly nonlocal from the bulk point of view. We consider a thought experiment that is a small variation on that of Ref. [4], differing in that it uses the naturally produced Hawking pairs rather than introducing additional entangled ingoing bits. This leads us to a rather different conclusion, that the thermalization time does not protect us from an inconsistency of BHC. Rather, if the experience of the outside observer is as we have assumed, then the infalling observer must encounter high energy quanta at the horizon. Our first thought experiment requires these only in low partial waves. However, a second thought experiment, using a detector lowered through the potential barrier to the near-horizon region, allows us to probe higher partial waves and come to the same conclusion about these. Thus, the infalling observer either burns