Black hole complementarity of Susskind and t Hooft was a consensus among string theorists for the interpretation of the information loss problem. However, recently some authors find inconsistency of black hole complementarity: large N rescaling, and arguments by Almheiri, Marolf, Polchinski, and Sully (AMPS) in 1207.3123. According to AMPS, the horizon should be a firewall (and hence, singular) so that one cannot penetrate there for consistency. There are some controversal discussions on AMPS. Apart from these papers, the author suggests an assertion on this problem using a semi-regular black hole model. It may be a good time to say some opinions on black hole complementarity by general relativists/loop quantum gravitists.