The references are very amusing. A whole lot of stuff is considered harmful, or no longer harmful:
A. Perlis, "Public-private key pairs considered harmful," UC Berkeley, Tech. Rep. 4221, June 2005.
P. Zhao and L. Subramanian, "Replication no longer considered harmful," in Proceedings of POPL, May 2003.
A. Turing, F. Zheng, and D. Ritchie, "Redundancy no longer considered harmful," Journal of Perfect, Large-Scale Theory, vol. 0, pp. 1-16, Oct. 1996.
J. Miller and R. T. Morrison, "Superblocks no longer considered harmful," IEEE JSAC, vol. 20, pp. 45-53, Feb. 2004.
A. Pnueli, J. Smith, and P. ErdÖS, "Multicast systems considered harmful," Journal of Pseudorandom, Homogeneous Theory, vol. 42, pp. 150-192, Apr. 1990.