How NOT to review a paper: the tools and techniques of the adversarial reviewer.
-- Graham Cormode, SIGMOD Rec. 37, 4 (March 2009), 100-104.

The reviewer can always complain that the data sets tested on are “unrealistically” small: if the data size is megabytes, demand gigabytes; if gigabytes, demand terabytes; if terabytes, demand chicobytes (A made-up scale of data, based on the Marx Brothers: chicobytes, harpobytes, grouchobytes, gummobytes and zeppobytes.)