Research showing how upward social comparison breeds competitive behavior has so far conflated local comparisons in "task" performance (e.g. a test score) with comparisons on a more general "scale" (i.e. an underlying skill). Using a ranking methodology (Garcia, Tor, & Gonzalez, 2006) to separate task and scale comparisons, Studies 1-2 reveal that an upward comparison on the "scale" (e.g. being surpassed in rank), rather than in the mere "task" (e.g., being outperformed), is necessary to generate competition among rivals proximate to a standard (e.g. ranked #3 vs. 4, near "the top"); rivals far from a standard (e.g. ranked #203 vs. 204), on the other hand, still tend to cooperate. Study 3 illustrates this finding with player trades in Major League Baseball. Study 4 further shows how an implicit scale comparison, instead of the commonly assumed explicit "task" comparison, may account for those classical competition findings in the literature. Study 5 then reveals how scale ranking becomes all important in the proximity of a standard, leading rivals to tolerate even an upward "scale" comparison to increase their proximity to the standard. Implications for the increasingly popular "forced ranking" management systems (e.g., at "General Electric") are also discussed. (Contains 3 figures.)