The Feb. 2 Federal Diary column noted that the administration would replace the virtually automatic "in-grade" longevity raises of federal workers with a merit pay system that would "give bosses more leeway to give bigger, more frequent raises to top performers regardless of longevity."

This proposal is both right and wrong. It's right because the government should reward and attract the brightest and the best. It's wrong because the very core of a performance award is its objectivity, but the average "boss" cannot be objective because of unwritten personal, social and organizational biases -- "office politics," if you will. Office politics distorts and discourages complete objectivity. Just try to do your job after ignoring a pedestrian but necessary deputy while repeatedly rewarding that deputy's stellar subordinate.

There is a better way. An evaluation scheme known as "peer review" is used in academic circles. It is a scheme in which the work, the performance or a proposal of an academic is evaluated, often anonymously, by a small group of the reviewee's peers. While no system is perfect, this one is generally successful in ensuring that the issue under consideration is assessed only on its technical substance, with a minimum consideration of personalities or hierarchical imperatives.

I would propose that the peer review approach be introduced into the federal government. It might be applied only to managers above some "significant" level of responsibility for either staff or budget. Given the large number of federal managers in Washington, finding peers would pose little problem.

A corps of highly successful, peer-reviewed managers would greatly simplify governmental functioning. These certified individuals could be given more responsibility for running their own shops, with a consequent reduction in the huge quantity of ponderous yet ineffective regulation that seeks to guarantee excellence by substituting decree for thoughtfulness and that, as we all know, is evaded at will. DAVID H. SLADE Silver Spring