ByLewis Perelman (© 1998, 2001)One of the most discussed aspects of knowledge management is the transfer of “bestpractices” and other efforts to “benchmark” within and between industries. However, Iam convinced that such initiatives are of questionable value and may in fact do moreharm than good. Intel Corp. offers an alternative—what it calls “best knownmethods”—that, for manyreasons, represents a more sensible approach.First, no one knows what the “best practice” is. There are multiple possible criteria of“best” and it is not generally obvious which “best” is “best.” Even if one knew what thebest practice was at a particular moment, something better is likely to come alongwithin a matter of weeks, or even days, hours, maybe seconds, depending on thescale and volatilityof the business.Intel’s “best known” puts the emphasis more where it belongs: on what we know, animmediate reminder that our ability to practice/perform is limited first by our ability toknow what’s going on. I even like “method” better than “practice”—it suggests a bitmore concern with understanding how to do than just with mimicking what is done. It’salso a more humble and realistic quest for the “’best,” since:
human nature being more inclined to report success than failure, “best practices” arelikelyto reflect more “spin” than reality; and
“best known” is elastic enough to embrace (at least in principle) bad methods—i.e.,worst practices, failures, mistakes, traps, and other “lessons learned.”Of course, failure and error generally offer more valuable lessons than success. As Inoted in my series on “worst practices” in Knowledge Inc., history in sports and othercompetitive fields suggests that victory usually comes to the player that makes thefewest mistakes, that eliminates the worst practices, errors, and such.One of the earliest theorems of cognitive science was that learning comes from trialand error. Statistically, because there are many more ways of doing something wrongthan doing it right, more knowledge is spawned by errors and failures than bysuccesses.Had Edison invented a workable light bulb after the first experiment, rather than the5000
, his prospects for commercial success might have been seriously impaired.Something that seemed that easy to do might have been harder to patent or to attractinvestors—who might have anticipated a flood of competitors. Edison also would havelacked the empirical database to further refine the design. Moreover, experimentalfailures in one time and place often hold the seeds to a breakthrough in anothercontext—Penicillin, AZT, the phonograph, and Java are a fewexamples.