hutcheson - 4:29 pm on Sep 19, 2006 (gmt 0)

Ted's approach seems to me to have the kind of solid good sense that would be borne up even by the most rigorous technical analysis.

Google HAS to look random in its treatment of any single site; any failure to do this, and it becomes subvertible, And in today's web, that subversion the death of a billion spammer-cuts.

So every post that says "I can't understand what Google is doing with this site" is a testimony to Google's continued success at defending its results against systematic subversion. (A perfect defense against SHOTGUN subversion is, of course, an impossible dream.)

Again, Google HAS to maintain that unpredictability, even if that requires purely random unjustified drops in site ratings, it would absolutely be worth it (from Google's perspective and in the long run from the surfers perspective.)

Think of Google as a rabbit in a meadow, completely surrounded by lines of men with semi-automatic shotguns. What MUST it do? Move to avoid the highest concentrations of firepower, learn to survive ingesting a quantity of lead equivalent to the average density of pellets, and still find something to eat.

The hunters are already doing a fairly good job of filling the entire space with randomly fired pellets, which is THEIR optimum approach: because if they concentrate fire in some places, that's less free-flying lead somewhere else, which is where Google will attempt to go.
Ted's "signals of optimized keywords" are zones of high concentrations of pellets, the rabbit is well-advised to be elsewhere, no matter how many good clover patches it misses out on.