occupies most of ). This will be the case if x's rate of increase is proportional to the observed average payoff of instances of x at time t or, roughly,
It will still be the case if the rate is proportional to the schema's "usefulness," the difference between and the overall average performance of instances in , so that (In genetics is called the "average excess" of x when x is defined on a single locus, i.e., when x is a specific allele.)
The discussion of "intrinsic parallelism" in chapter 4 would imply here that each x represented in should increase (or decrease) at a rate proportional to its observed "usefulness" . If this could be done consistently then each x would be automatically and properly ranked within as t increases. The reasoning behind this, as well as the proof that reproductive plans accomplish the task, will be developed in full in the next two chapters.
). This will be the case if x's rate of increase is proportional to the observed average payoff 
of instances of x at time t or, roughly,
of instances in 
(In genetics 
is called the "average excess" of x when x is defined on a single locus, i.e., when x is a specific allele.)
. If this could be done consistently then each x would be automatically and properly ranked within