Finally, we explain the reason for forbidding to mention Reserve explicitly in our rules. Terms Reserve(t) always evaluate to false, so evaluating Reserve(t) or setting it to false is useless. But why not to allow putting the value true into Reserve locations. Elements can be discarded from universes, of course; to discard an element (represented by a term) t from a universe U, use the instruction U(t) := false. Isn't the reserve a natural place for unwanted elements? Yes, it is. Notice, however, that moving an element into the reserve may necessitate numerous changes of basic functions in order to ensure that the Reserve proviso remains valid. Would such a move contradict the sequential character of our rules? Not necessarily. We could just mark discarded elements as reserve elements, but then it might be necessary to augment rules with numerous guards Reserve(t)=false, which would be too tedious. It is preferable to leave the discarded elements alone. This pragmatic argument was put forward originally by Egon Börger.
But shouldn't the computational resources of the ealgebra simulating an algorithm A closely reflect the computational resources of A ? Yes, but it is important to separate the following concerns: the logic of A and the relevant resources of A . Concentrating on the logic of A may allow one to come up with simpler rules for the simulating ealgebra. And if one needs to track the resources of A , a separate bookkeeping may be set up. This separation of concerns allows us, for example, to use infinite universes. And caring about only particular elements and universes, rather than the whole superuniverse, makes combining ealgebras easier.