On the incompleteness of a model of algorithms for computing estimates

It is shown that the Linear closure $\mathscr L\{A\}$ of the algorithms for computing estimates is invalid on a set of regular problems, and therefore the model of such algorithms is incomplete. However, for effectively separable problems $\{Z\}$, with respect to a given system of the reference sets $\{\Omega\}$, the class of algorithms $\mathscr L\{A\}$ is correct. A counter example showing that the condition of effective partition of the problems is not essential for the validity of $\mathscr L\{A\}$, is given.