... $c_{m_je}E_{m_j}$

As far as I see, this way of discounting of recommenders' event sets does not follow from definitions, e.g. $c^2E$ does not conradict them also. $cE$ is chosen for the sake of symmetry and balance.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... experiences

Good example of non-comparability is a case when some sender spams 1 mln addressees and sends 1 mln acknowledged mails to one addressee. Obviously, it has to be recognized as a spammer with absolute certainty.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.