You have a point, there are means and ends. I was using the term "means" as synonymous with "methods used to achieve instrumental ends", which I realize was vague and misleading. I suppose it would be better to say that consequentialism does not concern itself with means at all, and rather considers every outcome, including those which are the result of means, to be an end.

As for your other point, I'm afraid that I find it rather odd. Consequentialism does not need to be implemented as having implicitly summable values, much as rational assessment does not require the computation of exact probabilities, but any moral system must be able to implement comparisons of some kind. Even the simplest deontologies must be able to distinguish "good" from "bad" moral actions, even if all "good" actions are equal, and all "bad" actions likewise.

Without the ability to compare outcomes, there is no way to compare the goodness of choices and select a good plan of action, regardless of how one defines "good". And if a given outcome has infinitely negative value, than its negation must have infinitely positive value - which means that the negation is just as desirable as the original outcome is undesirable.