Jrgen, I had similar thoughts and confusions a little while back. I believe BlackNumero was trying to bring home the essential point that choice is context dependent.
You point out unless they (carrot and goat) were compliments(or the opposite) there is little reason to believe it necessary to strictly express the decision so strictly as the comparison of 2 bundles; you may as well go and just “sum” the utilities(converting the ordinal rankings into monontonically increasing numbers). Hence with 4 goods A-D, with ordinal rankings:
-
A
-
B
-
C
-
D
Then since v(A)> v(B)>V(C)>V(D), V(A+B)>V(C+D), so A and B are valued to C and D right? Now this MAY be the case but as you implicitly acknowedge, this may not always be the case, since A and B may be compliments(or the opposite, they may not “go well” together). Furthermore I would add, value scales themselves must be entirely context dependent, and so this would also be the case given goods that are substitutes in the sense that they are capable of serving the same ends to different efficacies(Rothbard generalises the notion of substitutes to the sense that all goods are essentially substitutes for one another, but I won’t go into that here). Hence if A and B were substitutes toward achieving the same end(for instance), then it would also not necessarily be the case that A and B would be preferred to C and D, since B might serve a much less valued end.
Hence we cannot impute the valuation of bundles of goods from individual goods and vice versa(as expressed cogently first by Menger). The tradeoffs in each situation of choice represents a tradeoffs of a different sort. Strictly, the avove value scale can only help us understand the tradeoff of one good, e.g. B for either A, C or D. We cannot use it to understand any other situation, since this would have to take into account how these can be combined to acheive an alternative set of ends in a situation of an entirely different context.
Hence value scales of the form shown above are defined strictly, only to help understand a given tradeoff and more generally to explain the consequences of the fact that men act and prefer one thing to another when constrained by scarcity. These value scales are simply a school of thought, and we cannot measure them, or conceive of them in any meaningful sense of existing “out there.” As Rothbard noted in his essay on welfare economics, the only thing that can be observed from action is the demonstration of preference from a tradeoff, for the thing acquired over that given up. However valuation of ends, and how these meaningfully affect the valuation of goods we cannot know, and changes from one action to another, and as I have noted above, would occur even if valuations themselves did not change.
In conclusion, i would highly recommend a careful reading of Mises’ section on marginal utility in Human Action. Do be wary however of one error he makes in his exposition however, which I pointed out here. Hope that helps.