Changes in limpet larval development in either direction (to or from direct development) would count equally in a phylogeny. This is true unless some compelling reason existed, such as change in one direction being much more likely than in the other direction. If loss of larval development is more likely than reversal, than the most parsimonious answer would not be the best, and the computer model could be skewed to favor losses.
Work Step by Step
Consider parsimony: it looks for simplicity, and having loss and reversal count equally would be simple. However, if we know that is not the case, we need to account for it in building a model since we really want the truth, not simply what is the simplest answer with the fewest number of evolutionary steps.