Reliance and predictability

Will wonders why, when "a majority of justices can all be convinced that the Constitution requires something different than some old justices did, ... they should stand by error." In order to sustain this as an argument against a strong presumption of Stare Decisis one has to address the issue of retrospective versus prospective judicial decision making. This is especially a problem because, while previous courts showed some willingness to engage in prospective decision making, the current court is quite loathe to use it. In order to make this more concrete, let's look at the same example Will brings up, Roe v. Wade.

Many states (I don't know how many) still have their pre-Roe anti-abortion statutes on the books. If Casey had found that Roe was wrongly decided and that there is no Constitutional right to choose whether or not to have an abortion, all abortions that took place in those states between the time of Roe and the time of Casey would be criminal acts, and prosecutable if the statute of limitations had not yet run. This is because, without prospective decision making, Casey would have had to hold that their was never a Constitutional right to abortion and those statutes were always in effect. The interest in reliability is a reason for a strong presumption of Stare Decisis or for increased prospective decision making, but I don't see how you can weaken the presumption and leave retrospectivity in place as strongly as it currently is.