Maybe the logistic success curve should actually be the cumulative normal success curve.
I guess so! Is there reason to favor logit?