⁓ maybe we need one algorithm for 'get the item right for the first time' and one for 'repeated practice'