You're not wrong. I wanted to focus on one implementation first, given the challenges, but I'd like to go the spectral phase route next.