My thoughts:
1: Round one failed because of meager low level helicity. I’ve seen it many times and this is what happens when 0-1km helicity is 150-200. You have storms spin like a top, with little ground truth unless the storm rides a boundary or it strengthens enough to finally get a tornado down for a short period of time. This was pretty well modeled.
2. I believe round two failed because of unidirectional flow aloft. This lead to messy storm modes which also helped work over the area. Most of the modeled helicity was speed shear not directional shear(the opposite happened over AL)This was also well modeled.
3. Initiation of supercells the occurred earlier and speed of the squall line also helped. The past few days had supercell initiation over central AL at or around 21z to 0z, however supercells were being initiated by 11am this morning. Slow them down and the shear increases some, although I don’t know if it changes the event.
4. The CAMs were throwing out warning after warning with the overall lack of UH streaks over the high risk area. These are never fully accurate and bust often, but when models that absolutely love to nuke the area don’t nuke the area, they may be on to something.
