> Even if the approach I describe is broken because groups board together, it becomes at worst a random boarding order which is still much more efficient than t

> Even if the approach I describe is broken because groups board together, it becomes at worst a random boarding order which is still much more efficient than the current nonsense. I think it can quickly become worse than random. Imagine you have the perfect sequence for max parallelism, but right in the middle is one 6 person family group who are going to take 10 minutes to get seated while they arrange toys and stuff for each child, blocking the aisle the whole time. With the perfect sequence, this halts all boarding for those 10 mins, since people need to pass the blockage for the scheme to work. Whereas with random boarding, there will actually be many people who manage to get seated during those 10 mins, because there will be a bunch of people who, by chance, are seated between the door and the blockage and can get to their seats.