Why do airplanes use zones to board passengers?

I just got back from a trip to San Diego. One of the (very few) great things about traveling with an infant, is that you get to board right after first and business class passengers are through. Even so, I made a mental note to try and learn something about boarding zones, after I got back.

Boarding an airline turns out to be a nontrivial optimization problem.

Clearly, an appropriate function to be minimized is total boarding time (with special consideration for "special" passengers, as a constraint). Key variables are the number and distribution of seats on the aircraft, the fraction of total seats occupied, and the stochastic nature of the boarding process itself.

It turns out that there is no clear winning formula, although some strategies are better than others. Here is a nice description, which animates some common strategies, and links to some prominent primary sources. An interesting picture from simulation studies (from that site):

Two interesting points. If the passenger inter-arrival time is more than 10 seconds, all the strategies seem to coalesce. Secondly, although reverse-pyramid is one of the more efficient choices, no commercial airline seems to use it. Instead, many airlines seem to prefer back-to-front which is among the more inefficient methods.

Why? I don't know the "real" answer, but it seems to me that people traveling together in groups typically choose adjacent seats. Reverse-pyramid or outside-in (United Airlines) would split them up.

Should modelers incorporate this criterion into their optimization models?