Hmmm. Big topic. But a good one.
Last item first …The video shows a fairly ordinary day from 0200 to 2200. It happens to be the day before Thanksgiving, but some other day that same month back in 2013 would look *mostly *indistinguishable. It’s definitely lighter traffic than a random Wednesday in, say, October 2016.
What really stands out to me in the vid are how the flows differ by time of day. e.g. In the wee hours we see lots of eastbound but little westbound. Once it’s daytime the whole thing gets dense in every direction, but density on either side of the Mississippi is a lot different. Lastly in the early evening we see the NYC area launch a continuous hail of machine gun fire northeast toward Europe.
Trying to tackle your broad ideas more or less in order …
Demand:
Thanksgiving and Christmas have very different demand patterns. New Year’s is essentially a first order echo of Christmas.
Most airlines rely on both business travelers and leisure travelers to fill the jets profitably. Over all three holidays the two traffic demands ebb and flow in generally opposite directions but not exactly on the same timing or to the same degree. That opposition damps the peaks and valleys … a bit.
In Thanksgiving we see business travel demand collapse on Mon-Tue before the holiday, with little increase in leisure. On Wed leisure explodes. On Thu-Sat both are dead. On Sun-Mon leisure explodes again. With businesss picking up on weakly on Tue and increasing through the week. Business travel remains rather weak until about Dec 15th when it drops to near zero until midweek after New Years. Which might be the 3rd or might be the 7th depending on what dates are what days of the week.
After leisure travel’s post-Thanksiving Sun-Mon huge surge it collapses to nil until Dec 15th. Then begins a steady climb to the 23rd. It goes berserk on the 24th, drops back to near zero on the 25-26, then starts building again leading to a smaller peak on the 31st, then dead again on Jan 1, big peak on Jan 2-3, then falling back to the typical winter demand which is maybe 80% of summer demand.
As with business demand, the shape of leisure demand over Christmas & New Years depends a bit on how the holiday dates sit vs. the weekends. Thanskgiving doesn’t have that issue; instead we battle with the date being as early as the 22nd or as late as the 28th. Which affects how long the post-Thanksgiving to pre-Christmas lull lasts and how early in Nov we need to ramp for the holiday.
All in all demand is a real bucking bronco. Which we’re actively trying to shape by dynamic pricing. While hoping to outsmart those bastards painted the other colors who’re thinking the same about us.
Supply:
Airlines try to match supply to demand each as best they can. Supply is flexible, but not nearly as much as demand jumps around. Which is what leads to passenger frustration with over-crowding, high holiday fares, and all the rest.
Years ago it wasn’t as practical to make the schedule much different just due to administrative difficulty figuring it all out. Modern software and high capacity computing has removed that obstacle. Which means the limiting factor today is either logistical, or your company’s confidence in their demand projections; it remains the case that oversupplying seats in any given city pair on any given date destroys your margins all day, so there’s financial safety in being a bit conservative. Three knots above stall speed 
works so much better than three knots below :eek:.
As an example, my carrier’s internal propaganda tells us we’ll fly about 2/3rds of Dec’s total flights from the 16th-31st and only 1/3rd in the first 15 days. Given that the broad outline of the decision must be made in late September and finalized in early October, that’s leaning pretty hard into the shape of the predicted demand curve.
On the super demand days (the eves of all three) they can run a few extra flights. WAG 5% overall, 10% in key markets. By tricks like: make sure any aircraft going through short-term maintenance inspections are done a couple days before the spike and delay any others coming due by a couple days until after the spike using the grace provisions in the regulations. Decide to reduce the number of spare aircraft which are prepped to deal with day-of failures; but not too many or you may be starring in CNN’s annual angry stranded traveler disaster story.
Slot constrained airports like JFK, LGA, DCA, ORD and a couple others simply can’t hold more flights per day. They’re 100% full 365 days a year. The best you can do there is put in bigger airplanes. Or, ref that offsetting leisure/business mix, do things like kill some of the business-heavy flights to run more leisure-heavy flights to other destinations in the same time slot.
Air traffic, routing, fuel, etc.:
Baseline Nov and Dec traffic is/are already much less busy than summer. So on most of the days where we’re running extra for the holidays it simply means pushing a summer-days worth of airplanes through winter day. I don’t perceive it as particularly busy. In the northeast quadrant things are always near saturation anyhow. As mentioned above, the slot constrained airports look/feel the same every day of the year.
GPS hasn’t really changed much as to traffic. It *does *mean the enroute portion is straighter, quicker, and less fuel burn than the Olde Waye of tacking from VOR to VOR.
And yes, we’re more spread out. When we first got GPS and just used it to fly VOR airways more precisely we’d see a lot more close head-on passes that used to be offset a couple of miles when we were manually chasing wandering VOR needs with +/-4 degree tolerances.
Nowadays the head-on passes have declined and instead we see very gentle crossing paths where we’re both going direct from and to some random place whose routes happen to converge at some random shallow angle at some random spot in the sky.
But we’re not building new airports or runways. All the myriad GPS direct routes into a major airport terminate at just a handful of fixes 100ish miles from the field. Then everybody goes nose-to-tail at minimum spacing to landing. Minor airports are the same except the final merging / metering fix may only be 20 miles from XNA whereas its 150 miles from DFW. Said another way, the US airspace system isn’t short of airspace; it’s short of airports.
The potential future benefits of NextGen and full up GPS-centric airspace and traffic manangement design are another mongo post for another day.
Departures vs. arrivals doesn’t matter much.
One thing that does change is we carry a bunch of extra fuel on the peak days if possible. However harmful a weather slow-down into a major station would be on a holiday, it’d be 10x as bad if many aircraft end up diverting to outstations. Far better to bite the economic bullet and carry an hour’s hold fuel instead of 15 minutes; it’s (relatively) cheap insurance against starring on CNN. Insurance you couldn’t afford to carry 365 days a year but worth it on just 3 or 6 or whatever.
Weather is the big wild card. Winter snow is no good. But neither are summer thunderstorms. Extra fuel on board and extra ground crew including customer service staffing at the airports are our best defense. Any given disruption has the potential to go exponential. The quicker we can damp it back down the less chance it has to go divergent.
Crews:
The 1/3 2/3rd split for our Dec schedule I mentioned above implies most of the days off crews will have will be in the early part of the month. Days off during the last half will be relatively few and far between. And so go to the senior folks. You allocate your reserve crews extra lopsided versus the already lopsided schedule to absolutely maximize coverage on the key days of the key weeks. For reserves it tends to be **General Quarters! All Hands on Deck! **on the peak days.
As an example, I’m supposed to be off the 24th and 25th, but working several days straight on either side. If they call for extra volunteers (for overtime pay) on those key days I might decide to take them up.
As with generating extra airplanes by properly timing their recurring maintenance we can do the same with crews by timing their refresher training. Everybody spends 2-3 days every 9 months at the schoolhouse / sim. If we put a blackout on that during the roughly 3 spread-out weeks of holiday time we free up those pilots and the instructors and check pilots to fly real airplanes on those days instead.
Flight attendants are similar.
Paradoxically the most prominent effect crews have historically seen over holidays is all the flights that *don’t *operate on the dead days, such as Thanksgiving, Christmas, and New Year’s themselves. Which has the effect of “stranding” crews in hotels on the magic day when they’d otherwise have gone home a day sooner. It’s bad enough to be out over the holiday, but it feels especially futile / insulting to spend it sitting around a hotel doing nothing productive.
OTOH, I can recall at least the early stages of some truly epic parties involving several carriers’ with multiple crews each, the hotel staff who’re equally stuck, and the limitless supply of food and booze available at a big hotel once everybody loses interest in keeping track of it exactly. 