Furthermore, how long has Apple and the carriers -- certainly AT&T -- been doing this? If next year is the 10th anniversary for the iPhone, that's ten years. While I love Apple devices and I really like AT&T's ordering process, you'd think that the actual DevOps behind the sites would get a process down by now?
Personally, I don't understand why AT&T, Apple and/or whomever don't use Amazon Web Services (AWS) or similar scaling architecture to scale up as the traffic gets crazier, then scale back down as the traffic simmers down. That's the whole point of renting space at data centers, and if it's not Amazon, I know there are other vendors that provide rentable machines/capacity on their data centers.
I have never understood why that doesn't happen. Then again, maybe it's not a site issue, per se; it could be a billing systems issue or some such. In telecom, it's called "busy hour," the theory that states that you should provision how much capacity you need to reasonably handle the busiest time of your business -- without a meltdown.
Well, there are lots of pieces to the puzzle. Place an order on Apple's website, and it has to handshake with AT&T or one of the other carriers to confirm your eligibility and contractual details. So on and so forth. They could all go nuts and put in 10,000 x the normal required capacity on each part of the system to handle these once-a-year peaks, and spend a ton of money doing it, but they won't. They will go part way and call it good enough. They will figure "well, we only have enough stock for 5% of the initial orders anyway, so why make it possible for even more orders to flood in?"