For AE 4:
I'm thinking we should separate the concept of ticket sales and actual flights operated. Say if an airline operates ORD-LAX, ORD-JFK, and ORD-SFO, he can choose which tickets to sell (LAX-JFK via ORD, SFO-JFK via ORD, etc). Passengers wanting to fly LAX-JFK would choose among all possible itineraries (LAX-JFK nonstop, LAX-ORD-JFK, etc) instead of actual flights.
This system has several advantages:
1) It easily facilitates a realistic connecting passenger model. Currently all demand is O&D, and connecting pax are generated based on hub traffic. Under the proposed system, the connecting pax would be real pax taking costs and flight times into account. Airlines flying LAX-DFW-JFK would compete for LAX-JFK passengers as well.
2) It allows for more flexible routings and stopovers. For example, if an aircraft doesn't have enough range to fly DFW-PVG, you can fly DFW-NRT and NRT-PVG and only sell tickets on DFW-PVG.
This sounds like a lot more micromanagement, but you would be able to choose default "rules" for new routes - sell tickets corresponding to flights only (no connecting pax), sell tickets on all possible itineraries (within timetable constraints), etc.
I think that would be best for routes like if I wanted to fly from Waterford Ireland to JFK I would have to fly to Shannon or Dublin first on a smaller aircraft cause a bigger 747-400 would not be able to takeoff from Waterford with out crashing into trees