The plague of "frequency" was not really conquered in passenger allocation, only masked by a set of penalties, such as tech stop penalty and too small penalty, too many flights between airports penalty. This series of penalties brought AWS closer to real world with players flying aircraft resembling aircraft of real world.
But behind these penalties, the allocation engine still has only 1 variable: Frequency, with some small, barely perceptible side effects that make almost no difference.
In practice:
- Airline A can be running a single flight ( 1 ) the newest fast jet, premium seating pricing at 50% discount.
- Airline B can be running a single flight ( 1 ) with worst peace of garbage aircraft, pricing at 50% premium
The system starts the allocation with 1 = 1 then applies the penalties, then the barely perceptible side effects of seating quality, time of trip (speed of aircraft). There is almost never a zero allocated to an aircraft. So if the crappy aircraft gets something allocated no matter how crappy, just run a small enough aircraft, so that this something results in 100% LF.
That is be basic flaw in the allocation system, and now, with Cargo, the penalties are out, and we are back to:
1 Boeing 777 Freighter = 1 Douglas DC-3 Freighter
So let's fully fill up DC-3 with its 2900 kg of payload, let's load the 777 Freighter with 2900kg of payload and maybe give some spillover to the 777.
So, while in real world, no sane airline would fly 7 x 737-700C Cargo back to back instead of a single 777 Freighter, when the dust settles, running 7 x 737- 700C will become one and only strategy in AWS. Or perhaps even 20x 737-200 C. So we are back to square 1.
Maybe the allocation of cargo should be revamped and starts from 1 single variable: price with everything else (especially frequency) be relegated to noise. At certain price, Zero is allocated. As of now, I have detected almost no price sensitivity / elasticity to cargo allocation. Default, 10% discount, 10% premium all seem the same. * (side note).
Since the price an operator running 60x Cessna transport (and breaking even) is perhaps 10x of economics of running a single 777 Freighter, the economics will dictate the outcome.
Otherwise, the skies will be covered by 10s of thousands Cessna Cargo planes.
* (side note)
There is, of course, no way to really test anything by the player. Variable that is on level of noise (price sensitivity around default) is completely overwhelmed by the most useless variable of all (which IMO should be eliminated) which is the system generated randomness. With the randomness making any testing next impossible, always unreliable, there are areas of the system that perhaps could open new strategies for players (say premium economy seating of 2000nm flight). The system introduced randomness is 50x stronger than premium seating, and with 7 day cycle, the player will just never know. Say, you need 20 Mondays as a sample to test a hypothesis (to hopefully overcome the useless randomness), such as effect on premium seating on 2000nm flight. Suppose the absolutely most determined player decides to dedicate 3 days of his life to test this, and then "damn a C check".
Whereas, without a randomness, you run with First approach once, you run with Second approach once and compare the results, draw conclusions. Scientific method. All of this is off limits to the player.
The best game world ever was a Test Game world where not only 7 day cycle was turned off, but also randomness. Perhaps 7 day cycle adds value, could we run 9 out of 10 people without randomness, and 1 out of 10 with randomness, maybe even amped up, so that absolutely now one has any idea of what is going on.
So we could run 9 out of 10 games based on scientific method, and 1 out of 10 run on imbecilic method.
* (end of side note)