This is called a Straw Man fallacy.

I never said that it was the MOST complicated, nor did I say that it has the best graphics engine. I said that "complicated software is never optimized at release." The logic in that statement is solid. I've been employed as a programmer since 1986.

You probably should not use logical fallacies when you are criticizing someone else's logic.

Software is, most of the time, optimized before release. Here are some examples of very complicated software which run fine at release:

- TESV: Skyrim
- Mafia II

Are those enough? or you want more?

Also, both games have HUGE worlds. With hundreds or even thousands of actors (AIs) and objects. Both of above games have also complex game subsystems.

Both games published by much larger and more financed companies. They also aren't as complex as calculating flight physics and damage modelling, engine management etc.

Apparently we are using different definitions of "optimized". "Running fine" is NOT the same as "optimized".

To me it means "Make the best or most effective use of a resource". That means the code should be as efficient as is possible. I have no idea why you think that either of those games were as efficient as possible at release, but I would bet that you're wrong.

We find ways to make code more efficient all the time.

The complexity of the simulation and of the damage model is really a matter of faith until you have a way to test what is doing the simulator.

You can have that on some simulators
like xplane and rfactor in a very clear and clever way
xplane:
http://www.youtube.com/watch?v=O_NDeSPCMks

rfactor 2 tyre consumption and dynamic simulation
http://www.youtube.com/watch?v=0ZWeEoOxKKw
http://www.youtube.com/watch?v=Oha1bQ9C174


that is a kind of complexity that you can compare with the real life simply because you see what is happening in the simulation.

The problem with Cod is that we don't have at all that kind of insurance on the physically correct simulation.

I'm sorry and I would like very much to say the contrary, but for now we have a Physic simulation with huge problems (velocity and ceiling of aircrafts is the very basic of the parameters to judge a correspondance with the reality) and a Damage Model that allows the Hurricane to fly with half wing cut off.
We were promised to have the acrobatic airplane to be able to admire the physic model but there is anything about it more than the promise before of the release.
About the damage model we don't know how it is made and we have not at all any analisys instrument to understand it (I hope to be corrected asap).
It seems that the convergence of the weapons is bugged, but how do we know that the ammo damage is made in a correct way ( or at least with a clear logic on which you can debate but at least you know it)

Those are both games with very little depth in them. They are not complex like a flight sim. At least try and be in the same ballpark

There is an huge difference between a software that needs to be optimized and a sowtware that needs a huge remake of a core system like the graphic engine... but perhaps I was thinking that was obvious but is not.

Sorry, but neither of those are anywhere near as complex as CloD. Nice character animation, sure, but a far stretch from modelling land sea and air and all the complexity and physics of a whole bunch of WWII aircraft and weapons systems.

You'll be telling me they model the ballistics of their pistols next.

do you have an idea of how complex is to manage only the facial expressions of a so big amount of characters?

I provided more similar examples, but talking about complexity it is not true that a flight simulator is more complex; more physic calculations is not meaning more complex. IMHO.

Sometimes you can optimize code by making small changes. Sometimes it requires big changes. It's not always possible to know that you made bad choices early in the design process.