when you talk about the full motion sims are you talking about the ones that are effectively chairs on the end of robot arms? or different styles?

For the record, I'm a commercial pilot.

Simulators have their place - but it is certainly nowhere near the experience as a real aircraft. Speaking from a commercial background, simulators are great at two things:
1) Procedure
2) Techniques

Simulators are great in showing pilots how things work. Want to know what to expect in a fogged in approach to an airport and are learning how to use the ILS etc? A simulator is *great* in this role. You can do things in this combo that are GREAT for education. Does it come anywhere close to the real thing? Hell no.

The other thing that simulators excel at is teaching things such as instrument scans - basically train you to keep an eye on all your instruments at the same time by developing an effective scan of them. No pilot flying on instruments will use a single instrument - flying is very complex and cannot be done like this. An effective instrument scan (A/H -> Airspeed -> A/H -> Altitude -> A/H -> VSI -> A/H -> DG etc) is very hard to grasp when first starting - and it is the bread and butter that keeps pilots alive when the weather is starting to deteriorate or you start to fly faster and bigger aircraft.

Your standard 737 pilot will probably spend about 15 minutes out of every flight looking out the windows. The rest is monitoring instrumentation. I cannot understate how important this skill is - and simulators are perfect at developing those skills.

So are simulators replacement for a real aircraft though? Nowhere near. Simulators should be treated as an addition to inflight training - not as a replacement for it.

Im also a military aviator, and "for the record" helicopters are a hell of a lot more dynamic than all those "F" series bombers. Tactical? what, flying circles at fl230 and waiting for a non pilot to tell you to drop a jdam? yeah, must be hell, no way a simulator can give that realistic feel of pushing buttons and doing the darth vader on the radio.

    Go listen to some bad music and choke down that cock flavored bourbon you all seem to love.

I second the parent.

Here at McChord AFB, we fly the C17.

Sure, there are many things you can practice in the sim.

But there are many more things that simply require the actual aircraft in actual conditions.

I think g-force is the biggest one. We had simulators that tilted and shaked back when Afterburner was new. Still doesn't mean I'd know what I'm doing or be able to do the movements I want to under 6 gs in any random direction.

I don't remember any sims in Star Trek that simulated in-flight g's. But then again they use artifical gravity for everything, so not much in the way of g forces to train for.

Perhaps the real lesson here is that we just need better inertial dampers and artifical gravity generators for our fighters.

For the record, I flew hand built RC planes about 10 years ago. This qualifies me to comment on all aspects of aviation, both military and commercial.

Is there any reason why a simulator mounted on a gimbal on the end of a spinning arm can't produce any amount of G in any direction?

(OK, it can never simulate zero G but I"m guessing that's not very important for dogfight simulators)

I don't believe you need to be in combat, nor on any kind of military mission to have stress levels increase - sometimes to breaking point.

I've seen stories of relatively inexperienced first officers on 747s go crazy on an approach to an airport among thunderstorms and bad weather. My favourite one was an asian first officer who freaked out, started singing some song in korean and forced the captain to do the workload of two pilots to land. Most of the training leading up to this was in simulators.

I've seen stories of training captains throw real world scenarios to students learning to fly airliners that screw up, crash what should be a very easy recovery, go "oh well" and hit the reset button to the sim. They then continued to do other exercises while never really learning the lesson to the situation being simulated.

You can then add to the list of problems the reliance on computers in aircraft which if training is not up to scratch can really cause problems. The excellent presentation "Children of the Magenta [youtube.com]" highlights a lot of these - and while watching the video you might think that a lot of what the presenter points out is common sense - it is at odds with most training done in airlines these days.

The procedural training in a lot of airlines is (basically): Take off -> Climb to 500ft AGL -> confirm aircraft is in a stable climb -> Engage VNAV -> confirm no sudden unexpected movements -> Engage LNAV -> Monitor instrumentation until on final to land.

Its a matter of time until all these issues cascade into an unfortunate event. Its up to the pilot individually how they choose to avoid these events and how they keep their skills in tact to work effectively.

Excellent point. And lets not forget, these arent company reps and engineers working on them usually. It's airmen and sailors and Marines. A lot of us are damn good. But we're not perfect, and these complicated machines. Sometimes it may take a month to get a really ornery gripe sorted out.

"New" cobra and huey models are coming to the Marines now. Some of those are all new airframes. Others are upgrades of existing airframes...and of those, some of those were upgrades of previous airframes too!

The mystery, intermittent gripe, the way "29" always seems to need longer to get the engine started, or the way "12" always pulls to the left a bit around X airspeed. Aircraft develop idiosyncracies.

Guess you did not have the same instructors I did...shouting, weird failures, then just for extra fun letting off a fire extinguisher in the cockpit while you're on short final, with nasty wind shear and one engine out. I don't know if I was scared, but I was certainly sweating ;)

(For the record, the first two times I buried it just before the paint).

Until they all go on virtual strike and a political enthusiast logs in as the President and fires them all.

The Big Picture: Salute To The Canadian Army [youtube.com]

2 PPCLI on parade, with Brigadier Gault in attendance.

Don't worry, the Soviet Army is gone now. Thankfully NATO was able to outlast the whole rotten system of militant, milatarized, oppressive Soviet Communism.

And what a nasty giant they were back in the day too.

Soviet Military Doctrine [foreignaffairs.com]

After the Soviet Union fell, the US was able to cut its defense spending [heritage.org], which had been falling over time anyway. Even with the cuts, the US was subsidizing Western Europe's defense.

NATO BURDENSHARING AFTER ENLARGEMENT [cbo.gov]

Or were you thinking of someone else? If so, could you be more specific? It is a little hard to reconcile international relations with playground rhetoric. It is made even more difficult by the tendency of some people to forget who their friends are.

You can somewhat simulate coming out of the sun - even some home sims like Falcon BMS do it. And yes, there is no simulation for G of course (I've been up in an old WW2 era fighter and even 4-5 G was like ... whoah... props to the guys in jets!) - which is why cockpit time will still be necessary.

But things like radar management, weapons deployment, formation flight, emergency procedures, etc. can all be performed in the sim. There's a hell of a lot more to a modern fighter aircraft than point and shoot with guns :)

Yeah, if the ILS is working correctly. Everybody knows they NEVER break down...

Combine that with something (fairly) common such as a bird strike, and I'll stick with a dude up front, thanks.

Of course, too many accidents (~25%) are still CFIT, so human pilots are far from perfect, but maybe we're getting a bit off the 'sim' topic here.

Thanks for the thoughtful reply. I have a few observations of my own, so I'm returning the favor. :) The ability to acquire experience, and the capacity to choose an informed course of action (judgment) is a very useful trait, indeed. But let me ask you this: What does a pilot do after bringing experience and judgment to bear on an in-flight situation? The pilot implements a procedure calculated to address the situation. The situations a pilot can encounter can be very diverse, as you indicated -- weather, equipment failure modes, other traffic, etc -- and a pilot has a finite number of ways to acquire information about the situation (instruments, ATC, and his own eyes and ears.) There are a finite number of combinations of control inputs that a pilot has at his disposal to resolve those situations. It's choosing which ones and when to apply them that becomes the key issue, and human pilots rely (as you point out) on their experience to make a judgment about that.

AI is a crapshoot, to be sure, but there is a class of AI that has a lot of potential (IMHO) to mimic the human capacity to learn and to make decisions based on that acquired knowledge. This is called an expert system, in AI-speak. An expert system can be trained (think of it as acquiring experience in a domain) and then use a decision tree to implement some course of action based on input from that acquired experience. They have to be trained (just like you and I do) and their responses are constrained by the limits imposed by the domain, just like ours are in the cockpit. Expert systems are already used in medicine, finance, and the military to make very sophisticated judgments based on what they know to implement procedures to reach desirable outcomes.

For example, military sonar systems use an expert system to discriminate between all the objects detected near the hull of a ship. They can determine which objects more than likely pose a hazard (like a mine, or a torpedo) and ones that don't (like a discarded beer can or a friendly ship) and take the appropriate action. (NB: The potential for "collateral damage" that you allude to is the same whether you have a human or a computer in the loop. Learning to recognize the potential for collateral damage is within the capability of an expert system. Human pilots can learn how to do this, so the expert system can, too.)

Training an expert system to fly is not trivial, but not impossible either. Anecdotally, I have a camera-equipped quadcopter that is controlled by an expert system that has been trained to stay a given distance from a given object while avoiding collisions. It can also launch and land autonomously. When I'm riding my Ducati, it sends a bird's-eye view of the road ahead of me to a tablet computer bungeed to my tank. I use it to detect roadside revenue squads before they can detect me. Training the system to chop my throttle when it detects a roadside revenue squad is what I'm doing every time I chop my throttle when I see that squad car appear on my tablet. The expert system remembers what I did, and can respond the same way when it detects those circumstances in the future. Replace "chop the throttle" with "launch an RPG" in the decision tree, and I think you can see that transferring my expert system's capabilities in the domain of a guy trying to avoid a traffic citation to the domain of a military strike aircraft trying to neutralize a ground target *is* very possible, if not as trivial. As we have both noted, it already is being done.