I have just one request/suggestion in regards to Microsoft Flight (or any other new sim). Please improve The AI Engine significantly and/or give third party devs full documented access to the sim to allow them to do the same.BTW, since we're talking about a flight simulator, this relates only to flying AI. And I'm not talking about birds here... Just though I'd make that clear...:(

One would have thought that after 2 versions with AI, FS users would really want AI. Still, a better approach to AI would be nice to ensure that they interact with user aircraft works better than it does now. That also means an updated ATC system so the AI controllers can ensure better separation in the landing pattern and of course, ensuring better ground movements so AI does not disappear after a set time.

One would have thought that after 2 versions with AI, FS users would really want AI.

3 :PIn my opinion there's a few small nags that needs to be sorted out, before I can declare myself happy with AI state of affairs.

• Proper looking pushback procedure, or at least make it customizable. It still gets on my nerves, watching a 747 doing a 90 degree turn on a platter after being pushed straight back from the terminal.
• Better integration of parking procedures (ie. drive-through parking, taxi and turn parking)
• Allowing for stands with multiple parking locations, letting FS determine when to use one for a large aircraft or two side-by-side for smaller commuter types.
• Allowing AFCAD/ADE designers to designate a preferred taxi routing. Places like Mallorca drives me nuts, since the aircraft instead of following the outer, main taxiways, start taking shortcuts across the inner apron.
• Making ATC make less of a hash of cul-de-sacs. We've all been there, seeing 4-5 aircraft stuck on Taxi In to a cul-de-sac, because there's a single commuter jet going out from the same.
• Making the Flight Sim recognize when a runway is closed for takeoff or landing on one end only. This is pretty much my biggest pet peeve, since it completely screws up any airport with offset parallel runways (MAN & CPH are two glaring examples).
• And one really outrageous one, allowing users to design more complex approach ala the River visual into DCA, which can't be done with the current system.

More to be added as I think of them.

That also means an updated ATC system

Indeed and I think we should get rid of the notion that user and AI should be treated differently. If a new and improved AI engine and ATC system are going to work, user and AI need to be treated the same. Getting slightly off-topic, but how many times haven't people asked about user aircraft and parking codes?! If something works for AI, then it should for the user and vice versa.

•Making the Flight Sim recognize when a runway is closed for takeoff or landing on one end only. This is pretty much my biggest pet peeve, since it completely screws up any airport with offset parallel runways (MAN & CPH are two glaring examples).

Yes, it should be possible to configure runway ends independently from their reciprocal ends. That's high up on my list as well. Closely related, get rid of the need for fake parallel runways and don't hard link a set of runway ends to another set of ends. For example, sometimes it makes sense to have the engine use base end of runway A and reciprocal end of B. Instead of base A/base B or recip A/recip B (or more complicated setups thanks to the advanced crosswind tutorial).Two changes that would make a huge difference. No more need for multiple Afcads for an airport, just one... and one that will be adhered to. Closed means closed!

•Allowing AFCAD/ADE designers to designate a preferred taxi routing. Places like Mallorca drives me nuts, since the aircraft instead of following the outer, main taxiways, start taking shortcuts across the inner apron.

Exactly. And one-way taxiways, man, that would be great! No more shortest route from A to B, but realisting routing.An idea I've always liked is the ability to configure airports based on local time. So many airports have night ops (or other time dependent ops), why not let the Afcad designer set options to that extent. Day, certain runway config, night, other runway config. One Afcad.I'd like to see much better AI operation at one runway airports, which basically means better sharing of one runway by take-off and landing queues. Don't give one queue preference over the other, but merge them into one efficient smooth flow.Clean runway exits.Approach separation, enough said.

Unless you really enjoy flying with the sky to yourself there's no reason to dislike AI, especially if it can be disabled. The question is probably actually "do you use AI?". I would say that most real-world pilots (and I fly GA) would rather share the sky in the sim in as close a way as they do with other pilots in the real sky. That means some esoteric development and design decisions that for some peoiple might seem pointless (like some previous suggestions about taxiway, approach and runway closure behavior).The one thing I'd add to that is giving the AI aircraft the ability to take rapid exits if the ground speed is below some threshold (eg 50kts) and also to make rolling take-offs if cleared to do so. It would improve runway occupancy a great deal. Also allowing intersection departures for AI, that would also be useful.One nice-to-have ATC thing would also be to declare emergencies and have the simulated ATC handle it in some appropriate way - and that's where AI and user boundaries start to get blurred, because if we are going to declare an emergency then we are going to have to expect ATC to intereact with us, and I don't think its that big a step from awareness of the user aircraft to awareness of all aircraft, AI or otherwise.I fly in a sky with a great many other people, and a sim that doesn't simulate that is missing a lot of the flying experience that is in the real world. A sim without AI is half a sim, and the flying could just as well take place in a floatplane on a deserted lake, an aerobatic biplane doing displays at a deserted airfield, or a MiG-29 at a deserted air base in Siberia. The word deserted crops up a lot. There's no point simulating an approach into a deserted Heathrow, that's just not as real as it gets!And I'd add one more important AI deficiency that needs addressing, and that's airspace. Airplanes bound for Heathrow drag themselves across London at 2000' infringing any number of airspaces and prohibited areas en-route (Biggin, LCY) and airplanes below 10,000' in England get handed over to any number of en-route airports whose ATZ's extend to only 2000' AAL, when they really shold be in the hands of either a lower controlled-airspace controller, or flying with an FIR controller who may hand them off to airports if they will enter the ATZ, but otherwise allow they to fly with a flight information service.I usually fly in the US, but flying the sim in the UK has highlighted these areas a lot more to me, as I come to understand how the UK manages its airspace.And the biggest deal of all is to make the system as exposed to the users as possible, so that developers can pitch their expertise into the end result. If you want the ability to design airports, and airspace, and approach/departure procedures it would be great to have. Exposing the interfaces means that M$ don't have as much work to do, and many developers have said for years "give us the tools and we'll do the job".

Unless you really enjoy flying with the sky to yourself there's no reason to dislike AI, especially if it can be disabled. The question is probably actually "do you use AI?".

Well said. :(

And the biggest deal of all is to make the system as exposed to the users as possible, so that developers can pitch their expertise into the end result. If you want the ability to design airports, and airspace, and approach/departure procedures it would be great to have. Exposing the interfaces means that M$ don't have as much work to do, and many developers have said for years "give us the tools and we'll do the job".

I think this is definitely the idea that user and third party firms should be putting to MS as I feel that they will be looking to reduce costs and development time.

Unless you really enjoy flying with the sky to yourself there's no reason to dislike AI, especially if it can be disabled. The question is probably actually "do you use AI?".

The poll was a last minute addition to see if and how it worked, I didn't give it too much thought, so neither should you, but like is definitely closer to my point of view than use. It is the AI enthusiast's point of view, I don't just use AI, I dig it, if you know what I mean. Users won't care enough to press devs for changes, that's up to the smaller niche within the niche. :(

Take AI off the main thread, its a resource hog

Take AI off the main thread, its a resource hog

In general, I hope they will make more efficient use of multiple CPU cores. FSX did it for scenery compositing, but it could be done for other things too, like AI. We have sub-$200 6-core CPUs from AMD now, while the more expensive Intel equivalents can run 12 threads at once (two per core). In the coming years, these numbers will only increase.I think some are asking too much from the AI programmers. If AI was 100% realistic in the sim, what would be stopping it from also replacing real-life, human air traffic controllers and pilots? :)

The only new thing I would want to see with the AI traffic is helicopters added. Airplanes aren't the only thing that fly in our skies.:(

The only new thing I would want to see with the AI traffic is helicopters added. Airplanes aren't the only thing that fly in our skies.:(

I would have to agree with that sentiment.  :(

I’d like to see VFR aircraft flying proper circuits too :)

Heli traffic? http://www.flight1.com/products.asp?product=helitraffic

Heli traffic? http://www.flight1.c...uct=helitraffic

I know about that. I'm talking about Flight.

In general, I hope they will make more efficient use of multiple CPU cores.

I can't believe that they would be stupid enough not to make full use of multiple cores with MS Flight. That's the way forward for CPUs, so the software developers will have to follow.