Since we had this discussion about UAVs and F22s etc...

[MirariNefas]

Even a single celled organism is intelligent, so we must have missed something.

No.

[alexjrgreen]

Even a single celled organism is intelligent, so we must have missed something.

No.

See, for example:

Artificial general intelligence: an organism and level based position statement

Microscopical Substantiation of Intelligence in Living Cells

[Tom Ligon]

The driver in this air combat game is that the US presently dominates the air over any pipsqueak country it cares to. If they send up planes to defend we splash them before they get off a shot. What will drive UAV fighters is third-world countries that would like to stand up to us. If they can buy cheap off-the-shelf UAVs that cost nothing to train, cost a fraction of what a manned fighter does, don't have relatives to send sad letters to, and can worry a big air force enough to be a deterrent, that sort of UAV will sell like hotcakes.

How fast might UAVs advance? Are you familiar with the DARPA Grand Challenge? They set up a cross-country route thru the desert and had teams compete to see who could cross it. The first competition was a total cluster-reproductive-act ... essentially every competitor that even managed to cross the starting line was broken down or upside down within a mile or two of the start. But the second competition had, I think, five vehicles which reached the finish in the allotted time. The progress was stunning.

Driving thru the desert requires a fantastic amount of situational awareness. By comparison, consumer grade combat flight simulators do manage to produce dangerous AI pilots. The environment is simpler than the desert. If you don't know what you are doing, these can shoot you down quickly, using aircraft comparable to your own.

Can such an AI's weak points be identified? Absolutely. I can take on as many as 20 Me262's in Jane's WWII simulation, versus my lone Mustang, and usually be the last plane flying, sometimes without a scratch. The 262 in that sim is toothless at very low altitude. My limit against the FW190 is about 4 planes. If I were up against a plane half the size, same firepower, faster, more agile, which literally does not know the meaning of "fear", with just that amount of smarts, I would be in serious trouble.

What we have learned in exercises such as Red Flag is that a swarm of small fighters will defeat a few larger fighters. If we wish to continue to dominate air wars, we will need to deal with this. UAV fighters are coming. I suspect we will have some of the best, I suspect we will find ways to avoid the ones the other side has, and I suspect we will find ways to match or better their capabilities. But we will not be able to ignore them.

What I expect is we will team human-crewed aircraft with UAVs, using the humans to decide what to hit and how to hit it while the UAVs handle the suicide missions and hopeless furballs.

There is an old truth: fighter pilots make movies, bomber crews make history. We don't have fighters so we can dogfight other fighters. We have fighters to escort bombers, and to defend against bombers. Too much focus on dogfighting ignores the real mission.

[Diogenes]

Can such an AI's weak points be identified? Absolutely. I can take on as many as 20 Me262's in Jane's WWII simulation, versus my lone Mustang, and usually be the last plane flying, sometimes without a scratch. The 262 in that sim is toothless at very low altitude. My limit against the FW190 is about 4 planes. If I were up against a plane half the size, same firepower, faster, more agile, which literally does not know the meaning of "fear", with just that amount of smarts, I would be in serious trouble.


.

The P-51 mustang is classed as the greatest fighter aircraft of it's era. It defeated Me262s in real life, as well as everything else that was thrown against it. The FW 190 is the closest thing the Germans had to a P-51, but it was no match for the king of the skies. Even so, I very much doubt any p-51 could survive an encounter with 4 FW 190s. I suppose this has more to do with your critique of the AI in the program than any real criticism of the difference between the airplanes.





There is an interesting point that one of my Friends (Air force fighter Pilot/Engineer) told me about the P-51 mustang. The bump on the bottom is a radiator to dissipate engine heat. Due to the inlet and exit cowling design, the radiator actually produces positive thrust on the air craft.

[Tom Ligon]

Durn, when did Big Beauiful Doll switch to that stubby 2-blade prop?

http://www.cebudanderson.com/bigdoll.htm

My trick against the jets really is a fault of the plane. For some reason, Jane's makes the 262 slow down when near the ground. Consequently it carries a lot of angle of attack, and can't train its guns on a lower aircraft. If I stay low enough I'm bulletproof. The FW190 does not share the problem. I can just keep making head-to-head passes with clusters of them, pick out tail-end-Charlie, and blip my nose up to rip his belly as he passes over.

The effect is not total malarky ... a History Channel "Dogfight" episode of a huge WWII dogfight during "The Big Week" has an account of a P-47 pilot in pursuit of a FW190, both planes trimming the grass. The Jug kept shooting over the German's head, unable to depress his guns enough to hit the target. The FW finally pulled up to clear some trees, and the Jug nailed it.

The usual outcome of a Mustang/Me262 fight was the jet ignored the Mustang and just sped away. On a few occasions they were caught low and slow, usually while landing (Yeager got one that way). Had they engaged in a turning dogfight they would probably not done well, but they could choose their own game. It was not their job to dogfight, they were either fighter-bombers or defending against bombers. There is no way my particular scenario would ever have happened becaused massed jets would not have engaged the prop fighter. In this mission, the stupidity of the AIs is they do engage me.

I probably died fifty times before I learned to beat them.

[Skipjack]

I probably died fifty times before I learned to beat them.

How many times can a real life pilot die before he has a chance to learn about how to beat them?

Again, I think that the dogfighting days are over. Nowadays you fire your fire and forget rockets at a blimp on the radar screen. You destroy targets you have never even seen. UAVs are small and they are getting more and more stealthy. In addition they are usually slow aircraft. All that would make it very hard to identify a UAV as such without coming fairly close. A F22 is a comparably large, fast moving aircraft. Sure it has stealth, but in this game, I think it is about who sees the other one first.
I would put my money on the UAV. Two to three people piloting a thing that is much, much smaller than a single seat fighter jet.
Two to three people that have their eyes on all instruments all the time. Two to three people that have had their coffee in the morning and for brunch + a doughnut. People that are not tired, not feeling any physical or emotional stress...
I think any fighter pilot will have a tough time with that.

[alexjrgreen]

The P-51 mustang is classed as the greatest fighter aircraft of it's era.

By whom?

Comparing a Spitfire XIV (entered service January 1944) with a Mustang P-51D (entered service June 1944):

The Spitfire XIV was faster than the Mustang, more maneuverable, had a higher service ceiling, could climb better, and even had a better rate of roll, which was formerly the Mustang's only performance advantage over the Spitfire. It was superior to the P-51D in EVERY combat category except initial dive speed and range, and the only way range came into play in a dogfight is if the P-51 could fly around long enough for the Spit to run out of fuel!

British Spitfire or P51 Mustang?

[Diogenes]

The P-51 mustang is classed as the greatest fighter aircraft of it's era.

By whom?

Comparing a Spitfire XIV (entered service January 1944) with a Mustang P-51D (entered service June 1944):

The Spitfire XIV was faster than the Mustang, more maneuverable, had a higher service ceiling, could climb better, and even had a better rate of roll, which was formerly the Mustang's only performance advantage over the Spitfire. It was superior to the P-51D in EVERY combat category except initial dive speed and range, and the only way range came into play in a dogfight is if the P-51 could fly around long enough for the Spit to run out of fuel!

British Spitfire or P51 Mustang?

I believe I saw that rating on the military channel quite some time ago. They claim to have polled experts from all over the world, and the consensus was the P-51 mustang was the greatest fighter of it's era.

Also, don't sell it's long range advantage short. From a military perspective, that's a HUGE advantage. Also consider this. The heart of both beasts were Rolls Royce engines.

[Diogenes]

Durn, when did Big Beauiful Doll switch to that stubby 2-blade prop?

http://www.cebudanderson.com/bigdoll.htm


The usual outcome of a Mustang/Me262 fight was the jet ignored the Mustang and just sped away. On a few occasions they were caught low and slow, usually while landing (Yeager got one that way). Had they engaged in a turning dogfight they would probably not done well, but they could choose their own game. It was not their job to dogfight, they were either fighter-bombers or defending against bombers. There is no way my particular scenario would ever have happened becaused massed jets would not have engaged the prop fighter. In this mission, the stupidity of the AIs is they do engage me.

As you say, the Me262s were designed to stop the bomber attacks by relying on speed which couldn't be matched by other aircraft. They simply strafed, turned around, and strafed again till the targets were done for.
The Me262 suffered from a characteristic of jet engines, and a problem that still besets them today. (Karen Hultgreen) They don't like rapid airflow/throttle changes, and can flame out. The Me262s could not rapidly change from a landing throttle setting to a fighting throttle setting, so when they were coming in for a landing, they were dead ducks.

Another interesting story about the Me262 I heard from one of my History teachers. He was a turret gunner aboard a bomber heading into Germany for a bombing run. It was his, and the other gunners jobs to be looking out for enemy aircraft, and they had been hearing rumors about the German's secret weapon aircraft. The consensus was that if you got caught by one of these, you were pretty much doomed.

As he was manning his post, he saw several somethings coming at them at an incredible velocity. WOOSH ! and they were gone. ( Yes, he said Woosh!) All the gunners wondered "What was THAT ! "
Pretty soon, the strange aircraft turn around and came at them again. Everybody thought they were gonna die, but for some reason the Me262s didn't open fire. They come back for one more pass, and then they left.

He said he found out after the war that during that encounter, those particular aircraft had already exhausted their magazines against a previous attack, and had nothing left to shoot, so they tried to spook the air crews into abandoning their mission.

[Diogenes]

Durn, when did Big Beauiful Doll switch to that stubby 2-blade prop?

http://www.cebudanderson.com/bigdoll.htm

Looking at both pictures, i'm thinking the picture I posted appears to be that of a smaller aircraft. I suspect it is one of those 3/4ths sized kit (or replica) planes that are popular among aviation buffs. It is probably just painted to look like the original Big Beautiful Doll.

[alexjrgreen]

I believe I saw that rating on the military channel quite some time ago. They claim to have polled experts from all over the world, and the consensus was the P-51 mustang was the greatest fighter of it's era.

William Dunn (US fighter ace who flew Spitfires, P-51s, Hurricanes, and P-47s): "Now, if I had to make the choice of one fighter aircraft above all the others - one that I'd rather have tied to the seat of my pants in any tactical situation - it would be, without any doubt, the world's greatest propeller driven flying machine - the magnificent and immortal Spitfire."

British Spitfire or P51 Mustang?

[Betruger]

Yeah, that P-51 is a scale model. Look at the "pilot".


This should be in a new thread...

The XIV, XXI, and a few other marks of the Spitfire could be competitive with the Mustang, depending on their configuration. The XIV only arguably so, IMO. That engine reduced agility by too much. And you have to compare that XIV to a regular P-51D. Also, lots of souped up P-47s were just as competitive. In fact one pilot (Robert Johnson) explicitely said his lowly 47D-5 was the fastest jug he'd flown (and he'd flown all of em, incl the Ms of the 56th). It's an apples to oranges comparison, but consider the 47's firepower, ease of aim (8x50), reach, ruggedness, handling at pretty much any speed, and the XIV's performance at the cost of such cumbersome engine torque (such a difference the makers considered renaming it as a separate model) and comparatively small firepower and ammo duration, and it's easily starting to tip in the 47's favor.

That D-5 was a souped up early model D. The N and M models equally improved would've given the P-51D a good run for its money, even if it were also field-modified. I don't remember the charts exactly, but IIRC a stock N model jug's power/altitude curve is just one big clean curve all the way to 35 kft, with a speed record around 30 kft.

Exhaust thrust plays a significant part in total thrust for a number of planes. At lower altitudes it's negligible, but high enough it's pretty large. A Ta 152H-1 gets something like 200 HP around 30k.

German-side, the luftwaffe was just too attrited to compete. Overall, the P-51 is renowned as best of all only because it's famous.. Because it's a pretty plane. The real deal is very arguable. Considering the rich variety of performance and maintenance realities from model to model, and from airframe to airframe. Taking various pilots' opinion gives almost as many different answers. It's not impossible to aggregate all such opinions and thorough performance data to make a more comprehensive assessment.
http://www.wwiiaircraftperformance.org/
Also consider the F4U-4 in dogfighting circumstances. Or the 109K-4 running at 1.98 ata. Then you have a number of barely-made-it models that got on the scene just as their theatres closed, like the F7F (I hate it - it's the greatest most faultless prop fighter I've ever tested - paraphrase of Grumman or USAF chief tester) and F8F, etc. Or even stuff that would have been pretty revolutionary if it hadn't been cancelled because of jets' imminence (e.g. the X5FU). It's a real zoo. Nothing like the clear superiority of a single plane like the P-51's supposed to be.

The FW 190 is the closest thing the Germans had to a P-51, but it was no match for the king of the skies.

That's really arguable. Which 190 and which 51? The comparison can't be averaged from the range of all variables. You have two planes with pretty different lineage, different purposes, each one with a pretty wide range of models. Then you have a range of circumstances in which to pit them against one another. Initial altitude, merge range (e.g. whether or not one or both spotted the other, as frequently didn't happen in WWII), what loadout each one was carrying, etc.


Re: AI: I'm no specialist, but I doubt that consumer grade flight sim AI is comparable to military grade. Consumer PCs have to handle everything else (graphics, physics) in the same resource budget as AI, so that's one big limitation already. The military also doesn't have the cramped funding of video game developers.
That said, the problem with game AI in prop dogfighting is that despite any "perfect" execution of ACM, E-management, etc, AI doesn't learn from fighting you and is too predictible once you've flown against it enough. And/or you can use its own perfectly disciplined routines against itself.. Say trick it into "detecting" you as going for some ACM, and then hit its parry to the ACM it "detected" where it's vulnerable.
Where AI seems to be sufficient, in an equal plane/equal energy merge, is anytime it manages to establish a dominant position. Then all it has to do is textbook grind, till you run out of altitude/energy/angles and have no choice but to concede a guaranteed killshot solution. I don't know if this is valid in modern jet dogfighting, though.

What I've yet to see myself is long range combat AI in modern setting. You'd expect that the much simpler (relatively) mechanics at work at long range will be sooner conquered by UAV AI than close quarters dogfighting. Or very large flocks of birds, where AI coordination has to have some significant advantages over humans'.

[Skipjack]

I'm no specialist, but I doubt that consumer grade flight sim AI is comparable to military grade. Consumer PCs have to handle everything else (graphics, physics) in the same resource budget as AI, so that's one big limitation already. The military also doesn't have the cramped funding of video game developers

True, but from what I understand the military has to trade some performance for relyability, shielding, heat management etc.
So I think it will even out in the end. Also, where do the military programmers learn their stuff from and where do they come from?
I would say a lot of them have a game dev background and even if it is just some garage game AI- programming. There are plenty of papers out there to learn from.

Also I would not forget that UAVs are piloted usually. The AI would only take over, if for some reason the connection to the base (and the real humand pilots sitting at the remote control there) was lost.
I think that this balances the odds further towards the UAV.

What I've yet to see myself is long range combat AI in modern setting. You'd expect that the much simpler (relatively) mechanics at work at long range will be sooner conquered by UAV AI than close quarters dogfighting.

That is exactly what I see too. It is more long range fighting nowadays and less dogfighting. For that stealth and a small radar and IR profile are way more important than dogfighting skills.
And then, if you have a ratio of ten to one, I think the odds are much in favor of the smaller UAVs, at least even.

[Betruger]

True, but from what I understand the military has to trade some performance for relyability, shielding, heat management etc.
So I think it will even out in the end.

So the devil's in the details for this one. Rules of thumb aren't conclusive. It comes down to comparing exact numbers in context.

Also, where do the military programmers learn their stuff from and where do they come from?
I would say a lot of them have a game dev background and even if it is just some garage game AI- programming. There are plenty of papers out there to learn from.

Yes but this is also dependent on the bottom line.. The substrate the AI will run on and be constrained by. I personally (not a specialist, but that's the impression I get) doubt the hardware on a UAV is as low-end and unadapted as home/gaming PCs.
I can see it going either way WRT employed programmers. I do reckon a military UAV programmer has more leg room than someone writing for a video game. Also a lot more is on the line. The military has more leverage in attracting employees than the video game industry, I reckon.

Also I would not forget that UAVs are piloted usually. The AI would only take over, if for some reason the connection to the base (and the real humand pilots sitting at the remote control there) was lost.
I think that this balances the odds further towards the UAV.

Is that how it works? A human available to be put in the loop definitely complements AI to beat a human pilot on his own, in his man rated fighter. But again it's not just rules of thumbs.. I expect there's more at work than these few rules of thumb. What about the logistics, the definite handicap of seeing pixels instead of eyeballing things, is there some advantage to fighting for your life (while keeping cool) versus sitting in front of a computer.. And..

What I've yet to see myself is long range combat AI in modern setting. You'd expect that the much simpler (relatively) mechanics at work at long range will be sooner conquered by UAV AI than close quarters dogfighting.

That is exactly what I see too. It is more long range fighting nowadays and less dogfighting. For that stealth and a small radar and IR profile are way more important than dogfighting skills.
And then, if you have a ratio of ten to one, I think the odds are much in favor of the smaller UAVs, at least even.

How does it work out in practice, if manned aircraft were to adapt their loadout to counter UAV swarms, so that you had kill vehicles with multiple sub-vehicles like that anti-missile demonstrator from a couple months back? I suspect that's still not enough to counter UAVs in most situations, but I also don't think I know enough of the real picture to realistically argue any further. Modern air combat, from what I've seen, is much more interconnected than WWII. It's much less dependent on only the man in the cockpit as back then. I suppose UAVs are the latest development in this trend.

[MirariNefas]

Even a single celled organism is intelligent, so we must have missed something.

No.

See, for example:

Artificial general intelligence: an organism and level based position statement

Microscopical Substantiation of Intelligence in Living Cells

My apologies, I thought you were saying that single cells can think. Yes, they have well adapted systems which allow complex interactions and a range of responses to stimuli. Most people reading descriptions of what single-celled organisms do in these papers, would probably not call that "intelligent", so I suggest that you define the word "intelligence" when you use it in such a controversial way.