Are bombers obsolete?

[Laughing Mechanicus]

A question related specifically to the UK's nuclear deterrant. We are looking at replacing Trident like for like with a new ballistic missile system, but is that such a good idea? By the time such a system is designed, delayed and eventually deployed it seems like we will be looking at a world with effective ABM systems being deployed by many nations. Bombers are also a problem for the UK however - as we would only get 4 minutes of warning of missiles from Russia, and little more from those launched from many other countries; any bombers we built would be terribly vulnerable on the ground. Are there ways around this? It strikes me from that video of the B-52s that the main bottleneck in the speed of launching bombers is the runway - how feasible would it be to design an airstrip capable of handling many bombers taking off at once? Basically an extremely wide runway where every bomber just acclerates from its parking position, or something of the sort.

Now on another topic: a slight digression from the main thrust of the thread, but how would the proliferation of powerful solid state air defence lasers affect the practicality of hyper sonic high altitude bombers? Or, for that matter, the suggested alternative cruise missiles or plain old ballisitc missiles?

It seems like the technology to allow for very powerful solid state lasers is coming along, if slowly - and in the air defence role the technological challenges they present are reduced; they don't need to be very mobile or worry too much about power requirements. Now please correct me if I am wrong here, but it seems that with a laser if you can detect an object you can hit it - all the speed and altitude in the world (literally) won't help you; would this kind of weapon not cause a huge problem for all the current types of strategic weapon delivery systems? Might this perhaps force an emphasis on stealth rather than speed and altitude? Is it even possible to be stealthy enough to evade defences while remaining an effective delivery system in the face of the kind of radar technology that exists today?

[Stuart]

If you have hypersonic small vehicles (missiles, etc.) which is obviously the way things went since Kholod tests were first run, why would you need humans though? If the hypersonic two-stage missile is a combination of ICBM/cm tech to make a super-missile undefeatable by ABM, what's the point of putting humans on it? Too much cargo, too much stress on frame... expensive.

Expensive yes, but worth it. There is always a swinging balance between offense and defense; when a new weapons sytem evolves, so do counters to it. We can see this clearly with the development of the bomber and the related air defense systems, with the development of the ballistic missile and its related missile defense systems (remember we've had perfectly viable and effective anti-missile systems since the early 1960s; the only reason why the ICBM held on to its fabricated aura of invulnerability so long was that the counters to it were effectively prevented by political treaty). Now, the greatest weakness of the ballistic missile was that it can't fight back. Manned bombers can, that's why air defense systems are much, much harder to design than missile defense systems and have a much lower expectation of success than missile defense systems. The term fighting back here includes any means that prevent the countermeasures from negating the asset in question.

At our present level of technology, we can't automate or mechanize fighting back. We've tried, tried hard, but we can't. We can do demonstrations in laboratories which give glimmers of hope in that direction but in the hard, cold world of battlefield reality we can't. Nor, to be honest, is there any real prospect of us being able to do so. We've got drones that can fire missiles at targets with high success rates but if those drones come under attack by a manned system, they're gone. We tried using drones armed with air-to-air missiles and they died. Now, if we build our hypersonic cruise missile boosted with am ICBM booster and don't put a crew in it, what we have is essentially a maneuverable target that travels at very high speed, very high up. It can evade to a limited extent but it can't really defend itself. Also, the elements of abortability and retargetability are gone. To be operationally acceptable, those require, absolutely require, a crew on board. It would only be a quetsion of time before the unmanned hypersonic would be brought within reach of defensive systems. Push me and I'd guess that time would be less than a decade. Strategic bombing really started in 1938 and air defense systems were up and running by 1945. ICBMs first became practical in around 1955 and ballistic missile defense was technically a done deal in 1962. Seven years seems to be the time lag.

Essentially, the defense against a hypersonic cruise missile is simply the ballistic missile defense writ large and suffering from the same basic problem. The inbound can't defend itself so once it is within reach of the defenses, the probability of it being killed gets to be very high. Essentially, if we can hit it, we will hit it.

If we put a crew on the system, we change the paradigm from an anti-missile system to an air defense system. Immediately, the probability of success drops way down because the inbound can fight back. It can use evasive maneuvers thought up at the last second (and not in the pre-planned book), it can use countermeasures effectively, it can use anti-interceptor tactics and equipment. In effect, it can fight its way through the defenses, something no unmanned system can do and probably will ever be able to do within a reasonable time-frame. That pushes the chance of success way up and returns us to the bomber vs ICBM situation we have now.

[Sarevok]

However will we realistically see any hypersonic warplanes within a 20-25 year time frame ? Air power seem to have taken a HUGE detour from the high and fast path of the 50s and 60s. Stealth, information warfare, drones etc seem to be the buzzwords of the day. Given the circumstances it seems very unlikely a hypersonic bomber like Stuart envisions will ever take flight in foreseeable future. I am nowhere near informed enough to make judgment about whether the concept is practical. All I am saying is we live in a world where B-52s are going to be flying 30 years from now and backbone of US air power is going to be the problem ridden F-35. Its going to take a revolution in pentagon to order development of a something radical like a hypersonic bomber in today's world.

[adam_grif]

Stuart, I find your lack of faith disturbing. When the machines take over, I'd say "I told you so", except there wouldn't be any point since you'd be the one whose skull was being crushed underneath the glorious metal endoskeleton of our completely-automated robotic overlords.

[Shroom Man 777]

Stuart, I find your lack of faith disturbing. When the machines take over, I'd say "I told you so", except there wouldn't be any point since you'd be the one whose skull was being crushed underneath the glorious metal endoskeleton of our completely-automated robotic overlords.

There was a nuclear war. A few years from now, all this, this whole place, everything, it's gone. Just gone. There were survivors. Here, there. Nobody even knew who started it. It was the machines, Shroom.

Defense network computers. New... powerful... hooked into everything, trusted to run it all. They say it got smart, a new order of intelligence. Then it saw all people as a threat, not just the ones on the other side. Decided our fate in a microsecond: extermination.

I grew up after. In the ruins... starving... hiding from H-K's.

Hunter-Killers. Aerial and ground patrol machines built in automated factories. Most of us were rounded up, put in camps for orderly disposal.

Some of us were kept alive... to work... loading bodies into dumpsters and incinerators. The disposal units ran night and day. We were that close to going out forever.

But there was one man who taught us to fight, to storm the wire of the camps, to smash those metal motherfuckers into junk. He turned it around. He brought us back from the brink. His name is Stuart. Stuart Slade. Your son, Shroom, your unborn son.

[Sarevok]

Stuart, I find your lack of faith disturbing. When the machines take over, I'd say "I told you so", except there wouldn't be any point since you'd be the one whose skull was being crushed underneath the glorious metal endoskeleton of our completely-automated robotic overlords.

Computers and their decision making skills have not advanced enough yet to make them competitive pilots for very expensive warplanes. Its one thing to spam cheap drones and another to trust software to fly a hundred million dollar super advanced aircraft. The former can be easily replaced if the software glitches the latter is too precious to be risked in same manner.

[Starglider]

A question related specifically to the UK's nuclear deterrant. We are looking at replacing Trident like for like with a new ballistic missile system, but is that such a good idea?

I get the impression that there's a limit to how much we can get out of Stuart without (a) paying his extortionate consulting rates and (b) getting the entire board security clearances.

Bombers are also a problem for the UK however - as we would only get 4 minutes of warning of missiles from Russia, and little more from those launched from many other countries; any bombers we built would be terribly vulnerable on the ground. Are there ways around this?

Sure. We could base all the nuclear bombers in the Falkland Isles. It would be great for the local economy and send Argentina an appropriate message about the consequences of trying another invasion.

Basically an extremely wide runway where every bomber just acclerates from its parking position, or something of the sort.

The limiting factor for serial takeoffs is wake turbulence and that stuff has quite a wide area of effect. You'd be better off building a series of parallel runways, or better, just a large number of dispersal airfields around the main base. Of course the UK doesn't have the land for this any more since all the old RAF bases got turned into housing developments and shopping malls.

[Starglider]

We tried using drones armed with air-to-air missiles and they died.

Well of course, if they don't have fighter-grade radar, ecm, stealth, speed etc, they aren't going to win any engagements with actual fighters.

It can use evasive maneuvers thought up at the last second (and not in the pre-planned book), it can use countermeasures effectively, it can use anti-interceptor tactics and equipment. In effect, it can fight its way through the defenses, something no unmanned system can do and probably will ever be able to do within a reasonable time-frame.

So you're saying that we can expect to see unmanned nuclear bombers approximately ten years after someone writes a genuinely competent enemy AI for a real time strategy game?

Where do I send the demo?

[Stuart]

Also, by the Arihant possibly being an SSGN, do you reckon this means that the PJ-10 Brahmos will be on it?

That's one candidate; there are a couple of others. A lot will depend on the internal architecture and the Indians are playing that very close to the chest. The current belief is that the Arihand is based on the Project 670M (Charlie II) and if she retains the oversized inclined bow tubes, then PJ-10 is a good candidate. If she's been revised to include vertical launch tubes, then there are a number of cruise and ballistic missiles that could be on her.

Wasn't stealth originally the "New speed and altitude?" I thought all those hypersonic aircraft projects were mothballed because it was far easier to make a swarm of hypersonic SAM's that would punch slots in your high and fast bombers, so they changed tack from 'outrun the enemy' to 'hide from the enemy'

That's how the common story goes; however there is much more to it than that. Back then, the effectiveness of surface to air missiles was vastly overstated; the generation of missiles that could seriously endanger a B-70 Valkyrie never existed and still do not exist (and yes, that does include the S-400). The best Pk that even the most advanced missile systems could score against a B-70 is around .15. Nasty but in a nuclear environment acceptable. Once the hypersonics were gone, the race changed from "procuring the best bet" to "find something that might survive". A lot of us had doubts about "stealth technology" right from the start. It only works at night so its use is essentially conceding daylight to the enemy.

Are you saying the ballance of power ping-pong's between slow and stealthy and high and fast as your opponent optimizes defences against one or the other? Or just that hypersonic bombers would have been better in the long run from the beginning?

The latter; cancelling the B-70 was the worst strategic blunder the US made in an era that was studded with catastrophic blunders.

I'm familliar with the US experience of 'interceptor' fighters designed to dash around lobbing long range AA missiles getting caught out in dogfights, but SAM's vs hypersonic bombers too? Or Hypersonic interceptor fighters with missiles vs hypersonic bombers? I'm just working from the assumption that if you can build hypersonic bombers, someone else can make a enough smaller, cheaper fighters with hypersonic missiles to ruin your day.

There's a big problem, up high what matters is wing area, area of control surfaces and engine power. Highup the big bombers have it all over fighters of nominally superior performance. In the early 1950s there was a classic case of a F-86 that tried to intercept a B-36. There was a dogfight which ended with the B-36 getting on to the F-86's tail and chasing it around the sky. I am reliably informed that many years later the event was commemorated by an extreme altitude rematch which ended with a B-52 getting on to the tail of an F-16 and chasing it around the sky. In tests between a MiG-15bis and a B-36, the B-36 won hands-down; it was cruising around quite comfortably while the MiG was in constant danger of stalling out (and did so when it tried to fire its guns). For the same reasons, a B-70 is faster and more agile than an SR-71 at the same altitude (the SR-71 has 5,000 feet less ceiling but . . . . ). Nobody has ever seriously threatened an SR-71 despite major serious attempts to do so. So, the B-70 would have been less vulnerable still.

What about depressed trajectory ICBM shots Stu? How do they stack up against a hypersonic missile shot?

They don't is the honest answer. They give no benefit at all where warning time is concerned. At the same time they suffer major trade-offs in range, payload and accuracy. Even worse, the depressed trajectory puts them into the engagement envelope of a lot of surface-to-air weapons (and even a few fighters). Thoroughly bad news all around, that's why they were abandoned.

Won't the virtual attrition from a booster launched manned hypersonic bomber essentially be the same as a depressed trajectory ICBM anyway?

No, because the bomber can carry multiple weapons with extremely flexible targeting, something precluded by the demands of a depressed trajectory.

There's a late 1980s film, By Dawn's Early Light which is set in part aboard a B-52 flying a nuclear penetration mission (I think that's what it was called) during an accidental ramp-up to war between the USA and USSR. In the course of the story they dealt with penetrating air defenses and fighter attack. From what I remember it was a very convincing depiction (but I have to leave it to people with real expertise, to say how technically accurate it may have been).

If one ignores the dumb sub-plots, it was a good film. Accurate? Not very but better than most. Two very good films to watch are "A Gathering of Eagles" and "Strategic Air Command". The first stars B-52s and the second B-36s but I believe some humans might have been in them as well.

(snipped explanation).

Exactly right. A very neat, well-written summary of the situation (both posts).

. We are looking at replacing Trident like for like with a new ballistic missile system, but is that such a good idea? By the time such a system is designed, delayed and eventually deployed it seems like we will be looking at a world with effective ABM systems being deployed by many nations.

My point exactly and a lot of people in Whitehall hate me for saying just that. SLBMs aren't as obsolete as ICBMs but their day is ending. By the time the UK gets its Vanguard replacement, (pessimistically 2030) ABM systems will be commonplace, tried and effective.

Bombers are also a problem for the UK however - as we would only get 4 minutes of warning of missiles from Russia, and little more from those launched from many other countries; any bombers we built would be terribly vulnerable on the ground.

This is why the UK went away from any sort of surface-based deterrent, missile or bomber. There just isn't time to do anything. As an example Bloodhound was never given a nuclear warhead because the increase in reaction time that would cause meant that none of them would ever be launched.

Are there ways around this? It strikes me from that video of the B-52s that the main bottleneck in the speed of launching bombers is the runway - how feasible would it be to design an airstrip capable of handling many bombers taking off at once? Basically an extremely wide runway where every bomber just acclerates from its parking position, or something of the sort.

More runways is the easy answer. What determines MITO is turbulence from the preceeding aircraft. Note in that video how one aircraft was rolling badly as it picked up the wash from the one in front. That's at an average of 39 seconds. At 15 seconds the pilots are working like madmen to keep the birds under control (one reason why 15-second MITOs aren't done very often. Too risky). We do this by dispersing the birds. But, four minutes just isn't enough.

However will we realistically see any hypersonic warplanes within a 20-25 year time frame ? Air power seem to have taken a HUGE detour from the high and fast path of the 50s and 60s. Stealth, information warfare, drones etc seem to be the buzzwords of the day. Given the circumstances it seems very unlikely a hypersonic bomber like Stuart envisions will ever take flight in foreseeable future. I am nowhere near informed enough to make judgment about whether the concept is practical. All I am saying is we live in a world where B-52s are going to be flying 30 years from now and backbone of US air power is going to be the problem ridden F-35. Its going to take a revolution in pentagon to order development of a something radical like a hypersonic bomber in today's world.

There's more work than you might think going on in this area; the trouble is that we have taken such a large detour from the True Faith that regaining the knowledge is hard. But, the work is there and being done. So, it might not be as impossible as you think. What form teh aircraft will take is an entirely different matter though.

Stuart, I find your lack of faith disturbing. When the machines take over, I'd say "I told you so", except there wouldn't be any point since you'd be the one whose skull was being crushed underneath the glorious metal endoskeleton of our completely-automated robotic overlords.

What makes you think I won't be working for the completely automated robotic overlords by then?

Sure. We could base all the nuclear bombers in the Falkland Isles. It would be great for the local economy and send Argentina an appropriate message about the consequences of trying another invasion

Don't laugh. It's been suggested quite seriously (for ICBMs of course). Also South Georgia by the way. The primary target then would have been the USSR of course and it would have spread their defenses out very thin - essentially turned them through 90 degrees. Apparently, there were legal problems amongst many other things. It would have been cheaper than building SSBNs though,

Well of course, if they don't have fighter-grade radar, ecm, stealth, speed etc, they aren't going to win any engagements with actual fighters.

That wasn't it. It was the datalinks couldn't handle the volume of information. They slowed down so much that the AAM Predator more or less had a nervous breakdown and gave up.

So you're saying that we can expect to see unmanned nuclear bombers approximately ten years after someone writes a genuinely competent enemy AI for a real time strategy game? Where do I send the demo

Nobody is close yet. A lot of people are working on such things but the problems are systemic and fundamental. After all, most humans can't absorb the amount of data going through the system (which is why 5 percent of pilots score 65 percent of the kills and most fighter pilots never see the aircraft that killed them). There's a lot of undefinable "we don't know what they are but they're critical" bits involved. I'd say a closer guess would be that we'd get an AI equivalent to a pilot about ten or twenty years after we actually discover what it is about ace fighter pilots that keeps them alive. The truth is, we just don't know. Lot of theories, lot of people claiming to have "the secret" but the truth is we just don't know

[Simon_Jester]

That's how the common story goes; however there is much more to it than that. Back then, the effectiveness of surface to air missiles was vastly overstated; the generation of missiles that could seriously endanger a B-70 Valkyrie never existed and still do not exist (and yes, that does include the S-400).

The question that troubles me is: how much of that is due to the lack of a need? Once everyone concluded that nobody was going to be building Mach 3 nuclear bombers any time soon, I'd expect the incentive to develop countermeasures to drop off.

Even if you have the basic technology to build the right kind of air defense system, there's no point in developing applications specific to stopping high speed bombers that don't exist. Not when there are other kinds of defense (like ABM) that protect you against a threat that really exists (like ICBMs). So can we use the absence of counters to the B-70 today as proof that no such counters would exist had it been deployed?

There's a big problem, up high what matters is wing area, area of control surfaces and engine power. Highup the big bombers have it all over fighters of nominally superior performance.

Hmm. Sea Skimmer suggested that hypersonic interceptors would have to be about the size of a bomber; could we see interceptors that look structurally like bombers but with an air to air payload?

Though as Skimmer said, that makes interceptions hard because you won't be able to count on outnumbering the attacker.

[Stuart]

The question that troubles me is: how much of that is due to the lack of a need? Once everyone concluded that nobody was going to be building Mach 3 nuclear bombers any time soon, I'd expect the incentive to develop countermeasures to drop off. Even if you have the basic technology to build the right kind of air defense system, there's no point in developing applications specific to stopping high speed bombers that don't exist. Not when there are other kinds of defense (like ABM) that protect you against a threat that really exists (like ICBMs). So can we use the absence of counters to the B-70 today as proof that no such counters would exist had it been deployed?

This is an extremely good and very important point that a lot of alternate history discussions forget. No change takes place in isolation; they are all part of an interactive environment where changing one part affects all the others, often in unpredictable ways. In the following "missile" refers to surface-to-air missiles unless otherwise stated.

In the case of missiles, it is indeed the case that had the move away from high-flying, very fast bombers not taken place, missile development would have followed a quite different path. However, having said that, the problems that afflict aircraft (using the term in an extremely generic sense) at very high altitudes are hard to resolve. We come back to the basic requirements I pointed out earlier. Up high, where the air is so thin it barely exists, what matters are large wings, large control surfaces and lots of engine power.

The very thin air means low drag and this means high speed. More importantly, the thinner the air, the less it heats the airframe and so the faster the aircraft can go before it hits its heat limits. However, it also means that the wings generate less lift and the control become much less effective. There's another problem. That thin air means that the engines develop progressively less power and the aircraft's maximum speed starts to fall. This is what kills fighters high up when facing bombers. The bomber's big wings generate lots of lift; the small wings on a fighter generate very little. The bomber's multiple large engines still provide power, the one or two engines on a fighter mean that they are power-critical. With a fighter, that means the difference between maximum speed and stalling speed is very slight. Performing even simple maneuvers like a gentle turn or firing guns and/or missiles reduced that margin to less than zero; the fighter stalls out and drops like the proverbial stone. Sometimes the pilot regains control lower down in the thicker air, sometimes they don't and the aircraft goes subterranean.

Missiles are an extreme form of fighter. They have tiny wings, even tinier control surfaces and very often they don’t have any power at all. A hefty proportion of the world's missiles are boosted up to maximum speed early in their flight path and then glide the rest of the way. There are reasons for that. This configuration is adopted because most missiles have to engage targets flying relatively low down where they need agility so they are designed to minimize drag. Up at extremely high altitudes, they can't turn worth a damn and are effectively unguided.

Aha! I hear you say. Then let's design a missile that can work at high altitudes. We give it big wings, big control surfaces and lots of power. That's good, we can design missiles like that. There is just a tiny problem. Being SAMs, those missiles are fired from the ground (of course) where the air is thick and heavy. Pushing those big wings and tail surfaces through that thick, heavy air takes a lot of power and burns a lot of fuel. Given a normal-sized SAM, it takes so much fuel that the missile can't get up very high before it runs out. So, the only way around that problem is to build a much bigger missile that contains more fuel. That is possible but it means we end up with a very large, heavy and immobile (also very expensive) weapon.

Obviously, this causes people quite a lot of grief because missiles of that sort would be hard to afford in adequate numbers. They're also pretty good targets in their own right, both on the ground and in the air. They're also pretty much useless when used against anything flying at low altitudes. This means that their batteries have to be protected against attack by being ringed with other SAMs that protect the high-altitude missiles from attacks that come in low – like anti-radar missiles fired by the inbound bombers.

One way around that problem was to use thrust-vectored controls for aerodynamic direction instead of conventional control surfaces. This solves some of the problems but adds wholly new ones. Now, thrust vectoring means swinging the engine nozzle in order to steer the missile. That means the engine must be working all the way to the target. Only, most missile engines don't do that, the reason being that they are too small to carry the fuel needed. What they do is have a limited amount of very energetic fuel that burns off fast (hence energetic) so they accelerate quickly. Thrust vectoring means that this approach to rocketry doesn't work too well.

It is possible to devise a missile that counters these problems. One possibility would be to have a big carrier missile that carried a group of small (AIM-120-sized) missiles inside a streamlined casing. This is blasted off, gets to high altitude and then fires its salvo of small missiles at the target. That's feasible but it (and all the other potential solutions) are very expensive and could only be deployed in limited numbers. An aside here, one of the beauties of ABM is that all the serious money gets spent on reusable things that are on the ground – sensors, command systems etc. The expendable missiles themselves are pretty cheap. So, once the system is established, adding extra missiles to cover more targets is a relatively minor expense. With this putative SAM system, much of the expense of the system is in the missiles so adding extra missiles to cover more targets is a seriously serious expense.

So, the problem is really one of physics; the demands of firing missiles at high-altitude, very fast and maneuverable targets requires a mass of contradictory characteristics and one can't gain some of them without trading off others. Unfortunately, trading off any of them means that the systems ability to do the job is compromised.

[Starglider]

Well of course, if they don't have fighter-grade radar, ecm, stealth, speed etc, they aren't going to win any engagements with actual fighters.

That wasn't it. It was the datalinks couldn't handle the volume of information. They slowed down so much that the AAM Predator more or less had a nervous breakdown and gave up.

If they were using datalinks designed for the original air-to-ground mission that would be understandable, since I don't imagine the Predator (a relatively low-cost platform) was designed with a huge amount of spare link capacity. Can you give out any actual figures for this? e.g. I've seen estimates of 90 Mbits for F-22 aggregate datalink capacity, but I don't know the range dropoff or the spectral efficiency, so it's hard to compare with current COTS technology. I've also seen estimates for the aggregate programmable computing power of the avionics racks at 10 gigaFLOPs, which is about 1% of a contemporary gaming PC.

That said radio link capability has no bearing on the capabilities of on-board, fully-automated decision making. For the hypersonic nuclear bomber case, you already ruled out operating them via datalink for the same reasons that cruise missiles aren't in-flight abortable/retargettable (plus there's the added problem of them going defenceless if the relevant comms infrastructure or control stations are destroyed).

There's a lot of undefinable "we don't know what they are but they're critical" bits involved. I'd say a closer guess would be that we'd get an AI equivalent to a pilot about ten or twenty years after we actually discover what it is about ace fighter pilots that keeps them alive.

Well, that gets into black-box vs white-box approaches to AI, I won't derail the thread.

[K. A. Pital]

If you can fire cheap ABM missiles from ground to space, what is the problem, then? Just make them a little less energetic and put a second-stage to fly high in the air. Perhaps not as simple as that, but certainly possible.

Also, if Mach 3 targets are so un-engageable, how come the USA downed target-practice BOMARC drones?

[Stuart]

If you can fire cheap ABM missiles from ground to space, what is the problem, then? Just make them a little less energetic and put a second-stage to fly high in the air. Perhaps not as simple as that, but certainly possible.

The problem is that the ABM is designed to engage a very fast target coming in along a predictable course. Essentially, it doesn't maneuver and flies in a straight line itself. To make the system you suggest work (don't get me wrong, it's not implausible) one would need to fire the missile so that booster puts the second stage within an acquisition basket of the target. That's not easy and the Pk would be low. It's the same old story, very high speed and high altitude reduce the time for decision making to a minimum and increase the probability circle the aircraft might occupy to an impractical extent.

Also, if Mach 3 targets are so un-engageable, how come the USA downed target-practice BOMARC drones?

Because the Bomarc drones flew at under Mach 2.5 and around 50,000 feet. The Bomarc drones were good simulators for aircraft like the B-58 but they didn't come close to simulating the B-70. Also, the Bomarcs didn't fight back. They were, though, very good simulators for certain large anti-ship missiles.

[adam_grif]

Well, that gets into black-box vs white-box approaches to AI, I won't derail the thread.

Maybe he can clarify, but reading his post about it almost makes him sound superstitious about piloting skills. Obviously, there will be optimal responses to things that can easily be determined, and some that will be difficult to determine. But I don't see any reason why it's particularly beyond possibility, even in the near-term future.

[Starglider]

Maybe he can clarify, but reading his post about it almost makes him sound superstitious about piloting skills.

Not really. The kind of intelligence you use in a fight is a mass of layered heuristics, pattern recognisers and pattern generators. The obvious reductionist approach is to try and isolate each response under controlled conditions, hypothesis the mechanisms, then see if those can predict behavior in similar scenarios. Or build computer programs based on the theorised mechanisms and see how they perform. Unfortunately the basic techniques used in psychology to do this are very weak. We've been able to reliably verify quite a lot of cognitive biases and general decision heuristics, but that only works for relatively simple rules operating over relatively long timeframes. At the other end of the scale there is a mass of work on how perceptual works and instantaneous reactions. Air to air combat is one of those nasty complicated problems that combines decision processes operating on multiple different timeframes, in different domains, in real time. Certainly the classic knowledge engineering approach of asking pilots what process they use to make decisions won't work very well; they can give you some textbook maneuvers, but not how to use them.

So in short I would say that Stuart isn't being superstitious, it's just that the problem is resistant to white-box analysis. It doesn't help that some of the key tools (e.g. Bayesian networks and various data mining techniques) were not available until quite recently, all the 80s effort had to make do with horribly inflexible rule-based systems.

But I don't see any reason why it's particularly beyond possibility, even in the near-term future.

If the problem isn't really decomposable, then white box analysis won't help until you have a complete understanding of the brain. That will take a while, to say the least.

The alternative is a black box approach, or should I say recreating the capability from scratch without reference to the human model (all the popular techniques to do this are essentially opaque but some of the less popular alternatives aren't necessarily so). I am personally much more optimistic about this happening in a reasonable time frame.

[adam_grif]

If the problem isn't really decomposable, then white box analysis won't help until you have a complete understanding of the brain. That will take a while, to say the least.

The alternative is a black box approach, or should I say recreating the capability from scratch without reference to the human model (all the popular techniques to do this are essentially opaque but some of the less popular alternatives aren't necessarily so). I am personally much more optimistic about this happening in a reasonable time frame.

I don't even know why you would bring up a white box approach here since I never talked about that

In terms of building automated piloting systems, flying point to point is already in use. Building computers to handle maneuvering that could shred a real pilot needs an understanding of the physics involved, but is hardly insurmountable. The most difficult part seems like "identifying hostile targets" and "realizing when something is shooting at me". But humans don't eyeball these sorts of things, don't they already use automated systems to detect missile locks and identify enemy craft?

Maybe it can't outperform the greatest pilots in the world, but how hard is it to get a robot to fire off a missile when it detects hostiles?

[Starglider]

I don't even know why you would bring up a white box approach here since I never talked about that

That's what Stuart was talking about (unsurprisingly, an analyst would gravitate to that approach) and that's what I assumed you meant by 'determining the optimal responses'. For less challenging applications, we 'determine the optimal respones' by getting someone called a knowledge engineer to grill assorted experts until they have an appropriate set of rules or heuristics.

Maybe it can't outperform the greatest pilots in the world, but how hard is it to get a robot to fire off a missile when it detects hostiles?

Not hard at all, if you don't mind the fact that the enemy can easily spoof it into wasting all its missiles on ECM phantoms or decoys, plus the odd suspicious looking mobile phone mast etc.

[adam_grif]

That's what Stuart was talking about (unsurprisingly, an analyst would gravitate to that approach) and that's what I assumed you meant by 'determining the optimal responses'.

I was thinking more along the lines of a big book of situations that the drone recognizes, like a missile coming up it's tailpipe, and then knowing how it has to move in order to avoid that. And furthermore, knowing that it should avoid that. The trick here is getting it to handle multiple things at the same time, and mixing/matching responses for optimality.

Not hard at all, if you don't mind the fact that the enemy can easily spoof it into wasting all its missiles on ECM phantoms or decoys, plus the odd suspicious looking mobile phone mast etc.

I don't know how modern ECM and decoy systems work, but I'm just going to assume that the human operators who don't get fooled by this stuff (or maybe they do?) aren't just rolling a D20 and going with gut instincts to work out what to do, and that there are things to look out for with the spoofing.

You've raised technical challenges to be overcome, but nothing you've said really indicates that any of this is impossible, or even beyond what could be done in the next few years if such projects actually existed.

[Simon_Jester]

It is possible to devise a missile that counters these problems. One possibility would be to have a big carrier missile that carried a group of small (AIM-120-sized) missiles inside a streamlined casing. This is blasted off, gets to high altitude and then fires its salvo of small missiles at the target. That's feasible but it (and all the other potential solutions) are very expensive and could only be deployed in limited numbers. An aside here, one of the beauties of ABM is that all the serious money gets spent on reusable things that are on the ground – sensors, command systems etc. The expendable missiles themselves are pretty cheap. So, once the system is established, adding extra missiles to cover more targets is a relatively minor expense. With this putative SAM system, much of the expense of the system is in the missiles so adding extra missiles to cover more targets is a seriously serious expense.

So, the problem is really one of physics; the demands of firing missiles at high-altitude, very fast and maneuverable targets requires a mass of contradictory characteristics and one can't gain some of them without trading off others. Unfortunately, trading off any of them means that the systems ability to do the job is compromised.

Thank you. That said:

ABM systems, at the very least, have to be able to physically reach the altitude that high speed bombers fly at, because ICBMs live at those altitudes too (or, in mid-course, much much higher). I'd imagine that a two-stage "canister missile" concept like the one you describe could be based on that, using the ABM-type booster to loft the air to air missiles (such as the AIM-120s you named) into the general vicinity of the target.

The target can evade, of course, the question being how much evasion it can perform without losing speed or being effectively forced to turn back by the air defense net... I'm missing something, though; just wish I knew what.

The problem is that the ABM is designed to engage a very fast target coming in along a predictable course. Essentially, it doesn't maneuver and flies in a straight line itself. To make the system you suggest work (don't get me wrong, it's not implausible) one would need to fire the missile so that booster puts the second stage within an acquisition basket of the target. That's not easy and the Pk would be low. It's the same old story, very high speed and high altitude reduce the time for decision making to a minimum and increase the probability circle the aircraft might occupy to an impractical extent.

OK, more or less the track I was on already. I have neither the data nor the right training to estimate the size of the acquisition basket properly, so at this point I'm stymied.

I'd rather pay for a wall of two stage Solidbooster-AMRAAM SAMs than pay to patch up the crater left by a fusion bomb, but I can certainly see how the cost of the air defense system would go through the roof.

[Shroom Man 777]

Now here's a question: Why don't we use long-loitering big missile bus drones as a defense system instead of ground-based SAMs? Instead of shooting a SAM from the ground, we can have a big loitering drone shoot a long-ranged (hypersonic) AAM instead? Won't such a weapon have all the advantages of a ground-based SAM, but also have the advantages of a mobile aircraft? Imagine turning a B-52 or something into an airborne SAM. It doesn't even have to be a giant drone, it can be manned.

[CaptainChewbacca]

I'm pretty sure they tried that already as Project Pluto.

[Norade]

What about firing the missile from some sort of cannon with enough force to get it to an altitude where it will have the fuel to use vectored thrust to engage the enemy? Something like an electrically fired gun could be rather small and may do the trick.

[Starglider]

I was thinking more along the lines of a big book of situations that the drone recognizes... The trick here is getting it to handle multiple things at the same time, and mixing/matching responses for optimality.

Three problems there; 'recognise' is hard, the 'big book' is implausibly large (there are always special cases you didn't think of) and 'mixing/matching' responses is not a 'trick' so much as a 'horribly difficult problem' ('responses' isn't good enough, you have to have planning). Thus people spending decades on the problem with copious genius and funding and not getting very far.

The 'handling multiple things at the same time' shouldn't be an issue any more though, Stuart's protestations notwithstanding. Once you're into the teraflop realm, that should be enough for all the currently applicable techniques.

I'm just going to assume that the human operators who don't get fooled by this stuff (or maybe they do?) aren't just rolling a D20 and going with gut instincts to work out what to do, and that there are things to look out for with the spoofing.

You'll have to ask Stuart but I get the impression that much of the recognition part is subconscious. I would think ECM / ECCM depends as much on the psychology of the enemy operator and system designer (how are they trying to fool you) as the radar physics.

You've raised technical challenges to be overcome, but nothing you've said really indicates that any of this is impossible, or even beyond what could be done in the next few years if such projects actually existed.

Oh, people have been promising 'it will be done in the next few years' since about 1970. It's pretty clear by now that all the obvious approaches, and most of the not so obvious ones, don't actually work.

[Starglider]

What about firing the missile from some sort of cannon with enough force to get it to an altitude where it will have the fuel to use vectored thrust to engage the enemy? Something like an electrically fired gun could be rather small and may do the trick.

Will that actually save you any money over a huge missile though? A gun that big is going to be fairly expensive; even moreso if you want to reload it quick enough to fire another salvo, but if you skip that mechanism you'll need a barrel for every missile. Plus all the electronics now need to be rated for thousands of g instead of tens of g. I proposed this one myself in an earlier discussion about SAMs here, but on reflection the SRB solution is probably does the same thing cheaper (as seen in the British Bloodhound SAM; ramjet sustainer, 4 x strap-on solids to get it up to altitude quickly).