Gremlins were not "luck charms of many British pilots during World War II" as Darpa states. They were bad luck demons that could kill pilots.

"Gremlins were said to engage in such a myriad of bad behavior as sucking the gas out of tanks through hoses, jamming radio frequencies, mucking up landing gear, blowing dust or sand into fuel pipes or sensitive electrical equipment, cutting wires, removing bolts or screws, tinkering with dials, knobs or switches, jostling controls, slashing wings or tires, poking or pinching gunners or pilots, banging incessantly on the fuselage, breaking windows, and a wide variety of other prankish acts. "

https://mysteriousuniverse.org/2015/07/the-real-gremlins-of-...

There’s a Bugs Bunny animated cartoon in the Merrie Melodies series that depicts a Gremlin like the one you mentioned.

Falling Hare (1943)

https://www.youtube.com/watch?v=ZElJxTCIsJI

Gremlins sound like a Nightmare at 20,000 Feet.

Ummm what's bizarre is that link doesn't mention they probably were widely popularised by Roald Dahl.

Especially since they use pics in the article from his book?

https://en.wikipedia.org/wiki/The_Gremlins

Certainly that's what I think of first, over the film. Especially when talking about aircraft.

In the book they were not bad luck demons, more mischievous. Roald Dahl was a fighter pilot.

I am not an aerospace engineer, but the figure '20 uses' stood out to me as odd: if it's 20 uses, why not just build it to be easily serviceable indefinitely? With no pilot you massively simplify the set of engineering and materials problems, and if it's going to be reusable that many times you're still going to have to perform maintenance, refuel, re-arm, etc. What is so different between '20 uses' and 'decades'?

> What is so different between '20 uses' and 'decades'?

Millions of dollars, that's what. The ability to save costs by designing a limited-use airframe that cuts a few corners compared to a long-lasting airframe and has much lower servicing costs due to its short lifespan, while still being combat-effective within those constraints. Kamikaze aircraft in Japan were much more shoddily built than contemporary conventional Japanese aircraft; they were single use!

I guess a better question is “what’s the difference between 20 and 200 uses?”

If they're expecting a 5% loss rate per mission, then there's no point making them last longer than 20 missions. You can also make things cheaper and more durable if they don't have to be designed to be disassembled.

If there's a 5% loss rate per mission, then there's a 35% chance that it will survive 20 missions.

Adjust the loss rate and the point stands, though.

If you expect to only fly them during missions with that loss rate. I thought planes were used mostly for training. There are not that many active wars being fought right now.

You'd probably train in a simulator, or with more serviceable training drones, I guess.

This is the military, where stuff gets routinely destroyed in normal use. It's probably more cost-effective to engineer for a finite lifespan, because even if they were designed to last longer, they are unlikely to.

Perhaps they're using that to establish a goal line of sorts? There's also the matter of cost. Eventually diffuse wear and tear from the launches and retrievals is going to cause components to fail at a higher frequency. The scales of 'continue repairing' vs 'buy new one' are going to tip towards 'buy new one'. I'm sure the old one will still be useful for something as well -- perhaps as a target drone.

Honestly, to build for 20 uses means using no metal. Or maybe duct tape and tin foil at best.

In aviation, it’s going to be balsa wood, foam core ( ...and I guess some 3D printing with sparce infill, because darpa).

Basically flying garbage in times of war.

Oh boy! Can’t wait! Take me to the next Iraq!

This isn't just a concept. There's already hardware flying.[1]

As a concept, it makes sense. The USAF is being killed by the costs of the F-35. An aircraft that can survive in hostile airspace today is insanely expensive. The USAF just won't have enough of them to go up against anybody serious. Something cheaper is needed.

A cheaper unmanned aircraft won't survive as well. That's OK; if you can get 20 combat missions out of the thing, that's not bad.

[1] https://www.darpa.mil/news-events/2018-05-09

> Something cheaper is needed.

And if you can get it below (or in the ballpark of) the costs of shooting it down, that's another huge win.

The problem with the SUAS concept is energy and scale. You can't have something very small (small enough to fit a dozen in a C-130, so several metres in length at most) which has a powerful turbofan engine and significant range and excellent manoeuvrability and compact stowage.

Recovery especially is rather implausible due to the complex aerodynamics and risk to the people on the mothership. C-130s are fine for a demo I guess, but are particularly terrible tactically, as the adversary SUAS systems would easily be able to take them out.

> Recovery especially is rather implausible due to the complex aerodynamics and risk to the people on the mothership.

Why not recover the sUAs with a big UAV? You can do all sorts of crazy things re: vehicle design, if you don't need people inside.

For example, have each sUA just dive toward the UAV from above, decelerate to a zero/zero intercept, and cut propulsion, such that it goes ballistic falling past the loading bay; and then have the UAV target the sUA with a net out of its loading bay, catch it, and reel it in. Like a skyhook that assumes an uncooperative peer.

Or, have the sUAs throw themselves at random points on the the UAV's (very thick and solid and ferromagnetic) hull, and lock onto it magnetically, and then have the UAV pluck the sUAs off with external cranes and bring the sUAs into the loading bay. Like a monkey plucking ticks off of itself and eating them.

> Recovery especially is rather implausible due to the complex aerodynamics and risk to the people on the mothership.

Fly straight up, cut engines, fall ballistic. Maybe even add a parachute. The US recovered spy satellite footage in the 60s this way, with payload falling from space and being caught in-flight. So it should be workable today as well.

Technology is much more advanced nowadays, why not use that? I would envision a robotic arm that automatically collects UAVs when they approach sufficiently close to the mothership. The arm can be rather large and can be extended through rear loading bay.

> Technology is much more advanced nowadays, why not use that?

Is it though? I mean, it ostensibly is, at least on paper. Not so much in practice, though.

Maybe there is a cycle of expansion and contraction? In the 50s-80s we've pushed tech across all domains as far as we could, doing miracles. Then it got all scaled back in an attempt to optimize the most common use cases to be cheaper. If so, I can't wait for the next expansion phase.

That was a film cannister, not hundreds of kilos of carbon fiber aircraft (and weapons).

It you look at the picture in the article, UAV wings appear to be foldable, so many of them can be stored in smaller space.

>You can't have something very small (small enough to fit a dozen in a C-130, so several metres in length at most) which has a powerful turbofan engine and significant range and excellent manoeuvrability and compact stowage.

Why can't you? Wouldn't it just be, essentially, a reusable Tomahawk cruise missile?

I did think of tomahawk, but it doesn't have great aerodynamics for long endurance or speed. Certainly not very fast. Of course, TLAM accepts many constraints so it can be launched from a submarine.

Could the TLAM components be used in a better folding airframe? I guess so. It always seemed dumb to blow up a $2m airframe to delivery 450kg of HE. And of course it is, TLAM would be cheaper if it wasn't engineered for delivering nuclear warheads. Maybe just build $200k Tomahawk-lite that don't need air recovery, just ditch in the sea.

How reusable Tomahawk's engine is? That is, for a reusable engine of this size, what would be flight time until a serious maintenance is required?

I just figured recovery could be sky hook or similar, have a cable out there that the drone snags and then it gets pulled in. I cannot imagine the hell of trying to fly in the turbulence behind a C130

We tried that in the 20's with airship carriers. It was hard even at airship and biplane speeds.

And in the 30s, 40s, 50s. It’s an enduring dream.

One of my favorite parasite fighters https://en.m.wikipedia.org/wiki/McDonnell_XF-85_Goblin

The Fulton recovery system only had one fatality in its uses. And recovering a UAV is more durable than a squishy meat bag human.

I don't understand why you'd want to tell everyone how your weapons work. It may be a cultural thing, as I am not American. What is the reasoning behind it?

Programs like this won't stay secret anyway - somebody is going to see a test flight in the distance, or an employee at a contractor will leak.

Announcing it gives the military a chance to control the narrative. If a photographer snaps a test flight and it appears on Gizmodo with the headline "Military testing secret drone swarms to kill us all", there'll be a massive public outcry, a huge negative PR cycle, possible a Congressional inquiry, and in the course of doing damage control the military will have to reveal more secrets than if they just announced the project in the first place.

Americans are not shy about inventing details when they have none, and also have a tendency not to trust either their government or the press.

Advertising, given huge income from arms sales. Also perhaps generating fear and awe.

It's definitely a form of advertising. US companies often publish vast amounts of public R&D, and universities routinely do it. The US military sometimes behaves that way for the exact same reason.

When you can advertise that you're cutting edge, it helps attract the best minds. That's dramatically more valuable than worrying about eg China or Russia stealing the idea to Gremlins. Those ideas and technologies get out regardless, you can't protect them very long. Competitors and adversaries are going to catch up with a given piece of technology, even if they have to figure out how you did it the hard way. By recruiting the best minds to work on problems, you can ensure that you remain a lap ahead of the competitors that will steal or copy from you regardless. If something is very sensitive, then perhaps you can buy a bit more time by keeping it strictly secret, it's still going to get out. If your tech is good enough, powerful countries will do anything to get at it.

Good point! Advertising for "the best minds". One of my math professors was almost an NSA recruiter ;) I was never good enough to get the pitch, however :( I found engineering math to be much more useful.

vis. Google's open source offerings.

And even Microsoft's ;)

During the Cold War we had two government bunkers that only a handful of people knew about in my country.

30 years after the Cold War when Russia opened their soviet archives, it turned out that they were never secret at all.

Carrier has arrived.

<Protoss noises intensify>

The part I found most enjoyable was the fact that DARPA says the British love Gremlins, while Americans made them into horror icons.

"and retrieve them in mid-air" sounds pretty challenging.

Cruise missiles? Drones? With "AI"? Whatever, so damn predictable. But hey, when you gotta kill, you gotta kill. Might as well do it efficiently, with as little operator risk as possible.

But it does suck, when you're the target. And low operator risk makes wars a lot easier to sell to frightened masses. So likely, there'll be more targets, for less justifiable reasons. Resource wars, for example.

You can definitely argue that drones are a proliferation risk, and also risk conflict escalation. Because there is no pilot on board, A is willing to action riskier/provocative missions against B. B knows it is a drone, so it can shoot it down without loss of human life, and possibly analyse the fragments. A knows B will do this, so uses disposables to probe air defence capabilities. B sends its own drones to probe A's launch site/platform. Et cetera.

So it allows idiot governments to seriously escalate conflicts. I can totally see this in regional conflicts.

The disposable nature with low local expertise requirements might increase sales to poor countries too. While they might not be willing to risk their air superiority totems (MiGs of F-16s) in a border dispute, drones are qualitatively different.

You don't need to argue hypotheticals. This has been the US pattern for decades. Even in Iraq, the US casualty rate was far lower than in Vietnam.

It's easy for politicians to dispatch robots, no danger of service members being killed. Look at the proliferation of questionable targets and results that cruise missiles have given politicians when they need to make empty threats.

On the other hand, it's the military who have the money. Jet fighters and bombers came before jet airliners. So, maybe DOD's involvement will bring the Singularity closer.

I'm not sure that angry AI is something to look forward to. Unless I'm the angry AI, anyway ;)