It's not a matter of startups, or entrenched giants or any such thing, or the Innovator's Dilemma. It's a matter of communications, cost and performance.
A Predator can't fly as high as an F-35 or F-22. It can't maneuver nearly as well either. Can't fly as fast. It can't be refueled in flight, so it's range is lower than the F-35. So you'd have to hope that the missile it carried could compensate for these deficiencies. However, missiles are very finicky, and have launch parameters that help them improve their odds. The military looks at it in terms of PoK (probability of kill). So to be cheap (so you can afford to have swarms) you have to give up performance. This will affect PoK.
Next you have to have robust, un-jammable commlinks. So far, no one has those. That's going to be expensive to develop. Humans can make decisions when the comms are degraded, or non-existent. AI has a long ways to go, and AI is expensive. To remain cheap, you'll have to give that up.
With you napkin/elevator calculations, you would be trading 22 Predators for each F-35. 22 Predators is $88M, and the F-35 will eventually drop close to $100m as it reaches IOC. Still, not a bad bargain if it works.
But now, since you've gone cheap and off the shelf, you've got a slow, unmaneuverable aircraft that can't detect a target (no radar), can't communicate well with it's controllers, and can't climb very high, limiting the range of its missiles.
Then, even if you outfit this Predator swarm with an UberMissile, you'll have blown your cost equation out of the water when this swarm comes up against SAMs, or a squadron of F-16s. The swarm won't even see the F-35s due to frontal aspect stealth, and they'll be able to target the Predators and kill them like baby seals. They'll be able to engage, kill, and disengage before the low performance Predators will be able to react.
I want to write to back you up on this. I've spent most of (not all of) my career in military avionics and aerospace. And this was at small companies. The big companies win the big contracts, but so much is contracted out, and there are plenty of other options. But I digress...
This is a really hard, multi-dimensional optimization problem. The engineers at Lockheed Martin, Rockwell, etc., are not dumb. They are the opposite of dumb. They are rocket scientists, in both the colloquial and literal sense. The same holds true for us at the smaller companies. Heavy STEM educations, a lifetime of engineering and research. I've done a lot of work on next-gen battlefield stuff. It's not easy. It's not. It's control theory, filter theory, AI, computer vision, RF design, and so much more. If you can't play in at least 3-4 of those fields at a pretty high level you aren't invited to the party. I've seen far more professional, intelligent, and pragmatic engineers in that work than what I've seen pass for engineering in Silicon Valley.
You really aren't going to sit in an armchair and design up a system that'll work. OTOH, if you make something that does work, the military will beat a path to your door. Look into SBIRs, for example. I've worked on a lot of them. The door is not closed. The door is closed to half-baked schemes, but if you have an idea for putting some low cost hardware and software together, you can do something with it. I'm being more than a bit facile here; it isn't so hard to get through a Phase I and Phase II SBIR; getting your system deployed is quite a bit harder. Still, there are options. Just be prepared for the fact that they are going to take your product, stick it in a field, and shoot (literal) bullets at it. Ohh, failed the projectile test.... Yes, I take the point about swarms, but war is an extremely hostile environment (no pun intended). And as soon as you have tipped the equation towards "disposable", expect the enemy to change their systems to help you 'dispose' of them as soon as possible. Maybe your drone can operate in the theater today; what about 10 years from now? I honestly don't know, the calculus is anything but clear to me.
There's no arguing that the major aerospace companies are files with smart people. But these companies have historically not been very good at optimizing total operational cost. For a recent example, look at SpaceX. Every design decision they've made is to optimize operational cost instead of performance. The majors have been busy optimizing for performance, and they've ended up with very expensive rocket systems.
It is the same for military procurement.
Something else... If the drones are cheap and "disposable", then maybe upgrades can occur more frequently to keep up with current events. I don't think that trying to design weapon systems to least 40 years is the way forward.
I'm not at all against drones. They are clearly working for us and others; for me to try to claim otherwise would be trivially wrong. There will always be a tension between cheap-and-fast vs expensive-and-precise.
I've spent my career doing this stuff. I'm telling you, it is not trivial. It is being worked on, but you can't just strap a missile onto a drone and call it good.
A piece of context I left out. I don't want to kill people. Yes, despite working on war machines. I ain't tossing out cheap, disposable, optimized for operational cost out there, because that means you are, eventually, killing children and other innocents. No thanks. We aren't talking 'cool robotics here'. We are talking corpses, burned bodies, melted faces, families broken, futures destroyed, blood, pain, horror. There are no words for it. I worked on this stuff because I consider the alternative to be worse, but it's ambiguous. I'll spare you the "A Few Good Men" speech, as I'm sure you get what I mean.
"Move fast and break things" is a great way to build a Facebook. Not so much for war machines, IMO. Feel free to chalk that up to trying to preserve a career, but it's not (you, ansible, didn't say that, someone else did). It's the realization that everyday, when you go to work, you are designing something to cause unbelievable misery, often for political rather than ethical reasons. It's hard to live with. I'm not denying that there is politics and all of that other stuff happening, it certainly does, but it is not the only thing going on.
Sorry, I realize this is a downer for HN, where we are supposed to be chipper, enthusiastic, and supportive, but we are talking about killing people here. It's not a situation for being 'disruptive' or whatever SV meme you might think of (again, you didn't use that term, I'm responding more to the thread at large).
Given all of that, is there room for innovation? Sure. And it is being worked on. I've worked on next-gen autonomous robots and UAVs for war. You may or may not read about it some day. But it ain't easy. Another poster brought up the IUD analogy. Sure, if you blanket the sky with fire you can take down a modern US jet. And kill everything else as 'collateral damage'. I don't want to fight war that way, and I don't think we ever should, except perhaps in extremis. Because at that point we have lost whatever it means to be human.
Sorry, I think about this stuff. A lot.
edit: a big reason for the expensive airframes and other weaponry is because we decided we don't want to carpet bomb and otherwise use non-precision weapons. Spend a million and hit the window where the target is sitting, vs 50K to drop a bunch of bombs that destroy a city block. Precision is often efficacious, to be sure, but we are doing this for moral reasons as well.
There are a lot less civilians in the air than on the ground. As to cost, air to air missiles are not cheap and there not light so sending a mix of real and 'fake' drones is a viable option. At which point you don't need great air to air drones just good enough that ignoring them is a bad idea. My point is you don't need just cheap there is a lot of value in preventing hyper specialization.
However, there is no credible threat to the US military which changes things.
I do appreciate your perspective RogerL. And I do appreciate the fact that we're taking about "cool" robotics that are primarily intended to cause pain, suffering and death. Sometimes at a large scale.
The political reality of the USA right now is that we've got a huge military industrial complex, and it ain't going away anytime soon. One way or the other, we're going to be building weapon systems for air superiority and other uses.
Whether or not that is all a good idea is a discussion for another day. I think that over the last two decades, the USA could have been using more "soft power" to better accomplish our long-term strategic goals than with the precision guided munitions that seemed to be the preferred solution.
Anyway...
At this point, I'd just like to (A) see a better value for my tax dollar. And (B) I'd like to see our military prepared properly for the next war rather than the last one.
With regards to (A), on the cheap vs. expensive scale, I strongly think that we have erred on the side of expensive with regards to the F-35 (and the F-22). I think the Super Hornet (F/A-18E) is expensive enough as it is. Sure, the Navy bought it first, so of course the Air Force doesn't want it...
As far as (B) goes, with all the sabre-rattling we get from Russia and China, realistically speaking, we're not going to be another shooting war with either of them. I'd rather we have more weapon systems that can deal better with the threats we currently have, and will face next. I don't see how the F-35 can really help with that either, in part because the cost means we can't deploy them in the numbers we really ought to.
I realize that automation right now is not up to what a pilot can do, but it is rapidly improving. I'd rather we have a base platform that can accept upgrades easily as they become available, and that can be deployed in numbers to be effective.
> You really aren't going to sit in an armchair and design up a system that'll work.
A lot of the engineers at these companies could do that just as well as I could and succeed at it too. Just because they're technically gifted doesn't mean they'll climb to management and be given enough resources to see a hare-brained project through. I know they're smart; I worked with a lot of them when we were in grad school. Many of the people in my research group went on to work for NASA, Lockheed, and especially Raytheon.
> I've seen far more professional, intelligent, and pragmatic engineers in that work than what I've seen pass for engineering in Silicon Valley.
I would tend to agree that the breadth of technological prowess necessary to succeed at defense stuff is higher than in the valley. But don't confuse the incredibly smart and talented engineers with the often less smart and talented management. Defense contractors are often risk-averse and people trying to ensure long careers can be as well.
When things move as slowly as they do in the military expectations of what can be done by folks not necessarily in the know set the pace of development more than the actual technology does. If something isn't believable it won't get funded. And without that funding the feasibility can't be proven.
I appreciate your insight and honesty about the whole situation. I think if I could get over the moral aspect of making weapons the bureaucracy involved working with/for the military would still drive me insane.
I've made a pretty decent career of tackling projects that a lot of people wouldn't want to touch. Because I have a pretty decent grasp of what's technologically achievable and I'm too stubborn to admit defeat it usually works out OK. I could tell you that I could make something like that work but even with my track record would you believe me? Highly unlikely. I can't say that I would blame you, either.
Ultimately though you're judging the ideas I'm putting forth on US military standards. Someone trying to wipe the US military out would be incredibly foolish to try to win at symmetric warfare. The F-35 is designed to try and win those kinds of battles but because things are so uneven no intelligent enemy would fight us that way. Just look at the whole Iraq/Afghanistan IED situation. Someone is going to build the aerial equivalent of IEDs at some point and when that happens you're going to want survivable airframes over fancy ones in a BIG way.
Part of the reason why defense companies are risk-averse and bureaucratic is because things just CANNOT fail. It's not like, oh yea, my program crashed, I'm just gonna fix this bug and re-compile. No, if your jet crashes on its first test flight, the pilot is going to die. If one thing is wrong with a space satellite, yea no, you cannot pull it back down to Earth to fix it up.
If the thing costs $100mm apiece agreed that it's best if it does not fail.
If it costs $1mm apiece I am less convinced.
How much money is spent engineering to ensure that There Can Be No Failures (tm) of the first flight versus me accidentally destroying the first dozen $1mm drones? Which represents a false economy?
Remember that a drone doesn't have a pilot on-board so if it goes down in flames we're only out the money, not human life.
I appreciate your passion for the subject man, but I feel like you're purposefully taking stuff out of context and generally ignoring a lot of what I write. I've said nothing about space. And I've mentioned several times that I'm not talking about making things that would pass US military muster! The way the US military conducts business is not the ONLY way to conduct business. It's been fairly successful so far but it does not immediately follow that it's the only way.
We lost in Vietnam because it was a war of attrition that our military was completely unprepared to adapt to. We said that we won but in the end it was us that left, not the enemy.
No, after a critical change in political leadership (LBJ -> Nixon, and for that matter the general in charge, Westmoreland -> Abrams), we won, and the first attempt by the NVA to invade the South was utterly crushed, with only 40,000 of the 150,000 invaders managing to get back north, losing their equipment, which included more tanks than used in any single WWII battle. That was an achievement of the ARVN with air support from the US (the South had and used their own, BTW).
Of course the second invasion succeeded, because the US Congress by then was on the other side and had defunded the South, as I recall ammo was so low an infantryman had 1 grenade and less than 100 rounds.
Side note: the cost of outfitting the North with 3 complete armored/mechanized armies (first used up piecemeal, second in that first invasion failure) is now credited with helping to bankrupt the USSR.
> I think if I could get over the moral aspect of making weapons the bureaucracy involved working with/for the military would still drive me insane.
It's not so bad. Okay, it is bad, but I'm driven crazy by a lot of the stuff in the civilian world as well.
Someone, I don't think you, opined that we shouldn't be planning our war machines out a few decades. Not to be mean, but they are clearly woefully misinformed. I'll try to add some data for those not in the field.
I was just watching a youtube clip on a B-52 upgrade. A plane we plan to fly into 2040 or so, which is an 80 year life cycle. I started cackling while watching this 'modernization' as I watched a crew member pull a black rectangle out of the avionics. My gf asked me what was so funny. How to explain.
This was a DTM. I have done a lot of work with them. They are more commonly called a 'brick'. It's a tiny amount of flash memory used to move data on and off the airframe. Cost many thousands of dollars, slow, old technology - well, I'll let google tell you, I don't yack about military capabilities (https://www.fbo.gov/index?s=opportunity&mode=form&id=a1ff338...). But old, obsolete, and so expensive we kept them locked up in our SIL, with sign out procedures to actually get the chance to put your hands on one.
Replace that puppy with a USB stick! Pivot! Be agile!
Ya. Okay, so this is a device to move data around, and yes, one of the last things I was working on was a USB version of these. Oh, it was still the 'brick', but with some ports added. I also worked on previous versions that had a slot for PCMCIA cards. This is still the state of the art version, so far as I know.
Why so slow to update? Well, there is the one vendor problem, to start. The military is trying to move to open, but it is slow, and we have untold systems that are not open. It's sole source, and the lead time for orders are, interesting.
Screw that noise, I'm putting in a USB port. Sure you are. Equipment on a 1960's airframe should handle that, no problem. Okay, so we have a major development effort ahead of us, but it is not insurmountable. How many billions do you want to spend? Because it makes no sense to do this for a B-52, and not other airframes. So it needs to be robust, radiation hardened, capable of operating in dust storms at 150F, in Siberia during blizzards. It needs a self destruct capability so the crew can destroy it if they need to abandon their equipment. And so on. We aren't exactly talking 2 week sprints here.
Okay, we built it. Bolt that puppy on! No. Have you ever gone through the logistics of a retrofit? I have. So, a few of the things to consider. You have to ground each individual airframe - make time for the installation and tests. That means taking it out of service. What do you do with the crew in the meantime? Someone has to install it, someone has to test it. You need to give them training, and then test them to make sure they know their job. which means earlier some defense contractor had to author that training material, and someone else had to make sure that that material is correct. Then we have supply chain issues. Okay, so how many of these do you want me to build? At what price? Oh, ha, ha, add two zeros. Okay, for that price I will make a single line in a factory. Some of this is classified, so we are building a secure facility. Better add another zero to that price. It's built, let's start the run! Hmm, need workers, need to get them clearances, need to train them, need to do all the paperwork required by the government to prove you are handling classified materials correctly, prove that you aren't ripping them off (I honestly think we spend more money proving we aren't ripping the Government off than we save by avoiding the rip offs, but I digress). Okay, line is running. You only needed 1,000 units, you aren't paying to have these workers sit idle, so they are invited to pursue a career elsewhere. My line is mothballed. Oh, got a new order? I can maybe rehire and retrain and be ready in 6-12 months.
Okay, so we have our units. Who is going to maintain them, repair them, test them, store them. How do they get shipped to a war location? Who is going to track them?
Oh, forgot we actually have to get data on and off them. You aren't allowed to just stick a USB stick in a military computer (I hope the security reasons are clear). JMPS, the station the air crew uses to write and read data to these things, needs to be upgraded. This is a 'one size fits all' unit - they serve many different aircraft, so there are multiple data formats, multiple requirements. Which airframe do you want me to upgrade first? (insert massive politics here which don't necessarily bear much resemblance to the rest of the logistics). How fast and cheap do you want it? I can honestly give it to you fast and cheap - by using the same, obsolete protocols and formats from 40 years ago, tying you in deeper to that obsolete technology. But, I can do it faster, testing is easier. Oh, you want a new protocol? Hmm, wonder how many systems that is going to affect? Anyway, decision made, we hire some programmers, get them cleared, get that process going. Somewhere down the line we have to actually test this stuff, so at some point some aircraft will be idling on the runway while us software squirrels swarm around it, running out tests. Opps, some bugs. Can we keep this an extra two weeks? You have a war to fight? Maybe we can get some time in 2016? Awesome!
Software is written and tested, we have a 10 year installation plan, let's go! This is so friggin' agile! I mean, sure, USBs will be obsolete by the time it is fully deployed, but hey ho, we are current to the century!
Not so fast, young and foolish one. Time to rotate the air crew through training. They only have to understand 100 different systems, adding one more ain't no thing. They have to know the old system, and the new one, because it is not being rolled out all at once, and then there is the massive infrastructure for all the other airframes still using the old system. We have to interoperate. So, train air crew, either train the soldiers and marines who will be supporting this in various war zones, or wait for them to be rotated out and rotate in ones that have been trained in the states. Impose a whole friggin' logistics infrastructure over that to deal with having two incompatible systems fielded. Hire up IT folk willing to travel to IRAQ to install all the support hardware. It goes on...
I'm thinking, what, 3 sprints? Ya,right ;). Waterfall will become your best friend, and for a good reason, not obstinance and thick-headedness.
Anyway, this is a small view into how development and 'waste' goes on in the defense world. I put waste in quotes because a lot of this is unavoidable. At least, I've wracked my brain, and I don't see any easy, obvious way to stream line a lot of stuff. I was brief; the above implies a huge amount of logistics, and anything misplanned, any setback, has a ripple effect.
All of that has its frustrations, but it is also very challenging. It's a huge optimization problem. Sure, you are one piece, but your piece is inevitably trying to optimize for a very difficult environment (budget, schedule, capabilities, environmentals, you name it).
We can iterate, pivot, and throw away web apps. There was an article on Ars on how the old Android sw doesn't work because google is shutting down the servers for the obsolete capabilities. It doesn't, and cannot work that way in the military.
I don't mean any of this as a rant. I think it is a really fascinating world, and not many here have experience with it, unless they served (I didn't) or worked for a defense contractor. I do urge you all to rethink making easy potshots at government programs. I have my thoughts about the F-35, and other programs, but hey, until you've tried to run even a tiny program you really aren't in the position to make an informed judgement. I certainly don't feel competent to say "you did it wrong" to F-35 except perhaps in the most sweeping, broad things. That DTM-PCMCIA-USB thing? Real story, and a reflection of the tradeoffs of having different hardware in different systems vs one-size-fits-all. You make 10 disparate systems, well, you are just duplicating effort and wasting tax payer dollars. Try to make one system to do 10 things? Whoa, buckaroo, specialization is the way to go. Solve one problem, cheaply. Okay, I am going to use COTS (commercial off the shelf). I don't think so. It is not secure, hardened, etc. Okay, I'll pay someone to make a system. Oh, tack on 3 zeros to make the IP owned by the government? Yell at me for that. Okay, we will accept a proprietary format. Don't forget to yell at me in 5 years when the company goes bankrupt, gets acquired, or whatever. I should of foresaw that, right? Oh, we'll be agile when that time comes and just swap in a new system. How long could that take? (I refer to to the top of the post for that). When we have all that yelling done, don't forget to yell at me for not having the insight of going with secure wireless that the military just invented... Oh, you want a modernization program to replace the USB? Okay, how big is that checkbook? And don't forget to schedule some yelling time when the wireless gets jammed in the next war.
A Predator can't fly as high as an F-35 or F-22. It can't maneuver nearly as well either. Can't fly as fast. It can't be refueled in flight, so it's range is lower than the F-35. So you'd have to hope that the missile it carried could compensate for these deficiencies. However, missiles are very finicky, and have launch parameters that help them improve their odds. The military looks at it in terms of PoK (probability of kill). So to be cheap (so you can afford to have swarms) you have to give up performance. This will affect PoK.
Next you have to have robust, un-jammable commlinks. So far, no one has those. That's going to be expensive to develop. Humans can make decisions when the comms are degraded, or non-existent. AI has a long ways to go, and AI is expensive. To remain cheap, you'll have to give that up.
With you napkin/elevator calculations, you would be trading 22 Predators for each F-35. 22 Predators is $88M, and the F-35 will eventually drop close to $100m as it reaches IOC. Still, not a bad bargain if it works.
But now, since you've gone cheap and off the shelf, you've got a slow, unmaneuverable aircraft that can't detect a target (no radar), can't communicate well with it's controllers, and can't climb very high, limiting the range of its missiles.
Then, even if you outfit this Predator swarm with an UberMissile, you'll have blown your cost equation out of the water when this swarm comes up against SAMs, or a squadron of F-16s. The swarm won't even see the F-35s due to frontal aspect stealth, and they'll be able to target the Predators and kill them like baby seals. They'll be able to engage, kill, and disengage before the low performance Predators will be able to react.