How Hot is Ukraine Gonna Get?

[chrisl]

45 minutes ago, poesel said:

 Some ballpark calculations about drone evasion:
Assuming a muzzle velocity of 1000 m/s for the PD, that is 1 s time of flight at 1000 m range. Let's assume the drone can pull 6 g perpendicular to the PD (I haven't found any data about angular velocity of racing drones, so this is just half of the 12 g we had). For 1 s that is 3 m of movement. That sounds survivable for the drone.
At 500 m that is 0,5 s TOF for the bullet and only 0,75 m for the drone.
Half that again to 250 m and 0,25 s and the drone moves just 0,19 m. And that puts it already inside a buckshot circle.

Although the zigzagging of drones looks insane to humans, it is not that fast wrt to bullet speeds.

 

Have you ever been to Alaska?

The mosquitos are big and slow.  Enough so that you can mistake them for small birds and might want to use a shotgun on them.  They're extremely easy to spot, hit, and kill.

But there are so many of them that there's no way you're avoiding mosquito bites unless you have sufficient armor that they can't penetrate.  Bike shorts are useless.  Thick knit gloves aren't much better - their probosci just slip right through.  Gore-tex works pretty well because it's a really tiny pore size.

That's where things are headed with UAVs/UGVs, except they have much more effective penetrators than mosquitos do.

[ArmouredTopHat]

5 minutes ago, Probus said:

You guys have gone over and over this topic in this thread, but I am still astounded by Putin's willingness to throw his country down the tubes.

Its why until he is removed by the Russian people I dont see the country's stance changing. (And even then it might just be maintained via intertia) One can hope that the increasing financial strain of the war for Russia gets to a level where the otherwise politically passive Russian people finally have enough. Or the oligarchs start actually making some calls instead of letting themselves get killed off. 

Having spoken to a few Russians in the past, the level of ambivalence and straight out political passiveness that comes out from them is truly shocking at times and is a testament to the level of control Putin has over the country, even if things like the Wagner episode show its brittleness. My experiences are of course entirely anecdotal but it does seem to match the general trend that until this war really starts to hurt the average Russian, they really do not care all that much about the attempted destruction of the Ukrainian nation, or at least are happy to buy the reasons why the Russian government wages such an overt war of conquest. 

The level of propaganda is ingrained even to those outside of the country to the point where its eerily similar to the sort of indoctrination we saw with the Nazis, if not even more insidious. The common consensus with these people I talked to was that after battering down the political propaganda points they usually first spouted was that to them its all pointless anyway because they cannot do anything about it anyway, a point that is mildly true yet they use it as an excuse not to denounce the invasion. There is no easy solution to address this. 

[ArmouredTopHat]

4 minutes ago, chrisl said:

The mosquitos are big and slow.  Enough so that you can mistake them for small birds and might want to use a shotgun on them.  They're extremely easy to spot, hit, and kill.

But there are so many of them that there's no way you're avoiding mosquito bites unless you have sufficient armor that they can't penetrate.  Bike shorts are useless.  Thick knit gloves aren't much better - their probosci just slip right through.  Gore-tex works pretty well because it's a really tiny pore size.

That's where things are headed with UAVs/UGVs, except they have much more effective penetrators than mosquitos do.

The point being made is that if AI are being used to target and attack things, then AI can be used to defend them as well.

Current point defence is limited perhaps most of all by the fact its got a large human presence in the envelope who have to identify and task the appropriate asset as well as pull the trigger. AI and ML will in theory greatly reduce the time needed to acquire, track and validate a target before engaging. Certainly compared to a hypersonic missile coming at you a computer has all the time in the world to detect, track and engage a drone in good time with a fair degree of success. I imagine for the close in PD systems that might be seen in the next few decades that they might be entirely autonomous in operation and simply have to be managed like any other system in a vehicle.  

You go from a system that can maybe deal with a couple of incoming threats to one that can deal with half a dozen or more in seconds. Then you think about that system being networked to a a group of similar devices (Say in a grouping of vehicles for instance) for optimal hard kills that minimise collateral, combined with drone interceptors that would of greatly constrained a swarm attack to begin with. The efficiency implication is just as terrifying for the defence as it is the offense imo. 

[chrisl]

25 minutes ago, ArmouredTopHat said:

The point being made is that if AI are being used to target and attack things, then AI can be used to defend them as well.

Current point defence is limited perhaps most of all by the fact its got a large human presence in the envelope who have to identify and task the appropriate asset as well as pull the trigger. AI and ML will in theory greatly reduce the time needed to acquire, track and validate a target before engaging. Certainly compared to a hypersonic missile coming at you a computer has all the time in the world to detect, track and engage a drone in good time with a fair degree of success. I imagine for the close in PD systems that might be seen in the next few decades that they might be entirely autonomous in operation and simply have to be managed like any other system in a vehicle.  

You go from a system that can maybe deal with a couple of incoming threats to one that can deal with half a dozen or more in seconds. Then you think about that system being networked to a a group of similar devices (Say in a grouping of vehicles for instance) for optimal hard kills that minimise collateral, combined with drone interceptors that would of greatly constrained a swarm attack to begin with. The efficiency implication is just as terrifying for the defence as it is the offense imo. 

Yes, you get to use AI/ML on defense.  With the mosquitos (who aren't coordinated, but are extremely numerous) against coordinated networked targets with natural intelligence (humans inside of a screen tent working together as each person enters), the mosquitos always get through.

But if the targets of the attacker are the small numbers of big slow things, they have a much easier time of it than the defender, who has to defend against larger numbers of small fast things.  And has to get all of them, because any one of the small fast things getting through will ruin your day.  In WWII it was "the bombers always get through".

Hypersonic missiles are easier to intercept than drones.  We've already seen that.  Hypersonic is mostly a marketing thing to sound scary, but if they try to maneuver much at all the rapidly lose the "hyper" part of things.

I wouldn't spend a lot of money on PD, and not on making something fancy that can deal with lots of targets, because it's always going to be cheaper for the attacker to just send more cheap things to overwhelm it.  The real action is in layered networked defense where loss of any one or 10 nodes makes no difference.  And lots of bird block.  Spinning things hate tangly things.  Individual guys on the ground need web shooters like spiderman.

[Battlefront.com]

4 hours ago, ArmouredTopHat said:

A drone zig zagging is not going to be able to dodge bullets like Neo for very long, it might complicate a kill time, but that is probably where airbursts come in for heavier systems against a swarm. Again, layers of defence spring to mind. Evasion becomes harder the closer you get to target as well.

You need to think through your answers more and knee jerk defend the status quo less.  You have STILL not addressed what I laid out for you, so let me spell it out in painstaking detail and maybe you can see why I'm banging on about this.

Something (anything you like) that is protected by your fantasy mini-Phalanx system is sitting in any terrain of your choosing.  An enemy drone comes in at it from 2m off the ground.  Your mini- Phalanx system, because it is a fantasy item, can accurately track and hit anything that comes near it without regards to ground clutter, target speed, target agility, weather, jamming, etc.  It fires off a small burst of accurately placed ammo right at the incoming drone.  Every single shot is a hit, because after all this is a fantasy system.  So, here's my question...

After the rounds pass through the enemy drone, where do they land?  Because the laws of physics dictate that they are not going to stop when they hit a bit of plastic.  2m off the ground is Human head height.  It is less tall than most vehicles, soft skinned or armored.  It's a 1st story building.  It's any number of other tings that you probably don't want to hit.

Even if the fantasy weapon with all the other fantasy gifts I gave has another fantasy of precision airburst munitions, that doesn't change what I just asked about.

What do you say about collateral damage potential even under fantasy perfect conditions?

Now with that said, start chipping away at the fantasy and start introducing reality.  Limitations on tracking rapidly shifting targets, problems with ground clutter, misses, volume of fire, nothing blocking LOF, etc.  What about collateral damage then?

And finally, presume a half dozen drones coming in 2m off the ground from wildly different angles and see where that takes you.

 

4 hours ago, ArmouredTopHat said:

I am not assuming anything here. I am just pointing out we are largely talking about theoretical and we dont know what the practical result of AI swarm vs AI defence might look like because there is nothing close on that front to existing yet, in military terms anyway.

This is a steaming pile of BS and I'm going to continue calling it out.  You are more than happy to dream up defensive systems that aren't even close to being fielded and presume they are effective, but when it comes to AI controlled swarms (a thing that DOES exist right now) you say "well, it's all theoretical".  Sorry, that's a crap argument and it is also ignorant of where drones are right now.

Swarms exist.  Drones that attack things exist.  Combining the two is hardly a stretch and, in fact, some defense companies have already done so.  This was three years ago:

 

4 hours ago, ArmouredTopHat said:

I ask again, what is stopping an AI system from using defensive interceptor drones to nullify an incoming attack, with PD acting as a last resort and cleanup?

Point defense is a fantasy and an extremely expensive one to maintain.  Sending out interceptor drones has many limitations, but it certainly is more viable than some $1m+ defense system that can only handle corner cases at best.

Interceptor drones have a long ways to go before they can do what they need to do.  A major problem is that the speed of the incoming drones is likely to be as fast as the interceptors.  This means the interceptor must hit the incoming drone first pass because it won't be able to double back.  And since a last second dodge by the incoming drone is impossible to predict, hitting will be extremely difficult.  Airbursting the interceptor may work, but that gets into another host of problems.

4 hours ago, ArmouredTopHat said:

I am also pointing out that we are currently seeing increasingly smaller point defence systems that are capable of shooting down drones right now. These systems do in fact exist, right now. The technology is practical, its going to evolve just as much as drone munitions will. 

We are seeing nothing of the sort.  We are seeing carefully controlled, super expensive, very limited weapons systems that are knocking down unrealistically simplistic targets. They also have zero, and I mean ZERO, chance of being scalable to the extent they are needed to protect anything other than fixed facilities.  And even then, I don't think they hold out much promise.

Steve

[ArmouredTopHat]

4 minutes ago, chrisl said:

Yes, you get to use AI/ML on defense.  With the mosquitos (who aren't coordinated, but are extremely numerous) against coordinated networked targets with natural intelligence (humans inside of a screen tent working together as each person enters), the mosquitos always get through.

No defence is perfect, but having a defence is better than nothing. Think of how many lives are saved by Mosquito nets to continue your analogy. Just because they dont stop every mosquito doesn't mean its not worth pursuing. Anything that takes a reasonable effort that complicates a kill chain is worth pursing. 

Of course a munition is going to get through, PD will not be an infallible solution. The question is how much can a defence network reduce mission kills / damage / losses of assets for any particular operation. The alternative is quite literally death. 
 

9 minutes ago, chrisl said:

Hypersonic missiles are easier to intercept than drones.  We've already seen that.  Hypersonic is mostly a marketing thing to sound scary, but if they try to maneuver much at all the rapidly lose the "hyper" part of things.

Oh of course hypersonic are overrated by the vatniks, the comparison here is more that we have built machines for decades now that make rapid and quick calculations at a speed we cannot perceive.

I very much agree that a lot of the AA kit we see in Ukraine has not done so well against drones...but that's entirely to be expected. Most of these things were simply not designed to hit such small targets. Gepard is by all accounts a fantastic system for instance, but anything smaller than a shaed drone and its struggles. Same for Tunguska, Panstir and various other systems. 

The moment you look at systems more optimised for acquiring small drones using various means, you see the stark difference. (That video I linked showed how easy it was for a system to track and acquire one to kill it within seconds) 

15 minutes ago, chrisl said:

I wouldn't spend a lot of money on PD, and not on making something fancy that can deal with lots of targets, because it's always going to be cheaper for the attacker to just send more cheap things to overwhelm it.  The real action is in layered networked defense where loss of any one or 10 nodes makes no difference.  And lots of bird block.  Spinning things hate tangly things.  Individual guys on the ground need web shooters like spiderman.

Why not both? We have amazing and capable systems on ships that can reach out and hit targets at fantastic ranges, yet we have PD on them for a reason. Its not the desired outcome to have such a close call but sometimes a mosquito will slip through a few nets...and at that point you need something to squish it. Networked layers of defence would only be complimented by a last ditch system that can comfortably deal with strays while adding redundancy. 

[Battlefront.com]

4 minutes ago, ArmouredTopHat said:

Why not both?

Because point defense is ridiculously expensive and impractical.  Every Dollar that is spent on those systems is a Dollar not invested into something that has a much better chance of actually doing what it is expected to do.

Resources are finite.  The more that is blown on dead ends is money not spent on something that has at least some hope of solving the problems we need to solve.

Steve

[dan/california]

Like these. We watsed more money than most navies have ever spent on these fiascos.

[sburke]

6 minutes ago, ArmouredTopHat said:

Why not both? We have amazing and capable systems on ships that can reach out and hit targets at fantastic ranges, yet we have PD on them for a reason. Its not the desired outcome to have such a close call but sometimes a mosquito will slip through a few nets...and at that point you need something to squish it. Networked layers of defence would only be complimented by a last ditch system that can comfortably deal with strays while adding redundancy. 

The analogy doesn't work for a very simple reason.  The ships are in a task force, all the high priority items are a self-contained unit with enough supply in the task force for all its needs for quite a bit.  (Assuming that it actually can defend itself (which after the demise of the BSF to a country with no Navy one has to wonder.)   As @The_Capt has said so many times I have started hearing it in my sleep - the logistical chain is now the weak point.  The defense you are referring to now has to be on every link of that chain or it is pretty worthless. A combat brigade isn't going to do much if all its resupply is f'd.  That exposure is now 10's of KM deep.  The combat team runs out of resupply and it is stuck... now the artillery rain starts coming in.  Unless this fantasy defensive solution is covering everything, including the supply chain... well it isn't really a defense anymore.  it's just lots spent to protect the tip of the spear while the shaft of the spear is.. well shafted.

[Battlefront.com]

2 hours ago, poesel said:

Since you assume the attackers to be networked, I do the same for the defenders. So the defenders know where their own are. Easier for the vehicles than for the squishies of course. But when I doubt, shoot it. The explosives on the drone will do more damage.

Actually, I don't presume the attackers are networked.  In fact, I presume that they are fully autonomous.

But networking doesn't solve the problems of friendly fire coming from point defense weapons.  See previous post.

2 hours ago, poesel said:

Why? Drones emit a very distinctive sound. They are hot. They may emit radio signals. And last but not least, they are fast moving objects in front of a mostly static background

Detecting and tracking are two different things.  I can know there's a drone coming at me, but I don't necessarily know where it is within the 3D volume of space to the degree necessary to intercept it effectively.  And again, I am presuming full autonomy which means ZERO emissions. 

They might be fast moving objects against a mostly static background, but that background is a problem for targeting.  You can see that in any CM game where the target is moving behind/around/near other objects.

Lastly, the defense needs to be a very heavy emitter. That opens up all kinds of lethal possibilities for the enemy to exploit.

 

2 hours ago, poesel said:

That is your interpretation of what I wrote, and I think you are wrong.

Well, I tried to give you some credit, but if you insist on being 100% wrong then who am I to argue

 

2 hours ago, poesel said:

ome ballpark calculations about drone evasion:

Assuming a muzzle velocity of 1000 m/s for the PD, that is 1 s time of flight at 1000 m range. Let's assume the drone can pull 6 g perpendicular to the PD (I haven't found any data about angular velocity of racing drones, so this is just half of the 12 g we had). For 1 s that is 3 m of movement. That sounds survivable for the drone.
At 500 m that is 0,5 s TOF for the bullet and only 0,75 m for the drone.
Half that again to 250 m and 0,25 s and the drone moves just 0,19 m. And that puts it already inside a buckshot circle.

Although the zigzagging of drones looks insane to humans, it is not that fast wrt to bullet speeds.

The only way to get a bullet reliably onto a fast moving target is to be able to predict it's path.  Once the path is unpredictable, the chance of a miss are almost assured when talking about the sort of speed and maneuverability of a drone.  Airburst munitions mitigate that somewhat because "close enough is good enough", but the systems that can do that sort of thing are very expensive and big.  Two things that mean they aren't a practical solution for anything other than a limited number of fixed installations.

And it still doesn't solve for friendly fire situations.

Steve

 

[chrisl]

1 minute ago, Battlefront.com said:

 

And it still doesn't solve for friendly fire situations.

 

The mitigating factor for friendly fire situations is that I'm only putting drones/UGVs/robots in the field - there's way too much crazy dangerous stuff out there and I'm going to sit in the basement with fiber links to a scattered network of receivers to watch how it goes down.

[Butschi]

19 hours ago, chrisl said:

All the ChatAIs are basically just "fits" to all the garbage typed into the internet, so they come up with some spectacularly bad "hallucinations".  It's because there's no actual model of reality behind them.

 

15 hours ago, dan/california said:

The new style AI as Chris said is really just an enormously complicated pile of statistics that guesses at the most likely "average answer" to your question. Its guesses are only as good as its input data base. Since three quarters of the internet is questionable at best, models trained on it are a neat parlor trick, but almost useless if you need to trust the answers.

I think you are both oversubscribing a little too this popular idea that large language models "just" learn to predict the next word (and based on garbage Internet data), so they are stupid.

While the predicting the next word part is true, these models train to predict the next word based on learning which words are important to which other words. So they are basically learning context. The novel thing is - and this is why those models are increasingly popular besides just chatting - they can reason about their answers. Sure at the and of the day this is still just fitting parameters to data but our brains work in a not much different way.

Also, while ChatGPT for instance was trained in an unsupervised way with Internet data, a lot of supervised training came afterwards - talking to humans who reviewed its answers.

As for hallucinations - I doubt this is just "well Internet is garbage so chatbot answers garbage." I agree that it is about a model or concept of reality. Humans generally (but not always) know the difference between reality and imagination. That is because we have sensors, i.e. eyes, ears, etc. A chatbot doesn't have that... yet.

[ArmouredTopHat]

54 minutes ago, Battlefront.com said:

You need to think through your answers more and knee jerk defend the status quo less. 

I think you need to stop thinking that I am defending a status quo when literally talking about future weapon systems. How does that even make sense?

 

54 minutes ago, Battlefront.com said:

Something (anything you like) that is protected by your fantasy mini-Phalanx system is sitting in any terrain of your choosing.  An enemy drone comes in at it from 2m off the ground.  Your mini- Phalanx system, because it is a fantasy item, can accurately track and hit anything that comes near it without regards to ground clutter.  It fires off a small burst of accurately placed ammo right at the incoming drone.  Every single shot is a hit, because after all this is a fantasy system.  So, here's my question...

After the rounds pass through the enemy drone, where do they land?  Because the laws of physics dictate that they are not going to stop when they hit a bit of plastic.  2m off the ground is Human head height.  It is less tall than most vehicles, soft skinned or armored.  It's a 1st story building.  It's any number of other tings that you probably don't want to hit.

Even if the fantasy weapon with all the other fantasy gifts I gave has another fantasy of precision airburst munitions, that doesn't change what I just asked about.

What do you say about collateral damage potential even under fantasy perfect conditions?

Can we stop being so rude with the language please? Calling peoples ideas fantasy despite them being rooted in reality is a little grating. There is no reason to be this abrasive. I would appreciate a little more decorum and it would improve your points quite a bit.

We have used far more destructive weapon systems in such environments and view them as an acceptable risk for collateral. You are talking about bullets being fired in a warzone where there are already plenty of bullets and artillery being flung around in most warzones. Presumably an autonomous system would have safeguards to ensure that a person is not shot instead of a drone. Surely the same limitations are going to be put into drone munitions using swarm tactics to prevent friendly fire when operating in an area filled with friendlies? 

All of this ignores the primary point that drone based interceptors are going to be a primary means of defence. 
 

54 minutes ago, Battlefront.com said:

What do you say about collateral damage potential even under fantasy perfect conditions?

ERA is not exactly great when it comes to collateral yet its used heavily anyway. We have long since adopted basic ideas such as keeping dismounted infantry away from a vehicle, the same principle applies here. Ukraine fires tens of thousands of rounds into the sky against shaeds with a variety of weapons to down them. Those bullets also come down eventually. Collateral happens but the sad reality is that its part of the war. Is your argument really that because someone might get shot by accident the whole effort is worthless? I could argue collateral being a problem for offensive swarm drones but I dont use it as an argument that their whole enterprise is a waste of time. 
 

54 minutes ago, Battlefront.com said:

This is a steaming pile of BS and I'm going to continue calling it out.  You are more than happy to dream up defensive systems that aren't even close to being fielded and presume they are effective, but when it comes to AI controlled swarms (a thing that DOES exist right now) you say "well, it's all theoretical".  Sorry, that's a crap argument and it is also ignorant of where drones are right now.

This was three years ago.  Swarms exist.  Drones that attack things exist.  Combining the two is hardly a stretch and, in fact, some defense companies have already done so.

Ok I am sorry but pointing to a video of some drones doing a preprogramed show routine with fancy lights is nothing near an autonomous swarm system designed to hunt and kill a variety of targets on a battlefield. Your telling me I am making stuff up but the fact of the matter is there is no lethal drone swarm system in service and there is unlikely to be one for a while yet. 

https://mwi.westpoint.edu/swarm-clouds-on-the-horizon-exploring-the-future-of-drone-swarm-proliferation/#:~:text=In May 2021%2C Israel used,to the Israel Defense Forces.

The closest we have had is this, and its entirely non lethal system. 

https://bluehalo.com/bluehalo-selected-for-u-s-army-next-generation-c-uas-missile/

There are lethal systems in the works, but in the case of the US, the very same system is now being selected as a C-UAS role. So defence swarms are likely to be fielded in combat at the same time as offensive ones. Either way its still very early days for swarms. 

54 minutes ago, Battlefront.com said:

Point defense is a fantasy and an extremely expensive one to maintain.  Sending out interceptor drones has many limitations, but it certainly is more viable than some $1m+ defense system that can only handle corner cases at best.

Again, you are taking the example of a big expensive / static systems that were designed to down anti ship missiles in the 80s and applying it to new designs being made in the last decade. Of course a CIWS is big and expensive. No one is suggesting we plaster one of those onto everything. 

https://battle-updates.com/small-calibre-solutions-for-c-uas-systems-by-julian-nettlefold/

There are a plethora of promising systems literally in existence that could become very promising with the appropriate development. All the ingredients already exist and people are actively cooking with them. These systems are only going to get cheaper and more prolific. 
 

35 minutes ago, Battlefront.com said:

Actually, I don't presume the attackers are networked.  In fact, I presume that they are fully autonomous.

A swarm operating has to communicate with others in its group in some way, otherwise what is stopping the entire group from hitting one target instead of many more efficiently? How else is the swarm supposed to receive orders or be managed on the battlefield? If the system is entirely autonomous that implies a lot of hurdles have been jumped. Most envision a lot of 'autonomous' systems to still be managed in some way shape or form by humans. Because AI does in fact have limitations that can constrain. 

Of course, that would apply to a autonomous PD system as well, but if its integrated into a manned vehicle it literally has the crew to manage it.

39 minutes ago, sburke said:

The analogy doesn't work for a very simple reason.  The ships are in a task force, all the high priority items are a self-contained unit with enough supply in the task force for all its needs for quite a bit.  (Assuming that it actually can defend itself (which after the demise of the BSF to a country with no Navy one has to wonder.)   As @The_Capt has said so many times I have started hearing it in my sleep - the logistical chain is now the weak point.  The defense you are referring to now has to be on every link of that chain or it is pretty worthless. A combat brigade isn't going to do much if all its resupply is f'd.  That exposure is now 10's of KM deep.  The combat team runs out of resupply and it is stuck... now the artillery rain starts coming in.  Unless this fantasy defensive solution is covering everything, including the supply chain... well it isn't really a defense anymore.  it's just lots spent to protect the tip of the spear while the shaft of the spear is.. well shafted.

The point of something smaller and cheaper PD wise would be that you could in fact stick it onto every vehicle. an RWS system is not all that expensive compared to overall vehicle costs and certainly nothing like the cost of something like CIWS. The primary idea remains that a drone based interception system would still be handling the majority of incoming. Add a greater redundancy to logistics and a means for them to defend themselves, plus constrained enemy ISR that makes it harder for them to locate your logistics in the first place and its not unreasonable to assume this whole system is possible. 

 

35 minutes ago, Battlefront.com said:

Detecting and tracking are two different things.  I can know there's a drone coming at me, but I don't necessarily know where it is within the 3D volume of space to the degree necessary to intercept it effectively.  And again, I am presuming full autonomy which means ZERO emissions. 

Even a fully autonomous system will be noisy and easily found on a thermal or IR system. 

35 minutes ago, Battlefront.com said:

They might be fast moving objects against a mostly static background, but that background is a problem for targeting.  You can see that in any CM game where the target is moving behind/around/near other objects.

For humans maybe. 

46 minutes ago, Battlefront.com said:

Because point defense is ridiculously expensive and impractical. 

Drones started out pretty expensive, plenty still are. Why are we assuming a potential pd system has to have the cost of a CIWS? Why do drones get to evolve as a system but nothing else can?

54 minutes ago, Battlefront.com said:

You need to think through your answers more and knee jerk defend the status quo less.  You have STILL not addressed what I laid out for you, so let me spell it out in painstaking detail and maybe you can see why I'm banging on about this.

Perhaps if you were to be a little less condescending about what remains a largely theoretical possibility of what future warfare might look like we could have a better debate on it.  

Edited July 7 by ArmouredTopHat

[MikeyD]

The drone argument seems to have come to the conclusion that 'Well, warfighting is now impossible, no more war'. Except that wars are not over by a long shot. The wars in Somalia and Sudan are not over. The war in Israel isn't over, the war in Myanmar isn't over. We need to define our terms better. What sort of war in now impossible? May I suggest near-peer large scale long duration mechanized combat. But how many wars fall under that definition?

[chuckdyke]

2 minutes ago, MikeyD said:

May I suggest near-peer large scale long duration mechanized combat.

Well we play CM let us do D-Day with the formations from Black Sea or SF2. 

[ASL Veteran]

 

[photon]

I wonder whether more knowledgable people than me could hammer out the specs that a PDS system would need?

Let's say that you want to protect against a four drone hammer and anvil attack. Two pairs of drones coming in separated by 180° in the x-y plane diving in at 30° and 60° dives, with another pair in the same vertical plane coming in low and slow for the initial approach, and then bouncing between stationary and max speed for the last 50 meters or so less than a meter from the ground.

It's an angular mechanics problem. Your barrel needs to be able to repoint through maybe 110° of elevation and 360° of traverse, and your targeting platform needs to be able to track multiple incoming targets (with persistence for the slow blade slipping through the shield). If you set the drone max speed at 33 meters per second and an 100 meter engagement, range, you've got about three seconds per drone to target, point, and engage. Is that feasible? How fast can CROWS track?

Edited July 7 by photon

[chrisl]

1 hour ago, Butschi said:

 

I think you are both oversubscribing a little too this popular idea that large language models "just" learn to predict the next word (and based on garbage Internet data), so they are stupid.

While the predicting the next word part is true, these models train to predict the next word based on learning which words are important to which other words. So they are basically learning context. The novel thing is - and this is why those models are increasingly popular besides just chatting - they can reason about their answers. Sure at the and of the day this is still just fitting parameters to data but our brains work in a not much different way.

Also, while ChatGPT for instance was trained in an unsupervised way with Internet data, a lot of supervised training came afterwards - talking to humans who reviewed its answers.

As for hallucinations - I doubt this is just "well Internet is garbage so chatbot answers garbage." I agree that it is about a model or concept of reality. Humans generally (but not always) know the difference between reality and imagination. That is because we have sensors, i.e. eyes, ears, etc. A chatbot doesn't have that... yet.

Yes, I'm well aware of the whole context thing.  That's really the only way they could predict other words, isn't it? And there really isn't any reasoning behind the answers.  They just are mimicking confidently wrong answers on the internet.

My general observation is that they nosedive fast, and are subject to repeating a lot of incorrect information because there's a lot of wrong stuff that gets repeated on the internet. They have no reality model behind them, and limited (at best) weighting of the relative value & validity of their input sources (because they have no model of reality).  I've seen AI articles that were analyses of songs that included discussion and analysis of lines that weren't even in the song in any version, ever. And now the internet is being heavily populated with AI pages so that future AI training will be based on the superficial summaries of this generation.  It will converge towards meaninglessness.  

[sburke]

54 minutes ago, ArmouredTopHat said:

plus constrained enemy ISR that makes it harder for them to locate your logistics in the first place and its not unreasonable to assume this whole system is possible. 

What magic bullet are you using for that?

It is all reasonable if you allow for a complete suspension of belief at how unreasonable it is.

Quote

There are a plethora of promising systems literally in existence that could become very promising with the appropriate development. All the ingredients already exist and people are actively cooking with them. These systems are only going to get cheaper and more prolific. 

you want to give some examples as all I saw there was a hand wave that "hey people are already doing this", but nothing to show that is actually the case.

The problem I see (not with your concepts, but just the overall discussion) is that things are moving so fast that we are kind of fumbling in the dark throwing stuff at the wall in the way of ideas that are likely becoming outdated as soon as we think of them.

[The_Capt]

33 minutes ago, MikeyD said:

The drone argument seems to have come to the conclusion that 'Well, warfighting is now impossible, no more war'. Except that wars are not over by a long shot. The wars in Somalia and Sudan are not over. The war in Israel isn't over, the war in Myanmar isn't over. We need to define our terms better. What sort of war in now impossible? May I suggest near-peer large scale long duration mechanized combat. But how many wars fall under that definition?

We have discussed this at length. A drone swarm can only really be countered by another drone swarm. This is like WW2 where AAA was never going to stop AirPower.  Only other air power could interdict.  So we will likely see swarm clouds colliding with other clouds until one collapses.  Then everything and anything left in the collapsed cloud will die.  Further, this won’t only be in the air.  It will be on the ground/sea and subsurface at sea (maybe on land too).

This is the end state expression of where this thing is going but we will see iterations and variations along the way. It is not the end of war anymore than the introduction of the machine gun was. It is the beginning of a new form of war.

[ArmouredTopHat]

4 minutes ago, sburke said:

What magic bullet are you using for that?

It is all reasonable if you allow for a complete suspension of belief at how unreasonable it is.

We are literally seeing Ukraine smack down UAVs with FPV drones at a growing rate. The potential is there for unsustainable attrition of long range recon assets. 

Its possible that UAVs will have a lot more to worry about going forward if more dedicated counter systems come online that are even more efficient. 

 

7 minutes ago, sburke said:

you want to give some examples as all I saw there was a hand wave that "hey people are already doing this", but nothing to show that is actually the case.

I literally supplied a link showing a bunch of systems in that same post....

https://battle-updates.com/small-calibre-solutions-for-c-uas-systems-by-julian-nettlefold/

There is considerable interest in this right now. 

[The_Capt]

2 minutes ago, sburke said:

What magic bullet are you using for that?

It is all reasonable if you allow for a complete suspension of belief at how unreasonable it is.

you want to give some examples as all I saw there was a hand wave that "hey people are already doing this", but nothing to show that is actually the case.

The problem I see (not with your concepts, but just the overall discussion) is that things are moving so fast that we are kind of fumbling in the dark throwing stuff at the wall in the way of ideas that are likely becoming outdated as soon as we think of them.

Constrained ISR in the 21st century is essentially impossible.  Unless we deny space with large EMP, can control high-to-low altitude air, and every cell phone and hotspot (which means basically wiping out all civilian communications infrastructure) along with traditional military systems…ISR has come of age.  

Sensors are so small, processing so cheap, light and powerful, and communications so ubiquitous that any dependence of a military operation on an opponents ISR being denied either lead directly to escalations we will not do (eg erasing space-based) or is simply impractical.  Far better to invest in deception and stealth to buy time and space than to hope that every cell phone in a given nation suddenly stops working.

This is really the problem with the conservative position, they simply have not kept up with technological reality.  Finally, a lot of national information infrastructure is double purposed.  So “constraining ISR” crosses a line within LOAC. For example, say we want to deny communications to stop ISR.  We would likely have to also stop 911/medical services as well.  Cyber was supposed to provide options in these spaces but it is not able to produce results.

But hey, some are willing to wave magic wands and remove UAS and precision fires from the future battlefield.  Why not simply remove C4ISR realities as well?  Hell we should just prepare for another Gulf War ‘91 and all will be well.

[ArmouredTopHat]

42 minutes ago, The_Capt said:

Constrained ISR in the 21st century is essentially impossible.

Yet we might see Russian long range ISR hampered by drones. Time will tell.

I should have perhaps specified long range uavs but eh. 

[Sojourner]

2 hours ago, ArmouredTopHat said:

I think you need to stop thinking that I am defending a status quo when literally talking about future weapon systems. How does that even make sense?

 

Can we stop being so rude with the language please? Calling peoples ideas fantasy despite them being rooted in reality is a little grating. There is no reason to be this abrasive. I would appreciate a little more decorum and it would improve your points quite a bit.

We have used far more destructive weapon systems in such environments and view them as an acceptable risk for collateral. You are talking about bullets being fired in a warzone where there are already plenty of bullets and artillery being flung around in most warzones. Presumably an autonomous system would have safeguards to ensure that a person is not shot instead of a drone. Surely the same limitations are going to be put into drone munitions using swarm tactics to prevent friendly fire when operating in an area filled with friendlies? 

All of this ignores the primary point that drone based interceptors are going to be a primary means of defence. 
 

ERA is not exactly great when it comes to collateral yet its used heavily anyway. We have long since adopted basic ideas such as keeping dismounted infantry away from a vehicle, the same principle applies here. Ukraine fires tens of thousands of rounds into the sky against shaeds with a variety of weapons to down them. Those bullets also come down eventually. Collateral happens but the sad reality is that its part of the war. Is your argument really that because someone might get shot by accident the whole effort is worthless? I could argue collateral being a problem for offensive swarm drones but I dont use it as an argument that their whole enterprise is a waste of time. 
 

Ok I am sorry but pointing to a video of some drones doing a preprogramed show routine with fancy lights is nothing near an autonomous swarm system designed to hunt and kill a variety of targets on a battlefield. Your telling me I am making stuff up but the fact of the matter is there is no lethal drone swarm system in service and there is unlikely to be one for a while yet. 

https://mwi.westpoint.edu/swarm-clouds-on-the-horizon-exploring-the-future-of-drone-swarm-proliferation/#:~:text=In May 2021%2C Israel used,to the Israel Defense Forces.

The closest we have had is this, and its entirely non lethal system. 

https://bluehalo.com/bluehalo-selected-for-u-s-army-next-generation-c-uas-missile/

There are lethal systems in the works, but in the case of the US, the very same system is now being selected as a C-UAS role. So defence swarms are likely to be fielded in combat at the same time as offensive ones. Either way its still very early days for swarms. 

Again, you are taking the example of a big expensive / static systems that were designed to down anti ship missiles in the 80s and applying it to new designs being made in the last decade. Of course a CIWS is big and expensive. No one is suggesting we plaster one of those onto everything. 

https://battle-updates.com/small-calibre-solutions-for-c-uas-systems-by-julian-nettlefold/

There are a plethora of promising systems literally in existence that could become very promising with the appropriate development. All the ingredients already exist and people are actively cooking with them. These systems are only going to get cheaper and more prolific. 
 

A swarm operating has to communicate with others in its group in some way, otherwise what is stopping the entire group from hitting one target instead of many more efficiently? How else is the swarm supposed to receive orders or be managed on the battlefield? If the system is entirely autonomous that implies a lot of hurdles have been jumped. Most envision a lot of 'autonomous' systems to still be managed in some way shape or form by humans. Because AI does in fact have limitations that can constrain. 

Of course, that would apply to a autonomous PD system as well, but if its integrated into a manned vehicle it literally has the crew to manage it.

The point of something smaller and cheaper PD wise would be that you could in fact stick it onto every vehicle. an RWS system is not all that expensive compared to overall vehicle costs and certainly nothing like the cost of something like CIWS. The primary idea remains that a drone based interception system would still be handling the majority of incoming. Add a greater redundancy to logistics and a means for them to defend themselves, plus constrained enemy ISR that makes it harder for them to locate your logistics in the first place and its not unreasonable to assume this whole system is possible. 

 

Even a fully autonomous system will be noisy and easily found on a thermal or IR system. 

For humans maybe. 

Drones started out pretty expensive, plenty still are. Why are we assuming a potential pd system has to have the cost of a CIWS? Why do drones get to evolve as a system but nothing else can?

Perhaps if you were to be a little less condescending about what remains a largely theoretical possibility of what future warfare might look like we could have a better debate on it.  

Sorry, you want room 12A next door...

 

[MikeyD]

The article's a bit dated (almost 2 months ago) but it discusses the progress with laser defense in-service tests. Stryker Defender is not exactly being praised after 2 months in the field.

https://breakingdefense.com/2024/05/army-soldiers-not-impressed-with-strykers-outfitted-with-50-kilowatt-lasers-service-official-says/