chrisl

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

Big rolls of bird block.  It stops birds.  It stops cats.  I've even seen it hang up a 200 lb bear for a bit.

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.

It's not one of them doing that - a point defense system might be able to deal with that, even at crazy high speeds and accelerations.  It's a half dozen of them, communicating with each other.  On every single target.  The only way to be sure you hit them is if you're a frag grenade with a high density and uniform distribution of metal.  
eta: it's a good thing you design historical games and don't have to stick to modern games.  Because CM:Drone Wars will be a one-turn slaughterfest.

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.

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.

All of this is true, but before we spend tens of billions to equip the ground force with a sytems that claims it can do this the testing needs to be savage. The worst possible place is to spend a bunch of money on drone defenses that don't work, and then develop doctrine that assumes that they do work, and then lose the next war in a day.

Way easier with missile launched systems.  We call that "tuesday" around here.
Think about just about every rocket that's ever left the earth.  Every guided missile since what, the 1960s? Little kids can launch cameras in homemade rockets.  It's not even old technology, it's just a regular day.

PD is for that last desperate bit of defense if you have a multilayered thing that will statistically remove a huge fraction of the attackers before they get to you.  It's an officer's pistol.  
Artillery/rocket/airborne delivery of drones is a way to get lots of smaller drones farther faster than they could get on their own due to battery energy density limits.

PD is for that last desperate bit of defense if you have a multilayered thing that will statistically remove a huge fraction of the attackers before they get to you.  It's an officer's pistol.  
Artillery/rocket/airborne delivery of drones is a way to get lots of smaller drones farther faster than they could get on their own due to battery energy density limits.

Way easier with missile launched systems.  We call that "tuesday" around here.
Think about just about every rocket that's ever left the earth.  Every guided missile since what, the 1960s? Little kids can launch cameras in homemade rockets.  It's not even old technology, it's just a regular day.

Those swarms will be more effective on defense than offense because it will be much easier to lay a defensive drone field than to maintain a bubble around the attack that's resistant to massive amounts of cheap autonomous stuff.
About the best drone defense I can come up with is flying my own enormous drone army dangling bird block.  A ginormous mobile cope cage, but to stop propellers rather than shaped charges.

Aerovironment has done recon drones that launch from regular mortars.
Drone-40 puts a drone in a standard 40 mm grenade launcher package (not artillery, but accelerated with a bang)

It's not one of them doing that - a point defense system might be able to deal with that, even at crazy high speeds and accelerations.  It's a half dozen of them, communicating with each other.  On every single target.  The only way to be sure you hit them is if you're a frag grenade with a high density and uniform distribution of metal.  
eta: it's a good thing you design historical games and don't have to stick to modern games.  Because CM:Drone Wars will be a one-turn slaughterfest.

The gun launch wouldn't be pleasant, but it's a speedy delivery system and not that terrible.  I have plans to do worse to more fragile things.

It feels logical but I doubt it will cut it. Skip to minute nine in the video below and see why. Seeing that I felt the wave of sheer terror. And imagine swarms.
 

one minor quibble - if you're referring to MCAS with Boeing:  that wasn't even ML as far as I can tell.  It was a deterministic control law that got initially approved in an implementation that was probably not too horrible, then Boeing had bad configuration management and didn't review ir when it got updated to apply more extreme corrections where they would have found that it had fatal edge cases.  Bad system engineering and configuration management.

"Explainability" is a thing in AI.  Or more appropriately, Machine Learning.
When I use ML to act as a "remote graduate student" it evaluates a bunch of properties of what it's looking at that are based on real physical models, and then compares what it sees for any data set to other real human-evaluated data sets.  It then reports that "this new thing is x% like set 1, y% like set 2, etc" to make clear why it's making some recommendation.  What most of the AI chat bots (and apparently go bots) isn't this.
What a lot of the AIs that take enormous training data sets are doing is just picking the most likely move/phrase/sentence/whatever based based on a distillation of some enormous number of games.  Go is a combinatoric nightmare, so the computer has to constrain how far it looks around any point, so it makes sense that a big non-local surround would fake it out.  So it has no model, or an incomplete model.  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.

"Explainability" is a thing in AI.  Or more appropriately, Machine Learning.
When I use ML to act as a "remote graduate student" it evaluates a bunch of properties of what it's looking at that are based on real physical models, and then compares what it sees for any data set to other real human-evaluated data sets.  It then reports that "this new thing is x% like set 1, y% like set 2, etc" to make clear why it's making some recommendation.  What most of the AI chat bots (and apparently go bots) isn't this.
What a lot of the AIs that take enormous training data sets are doing is just picking the most likely move/phrase/sentence/whatever based based on a distillation of some enormous number of games.  Go is a combinatoric nightmare, so the computer has to constrain how far it looks around any point, so it makes sense that a big non-local surround would fake it out.  So it has no model, or an incomplete model.  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.

If I can attempt to refine this just a little bit, the older style of machine learning can give answers that very likely to be correct in their are of expertise. But those areas are narrow, and they take an enormous amount of carful refinement. I would argue that modern commercial jet auto pilots/flight control systems  are a version of this. Unfortunately Boeing has given us some epically bad examples of what happens when you try to half bleep this approach. 
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.
Having said that the approach can be extremely powerful when used on narrower, validated sets of information. Reading mammograms is the most common example, but there are a ton of other medical ones. A very interesting study just came out that strongly suggested AI could predict cardiac risk five and ten years out from a simple chest X-ray. Google's protein structure prediction work deserves a Nobel Prize, it completely revolutionized several scientific fields. We are just starting to feel that one.
I do have an actual relevant point...this war has generated more images of actual combat than all previous wars put together possibly. Many of those images and videos even have pretty good labeling. That is going to be used to create a database that can be used to train military AIs that I think we be rather useful. It should certainly help make some amazing games once the tech filters down.

"Explainability" is a thing in AI.  Or more appropriately, Machine Learning.
When I use ML to act as a "remote graduate student" it evaluates a bunch of properties of what it's looking at that are based on real physical models, and then compares what it sees for any data set to other real human-evaluated data sets.  It then reports that "this new thing is x% like set 1, y% like set 2, etc" to make clear why it's making some recommendation.  What most of the AI chat bots (and apparently go bots) isn't this.
What a lot of the AIs that take enormous training data sets are doing is just picking the most likely move/phrase/sentence/whatever based based on a distillation of some enormous number of games.  Go is a combinatoric nightmare, so the computer has to constrain how far it looks around any point, so it makes sense that a big non-local surround would fake it out.  So it has no model, or an incomplete model.  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.

Wait, what?
It's already been replaced, and the parking lots of old tanks just haven't been cleared out yet.  The rest of your same post that I cut from the quote pretty much summarizes it.
It's all about C4(or 5?)ISR+precision, and how the precision is delivered doesn't really matter.  Up until the mid 1970s, ISR on the ground was practically limited to LOS with binoculars plus some broad view from an airplane. In the mid 1970s, electro-optics+microelectronics began making the world, including the ground environment, more transparent and making it faster to distribute that information.  The tank remained useful because at the "guy standing in a field" scale it still could do local ISR+precision quite well and carried enough protection to last at least a few minutes on the battlefield statistically.  
Fast-forward to "I'm sitting in a home office within arms reach of a few hundred TB of storage and more computers than I can count that would have been considered supercomputers when I was in graduate school" and that's over.  High performance, low power, extremely compact computing power is essentially free.  SDR is practically free and fits on a thumb drive.  Battery tech has created bonkers energy density at low cost and high efficiency. High school kids can build and launch satellites as part of their space club.  They get capable ISR drones under the christmas tree and could build FPV drones in their aerospace club if they could get ahold of RPG-7 warheads (fortunately those aren't *that* easy to come by in the US). We're living in the future.
And unbelievably, about 10 years ago I even started having my thanksgiving menu and recipes planned out on a laptop that was in the kitchen during meal prep.

Hamas probably can't easily do anti-radiation.
A major problem with APS is the sensing - if you do active sensing with a radar, you're providing a homing signal to someone who, after a small number of attempts, can figure out how many simultaneous attackers it takes to overwhelm the APS.  You might as well paint yourself up like Daffy Duck "shoot me now! shoot me now!"
Passive sensing will give you less precision, lower SNR, and be easier to spoof once the attacker figures out the details of your camera system.  And the attacker will do it from their mom's basement BVR at no risk to themselves - just a bunch of cheap semi-autonomous munitions, while the tanks are out there getting hammered.  The problem with tanks using direct fire is range.  On a fully illuminated battlefield (full borg spotting, with partial object transparency) they become targets long, long before they get anywhere near where they will have LOS to a meaningful target.  And they're big, hot, noisy, spectrally-different-from-their-background targets.  Tanks might remain in a semi-direct fire infantry support role where they're more like flat-trajectory artillery firing over the horizon.

Current production is at levels of post WWII battleship production.  
The US produces ~130 tanks/year, mostly because we can and maybe they'll be useful some day, and it keeps some people off the street in Ohio.  But as noted by @photon regarding battleships - the US is famous for its excesses.  "we wallow in so many resources that we can commit to building ships that we'll just scuttle later to make reefs"
Russia loses more than that many per month in Ukraine.
People sailed battleships around long after they stopped being particularly useful.  Argentina even managed to get one sunk by a submarine after they invaded the Falkland Islands.
ETA: and with the levels of ISR available to the US, if the tank were that valuable to Ukraine, we'd be cranking up tank production for our own use.  There's a room full of E4s somewhere verifying tank kills that the AI has picked out from satellite imagery.

Seems like that would rate kind of high in the "bad idea" department - it would mean that an adversary that finds a hole (either technical or socially engineered) could just shut down a NATO defense.