How Hot is Ukraine Gonna Get?

[dan/california]

3 minutes ago, Battlefront.com said:

Not yet, for sure.  But under ideal circumstances (good condition Federal highways with decent weather) my impression is they are already pretty damned good.  Where they fall down, rather badly, is in complex dense areas.  There's just too much stuff to process too quickly.  For some reason even moderately competent drivers are able to handle it, not so much AI.

Which gets back to my point about a universal predictive model for warfare.  I think we can get excellent modeling for limited, carefully controlled battlefield environments.  I'd say Combat Mission fits that quite well. 

I agree with The_Capt though.  If we can get a predictive tool that can get even over the 50% mark, that would be quite valuable.  I think that's possible.  Predictive modeling for controlled battlefield environments can definitely be much better than that.  Again, I say CM is excellent at this. As long as CM is asked a question it is capable of answering, it could maybe get a customer into the 85% territory.  My professional bias aside, I really do believe this to be true.

Steve

Even the tiniest of bones about the next game you might want to share?

[Battlefront.com]

3 minutes ago, dan/california said:

Even the tiniest of bones about the next game you might want to share?

Sure, it will be better than the last one

Steve

[dan/california]

2 minutes ago, Battlefront.com said:

If they are talking about multiple batteries delivered quickly, my guess is trailer based.  And that would work pretty well for rear areas that need protecting.  Russia's ability to hit a stationary target of this size is about zero, so they're safe.  Within range of Russian artillery is a different story.

Steve

Yes, but if Ukraine had enough mobile systems to cover one decently sized offensive effort they would already have besieged Nova Khahova and cut the Russian forces on the west side in half. And short of my rather extreme suggestion above the best thing for every civilian in Ukraine is to win the bleeping war.

[dan/california]

4 minutes ago, Battlefront.com said:

Sure, it will be better than the last one

Steve

At least Steves sense of humor is holding up and he hasn't said that the next version will be national security customers only. That is something, I guess. Did you ever try to get ahold of the Ukrainian MOD. I mean you already have maps built for a fair bit of the place. 

Edited July 30, 2022 by dan/california

[Huba]

2 minutes ago, Battlefront.com said:

If they are talking about multiple batteries delivered quickly, my guess is trailer based.  And that would work pretty well for rear areas that need protecting.  Russia's ability to hit a stationary target of this size is about zero, so they're safe.  Within range of Russian artillery is a different story.

Steve

That exactly. What I'm hinting at is how useful 2 batteries would really be for protecting the country from air/ cruise missile attacks, and IMO the answer is: marginally. Perhaps it would be enough to protect Kyiv. Or Kharkhiv. Or any other local area or two, but in grand scheme of things it's impact would be mostly psychological/ morale boosting. For comparison, Poland's A2A modernization plan envisions 19 CAMM-ER batteries, which are more capable, and have to protect 3 times smaller territory.

OTOH, for creating an AA umbrella over combat zone, 2 batteries sound OKish, it should be enough to protect all the forces in Donbas...

 

3 minutes ago, dan/california said:

If anybody involved has any sense it will be for as much of both as they can get, and as fast as they can get them. Does anybody know if the missiles themselves have to be modified from the air launched versions? If not They could be pulled Existing NATO stocks, This is another factory that needs to be copied four times over, and run 24/7. It is looking like Taiwan will need an infinity of them, too.

The missiles are exactly the same - Russia will run out of cruise missiles/ aircraft before UA runs out of AIM-120, that's for sure

[OldSarge]

11 minutes ago, Battlefront.com said:

If they are talking about multiple batteries delivered quickly, my guess is trailer based.  And that would work pretty well for rear areas that need protecting.  Russia's ability to hit a stationary target of this size is about zero, so they're safe.  Within range of Russian artillery is a different story.

Steve

Has there been any indication which version these will be, the NASAMS-2 using the AIM-120 (30lm range) or the NASAMS-3 using the AMRAM-ER (50 km range)? The usual sources aren't providing much in the way of specifics.

[Huba]

2 minutes ago, OldSarge said:

Has there been any indication which version these will be, the NASAMS-2 using the AIM-120 (30lm range) or the NASAMS-3 using the AMRAM-ER (50 km range)? The usual sources aren't providing much in the way of specifics.

I'd say NASAMS-3 if it's to be a new build, but that's only regarding the electronics/ comms/ general configuration. For missiles, I'm not sure if there is even any number of the ER missiles produced, and if yes I'm pretty sure these were exported to the operator country. Regular  AIM-120 will be the main effector in this system, due to huge existing stock,that is a given.

 

Edited July 30, 2022 by Huba

[dan/california]

12 minutes ago, Huba said:

That exactly. What I'm hinting at is how useful 2 batteries would really be for protecting the country from air/ cruise missile attacks, and IMO the answer is: marginally. Perhaps it would be enough to protect Kyiv. Or Kharkhiv. Or any other local area or two, but in grand scheme of things it's impact would be mostly psychological/ morale boosting. For comparison, Poland's A2A modernization plan envisions 19 CAMM-ER batteries, which are more capable, and have to protect 3 times smaller territory.

OTOH, for creating an AA umbrella over combat zone, 2 batteries sound OKish, it should be enough to protect all the forces in Donbas...

 

The missiles are exactly the same - Russia will run out of cruise missiles/ aircraft before UA runs out of AIM-120, that's for sure

What he said, exactly, What they need to protect is a war winning offensive. It will take years to produce enough to cover most of the high density civilian areas, unless the Norwegians have four or five separate miracles up their sleeves. At least here will be ammo.

[Huba]

14 minutes ago, dan/california said:

What he said, exactly, What they need to protect is a war winning offensive. It will take years to produce enough to cover most of the high density civilian areas, unless the Norwegians have four or five separate miracles up their sleeves. At least here will be ammo.

From what I understand, the main strength of systems like NASAMS or Sky Sabre in the context of UA battlefield is the fact that active-homing missiles should be able to follow and acquire as targets all these low flying aircraft/ helicopters. It isn't possible with SARH or command guided missiles, as the line-of-sight from the guidance radars is there for only a moment when aircraft does the "toss". In this situation, AIM-120 should be able to be launched at the target and easily home on it from above. It would be really helpful if some of the assorted radars would be mounted on extendable masts, AFAIK this is a typical element of NASAMS setups. That would change the game completely, making main RU Air Force tactic no longer viable.

Edited July 31, 2022 by Huba

[dan/california]

I suspect this eagle and the Russian army could have a long conversation about biting off more than you can chew. 

[danfrodo]

44 minutes ago, dan/california said:

I suspect this eagle and the Russian army could have a long conversation about biting off more than you can chew. 

I guess the eagle is counting on the goat to beat himself to death?  But does the eagle know that means him too?  I am betting on the goat's bones being a lot stronger than the eagle's.  I could just picture Putin inside the eagle's head - Ha!  Going perfectly according to plan!  I have bruised the goat's ribs at only the cost of my own broken ribs and broken wings!   Start planning the victory parade!  I'll arrive on foot, by the way."

'course, the goat could just lay down on top of the eagle until the eagle suffocates, so aint neither of them geniuses.

[G.I. Joe]

9 hours ago, danfrodo said:

My favorite line for humans not properly assessing risk is "obese, sedentary, diabetic, non-seatbelt wearing middle-aged smoker very concerned about health effects of vaccines"

(edit: not to get all holier than thou, as I am drinking my 2nd diet coke of the day and god knows that can't be good for me)

Oh, that's awesome... Or, conversely:

"My definition of an optimist has to be the F-104 pilot who gave up smoking!"

-John Wiley

[sross112]

2 hours ago, Battlefront.com said:

Not yet, for sure.  But under ideal circumstances (good condition Federal highways with decent weather) my impression is they are already pretty damned good.  Where they fall down, rather badly, is in complex dense areas.  There's just too much stuff to process too quickly.  For some reason even moderately competent drivers are able to handle it, not so much AI.

Which gets back to my point about a universal predictive model for warfare.  I think we can get excellent modeling for limited, carefully controlled battlefield environments.  I'd say Combat Mission fits that quite well. 

I agree with The_Capt though.  If we can get a predictive tool that can get even over the 50% mark, that would be quite valuable.  I think that's possible.  Predictive modeling for controlled battlefield environments can definitely be much better than that.  Again, I say CM is excellent at this. As long as CM is asked a question it is capable of answering, it could maybe get a customer into the 85% territory.  My professional bias aside, I really do believe this to be true.

Steve

I've been following this back and forth on computer modeling and AI predictive tool stuff the last couple days. I just don't see how a computer could manage to do this sort of thing. Think about how much data you would have to put in on all the different factors and variables. Take this current war and try to put in all the data and variables that have already happened to bring it to the point it is right now. Even if you knew and were able to quantify everything that would have to be put in to get the exact model of everything that has happened so far you'd need such massive resources it is just not practical. For future predictive stuff, how could you possibly model the impact of the Audie Murphys and Chesty Pullers? The computer will tell you every time that a well entrenched chain of mg nests will stop that infantry platoon, LT William Deane Hawkins and many others like him would disagree. How would a computer model ever be able to quantify the individual valor, tenaciousness or audacity that has turned the tables, overcome the odds and thwarted the perfect plans in so many wars? At best a computer is going to be able to give us the same answer that so many pundits and intel specialists did before this current war. They, just like a computer, could never factor in the actions of men like Ukrainian Marine  Vitaly Skakun Volodymyrovych. 

You can't just look at the numbers and equipment and get a good model either. A computer would have predicted the Char B's slaughtering the inferior German tanks and France winning in 40. How can it quantify the Lafayette Pools, Kurt Knispels or the numerous successful UA tanks that actually started this war as RA tanks? How would the data to plug into the computer even be accurately obtained given the huge amount of variables? A piece of equipment in one person's hands can never be quantified the same for any other person. Simo Hayha with a M/28-30 is totally different than me with one (he would be considerably better for those that are wondering). So every piece of equipment relies on the human factor and on top of that you have to accurately model the leadership, logistics and political influences along with somehow figuring out the average amount of Sacrifice for the individual soldier, civilian and collective nation. How would a computer ever be able to model the 101st at Bastogne or the 1st Marine Division at Frozen Chosen? Or better yet the VC or Taliban? 

Even with tons of data and mountains of computing power I don't see how the predictions could ever get better than the CEP for the RA's 1 week wonder 152mm crews, definitely not a chance of HIMARS accuracy or precision. 

[Calamine Waffles]

29 minutes ago, sross112 said:

I've been following this back and forth on computer modeling and AI predictive tool stuff the last couple days. I just don't see how a computer could manage to do this sort of thing. Think about how much data you would have to put in on all the different factors and variables. Take this current war and try to put in all the data and variables that have already happened to bring it to the point it is right now. Even if you knew and were able to quantify everything that would have to be put in to get the exact model of everything that has happened so far you'd need such massive resources it is just not practical. For future predictive stuff, how could you possibly model the impact of the Audie Murphys and Chesty Pullers? The computer will tell you every time that a well entrenched chain of mg nests will stop that infantry platoon, LT William Deane Hawkins and many others like him would disagree. How would a computer model ever be able to quantify the individual valor, tenaciousness or audacity that has turned the tables, overcome the odds and thwarted the perfect plans in so many wars? At best a computer is going to be able to give us the same answer that so many pundits and intel specialists did before this current war. They, just like a computer, could never factor in the actions of men like Ukrainian Marine  Vitaly Skakun Volodymyrovych. 

You can't just look at the numbers and equipment and get a good model either. A computer would have predicted the Char B's slaughtering the inferior German tanks and France winning in 40. How can it quantify the Lafayette Pools, Kurt Knispels or the numerous successful UA tanks that actually started this war as RA tanks? How would the data to plug into the computer even be accurately obtained given the huge amount of variables? A piece of equipment in one person's hands can never be quantified the same for any other person. Simo Hayha with a M/28-30 is totally different than me with one (he would be considerably better for those that are wondering). So every piece of equipment relies on the human factor and on top of that you have to accurately model the leadership, logistics and political influences along with somehow figuring out the average amount of Sacrifice for the individual soldier, civilian and collective nation. How would a computer ever be able to model the 101st at Bastogne or the 1st Marine Division at Frozen Chosen? Or better yet the VC or Taliban? 

Even with tons of data and mountains of computing power I don't see how the predictions could ever get better than the CEP for the RA's 1 week wonder 152mm crews, definitely not a chance of HIMARS accuracy or precision. 

There is a good paper that just came out by B. A. Friedman about war and non-linear dynamics:

https://digital-commons.usnwc.edu/nwc-review/vol75/iss2/5/

[chrisl]

Looks like someone at the NYTimes (or maybe DOD and it's just sinking in at the NYT) finally is catching up with this thread.  Depending on how much you've looked at the times this month it may be paywalled, so pulling a few quotes:

Quote

The official described Russia’s recent efforts as a failure both on the battlefield and at home, where Moscow’s rhetoric about its ambitions in Ukraine has grown more bombastic in recent days. 

Quote

“Throughout July, the occupiers have been trying to storm Donetsk region,” Serhii Haidai, the head of the military administration in neighboring Luhansk Province, said in a statement. But unlike with the Russians’ push in the spring and early summer, when they could use their significant artillery advantage to flatten areas before advancing, Mr. Haidai said the Ukrainians’ destruction of Russian ammunition depots had “made it much more difficult for them to replenish arms stocks and maneuver.”

 

[dan/california]

38 minutes ago, Calamine Waffles said:

There is a good paper that just came out by B. A. Friedman about war and non-linear dynamics:

https://digital-commons.usnwc.edu/nwc-review/vol75/iss2/5/

The_Capn I am only on page four of this paper but I am absolutely sure you want to read. So does Steve.

[Battlefront.com]

41 minutes ago, sross112 said:

I've been following this back and forth on computer modeling and AI predictive tool stuff the last couple days. I just don't see how a computer could manage to do this sort of thing.

Here's how it would work...

1.  Set up a tactical environment that can work with any type of force or environment

Picture Combat Mission.  It's got a lot of hard data in it as well as vetted physics based modeling.  This means an Abrams behaves like an Abrams, complete with down range effects of its weaponry.  This stuff is not easy to do, but hey... we did it, so give us a couple hundred million and we can do it again no problem

For the physical environment we have to fudge stuff because there's only some physics involved and only a little bit of hard data.  However, it's a pretty "known" environment that can be fairly easily tested to ensure that the more known stuff (see above Abrams) appears to interact correctly.

Then we get into the squishy stuff known as "soft factors".  Experience, morale, leadership, physical fitness, etc. are all critically important, yet there's just about no hard data and almost no physics equivalent behaviors to transform into vetted equations.  However, like the physical environment there's enough literature and first hand information to massage the modeling into producing reasonable results.

And finally, we need to have autonomous simulation of Humans on the battlefield, but as individuals and as part of groups.  This is the trickiest of all things to do, but like the previous two things there are plenty of ways to know if the decision making (aka TacAI) is producing reasonably good results consistently across the board.

That's pretty much it for the tactical basis of the simulate.  You all have had access to this for relatively cheap money for years now, so think of what could be done with huge budget and no concerns about commercial viability.

OK, so what's next? 

2.  Larger scope of operations

You'd need to apply the same sort of rigor as with the tactical environment to things which support tactical operations.  Things like logistics, asset management, resource allocations, etc.  All totally doable and feature for feature relatively straight forward.  The challenge is the quantity of things that need to be simulated and having them interact with each other in a way that is similarly vetted as with the tactical environment.

3.  Autonomous higher level AI players

Ideally you'd have a really super duper AI player (aka StratAI) that behaves just like a single battlefield commander of a particular level.  This is doable, but it is extremely expensive.  We're talking a few million bucks and years of development to really get it solid.  However, it can be done.

Then you need a slightly different AI player (aka OpAI) that is capable of higher level decision making which feeds instructions to the StratAI.  Again, totally doable but very, very expensive to do.

This could be skipped over and have Humans provide this sort of decision making instead of it being automated, however this would significantly slow down producing the number of iterations necessary to get a solid range of results.

4.  Non-battlefield factors

This is the really tricky one.  Things that influence the war effort and are influenced by it need to be simulated.  Population will to fight, industrial capacity, rates of replacement, means of short cutting normal procedures, monetary budgets, political considerations, friction within the chains of command, etc.  This is highly subjective BUT feature for feature can be broken down and be reasonably modeled as with soft factors at the tactical level.

Whatever the huge budget for this simulation is, here's where a big chunk of the money would be spent.  And that's presuming that Humans would role play the various aspects and input values to influence the simulation.  Trying to have this be fully autonomous would likely cost too much and would likely be significantly inferior to having Humans do it.  Plus, it's eminently practical to have Humans provide this sort of input when talking about a professional simulation.  Unlike #3 above, this wouldn't slow down iterative testing that much because these sorts of factors don't change simulated hour by hour, or even necessarily day by day.  The sim could, therefore, do a lot before needing Human input.  And when needed, the input would be rather quick to get into the system and have it continue to chug away.

 

And there you go.  A system that can theoretically produce quantifiable, science based predictive modeling of any potential conflict.  A few hundred million Dollars and 10 years should wrap this up quite nicely

Steve

[chuckdyke]

8 minutes ago, Battlefront.com said:

A few hundred million Dollars and 10 years should wrap this up quite nicely

As usual money rules.

[Battlefront.com]

45 minutes ago, chrisl said:

Looks like someone at the NYTimes (or maybe DOD and it's just sinking in at the NYT) finally is catching up with this thread.  Depending on how much you've looked at the times this month it may be paywalled, so pulling a few quotes:

 

Not just NY Times, but also Washington Post:

Quote

Ukraine could be turning the tide of war again as Russian advances stall

...

Rather, Barros and many Western officials and analysts suspect that the Russians are close to exhausting their capacity to make further territorial gains as their depleted army confronts Ukrainian forces with newly acquired capabilities. Already forced to abandon their hopes of capturing the capital, Russian forces may soon have to reckon with their inability to conquer the entirety of the Donbas region — the only publicly declared goal of the initial invasion and the focus of current offensive ambitions.

...

“The Ukrainians have changed the character of the conflict with their ability to attack behind Russian lines,” O’Brien said. “What we will see in the next few months is the initiative swing back to the Ukrainians, and then we will have to see if the Ukrainians can push the Russians back.”

https://www.washingtonpost.com/world/2022/07/28/ukraine-russia-war-himars-missiles/?utm_campaign=wp_post_most&utm_medium=email&utm_source=newsletter&wpisrc=nl_most&carta-url=https%3A%2F%2Fs2.washingtonpost.com%2Fcar-ln-tr%2F378187e%2F62e40194cfe8a216010cd546%2F5b6a1f5bade4e277958a3cb5%2F15%2F70%2F62e40194cfe8a216010cd546&wp_cu=e17d566bfc90044414eb36edd5675ce3|72E0F70E3B4B42CEE0530100007F2E01

The article ended with a note of caution (I'm OK with that), but overall I haven't seen these two big publications put out positive forward looking articles in a while.

Steve

[billbindc]

5 hours ago, Battlefront.com said:

Not yet, for sure.  But under ideal circumstances (good condition Federal highways with decent weather) my impression is they are already pretty damned good.  Where they fall down, rather badly, is in complex dense areas.  There's just too much stuff to process too quickly.  For some reason even moderately competent drivers are able to handle it, not so much AI.

What I have learned from very deep dives into self driving AI is that human beings are insanely good at pattern recognition. I am pretty sure that cars, war and everything else that is not a closed system will remain human dominant for a long time to come.

[Battlefront.com]

11 minutes ago, chuckdyke said:

As usual money rules.

Yup.  This is an engineering problem and money is usually the primary means of addressing it.  It's how we went from V2 rockets to putting men on the moon.  Vision, will, and time couldn't have done it without lots of money.

Steve

[chuckdyke]

1 minute ago, Battlefront.com said:

Vision, will, and time couldn't have done it without lots of money.

I like to see an expanded manual. Which explains everything more in depth. Artillery for example it is the same tool whether we play Red Thunder Soviet Campaign or the US Campaign in Black Sea. I would like to see some suggestions for protocols for the player to use. Kind regards

[Battlefront.com]

1 minute ago, billbindc said:

What I have learned from very deep dives into self driving AI is that human beings are insanely good at pattern recognition. I am pretty sure that cars, war and everything else that is not a closed system will remain human dominant for a long time to come.

It's not just Humans.  I had three juvenile ducks snoozing in front of me the other day.  Mom was there watching out for predators, and although she kept an eye on me she didn't view me as one.  Then she took off for some reason and was gone for a few hours.  Miles away for all I know.  Then I heard some quacking and she landed around the corner from her brood.  She then paddled around several visual obstruction and hopped out of the water right next to her ducklings.

Imagine the amount of AI programming and technology that would be needed to replicate this simple decision making, not to mention getting something to fly, swim, and walk as part of its tasks.

Yup, computers are neat, but I've seen more intelligence in insects.  And AI programmers say the same thing, so it's not just me

Steve

[Calamine Waffles]

Every party has a pooper, that's why we invited you, party pooper
 

 

[Battlefront.com]

Just now, chuckdyke said:

I like to see an expanded manual. Which explains everything more in depth. Artillery for example it is the same tool whether we play Red Thunder Soviet Campaign or the US Campaign in Black Sea. I would like to see some suggestions for protocols for the player to use. Kind regards

The ability to hit what is being aimed at, the rate of fire, the amount of boom per shot, etc. may be different but the basics of artillery are still the same.  Even PGMs doesn't change things all that much in ways we care about from a game standpoint.

The big difference in the real world is how national doctrine for how artillery is used.  That, unfortunately, is hard for us to control because it's up to the Human Player to do most of that sort of stuff.

Steve