Improvement to Recon

[captainjack]

With my over-abundance of spotting heroes, I can confirm that 4 spotting is good on a kradschutzen and that 5 spotting is pretty handy on Panzer IVG, so a recon unit with 4 (and maybe 5) spotting and recon move should be useful.

Other things that could work within the existing game rules would include increasing the move to allow two or three stops on the way. +1 move heroes and tweaks to the equipment file show that this works quite well without overpowering the units.

Other options include adding the camo trait so that it's harder to spot or to increase the GD a lot to make it more difficult to damage (I think that a small GD increase was used by the developers to make recon more attractive to use). Be warned that I have not tried the last two approaches and have no idea how successful they would be or whether they have unwanted side effects - such as getting ambushed by a camouflaged recon with GD 25.

[rezaf]

Now I am a bit puzzled. So do you want to change the recons in the official scenarios, or you want to change the official scenarios themselves, so that recons could be used as intended?

First of all, I didn't start this thread, so I can live without any changes whatsoever (which is also the most likely outcome).
That said, if anything was changed, I think it'd have to be a fairly small tweak to the stats of the recon units themselves, it's too late in the game's development cycle to do anything beyond that.
And that I still wouldn't use recon is merely personal preference - I also don't use strategic bombers (though I have a couple in the core for select missions, especially when encountering ships) even though many people praise them to the high heavens.

I do not think that the AI is very brain dead, by the way. It is truly basic on its own, but if well scripted it can perform much better.

However, AI units can be told to move to a certain part of the map and they can be told to go on the offensive when they reach it or when certain conditions are met. Or to move to another part of the map. Or whatever. It can lead to some nice AI maneuvers, I can tell.

It was definately a massive improvement when the ability to script such things was added, but the AI wasn't improved one bit. It IS braindead. What you describe here with making clever scripts and switching between them in smart ways is not the intelligence of the computer enemy at play, it's the intelligence of the scenario designer.
With enough effort, the AI can be propped up to appear smart, but in reality this requires a human puppet player pulling it's strings.

If interested, you can test it in the BE mod.

Sorry McGuba, but I can't get myself to play those scenarios on gigantic maps. Not only do I feel their scope is just not the correct one for the PzC engine, they also lead to so outrageously long AI turns...
I'm sure you did a great job, though.

It could be easily cured by making the BT-7 unavailable after 1941/42. But then it would probably purchase T-60s instead.

Of course - I was merely trying to illustrate that the AI will buy stuff other than INF, but it will usually make retarded choices. As you say yourself, the logic that determines what it goes for is fundamentally flawed - in the early stages of the vanilla campaign, this leads to massive frustration due to all the filler units that at this point still are a force to be reckoned with, when the player has only fairly inexperienced units with relatively low power.
In the late game, it merely leads to "free XP" - low quality units the player can brush aside.
The only exception that comes to mind are the very last scenarios in the eastern branch, where the prestige grants given to the AI sometimes are so huge, it'll buy an IS-3 or something.
But even though that's a force to be reckoned with, it doesn't really make much of a difference in scenarios where a hundred russian tanks need to be dealt with...

I do not allow the AI to buy anything by setting the max number of AI units to zero in the editor.

Heh, that's sure an approach, but essentially, this means you're just disabling the buying AI alltogether. Sadly, I guess it's the best approach. In fact, I wonder why the official scenario designers didn't come to the same conclusion. Other than those early cases I described, where it just serves to annoy the player, what good do the AI's purchases ever do?
_____
rezaf

[BiteNibbleChomp]

No, it should be artillery, it is always good to have an extra piece of gun! Anything familiar?

WWI! Artillery Soft Attack 25+

- BNC

[McGuba]

Other than those early cases I described, where it just serves to annoy the player, what good do the AI's purchases ever do?

It certainly does not do any good and I have not heard of anyone saying: "oh, I love those silly AI purchases, it just so cheeky and funny!" Why the scenario designers left it like that is a good question, we should have to ask them, but nowdays there is very little to none communication between the designers and the players, so we might never get an answere to it. And even if we got an answere it would not make any difference as it is like that and will surely stay like that for game balancing reasons. (All the campaigns were tested with this feature on.)

It was definately a massive improvement when the ability to script such things was added, but the AI wasn't improved one bit. It IS braindead. What you describe here with making clever scripts and switching between them in smart ways is not the intelligence of the computer enemy at play, it's the intelligence of the scenario designer.

Obviously I see your point here, but it leads us to the very interesting question of "what is intelligence?" Can a person, who never goes to school, never learns the things, ever be intelligent? Intellingence is surely also about learning, years and decades of learning, and not just something we get when we born. I am not very familiar with the current stage of artificial intelligence development, but I am not sure that we will ever be able to create an artificial mind similar to ours. As far as I know all we can do currently is some kind of simulation, to make it appear smart, just like you wrote.

Let's get to the chess simulators. I am aware that nowdays there are some very nice chess programs which can beat even the best human players. However, it took long decades to create these programs and they does not work like human thinking at all. Deep Blue used "brute force" which means it examined all the millions of possible move combinations in seconds and only chose the most promising one. It is not really intelligence, as human chess players do not even waste their time think about silly moves, which make up the large majority of the possibilities. But if we compare chess to PzC, for example, I have to come to the conclusion that PzC is much more complex with all the different terrain types and units, weather, differing objectives, uncertain dice rolls, etc. etc. On the contrary, chess has its set rules and limited "unit types" on a very small "map" with no terrain differencies. I think that making a good AI for chess should be far easier than for complex wargames. And brute force could not help here as the amount of data that should be considered is virtually endless. Thus I do not think that the AI of these wargames can be made considerably better in the foreseeable future. Which leaves us not much more but overpowering the AI and / or using scripts to make it more challenging. Basically that's why I think that the creator(s) of PzC made a fairly decent job: we have a basic "brain dead" AI, only capable of following the most basic rules, but we have the possiblity to improve it with the help of scripting. From then it is up to the scenario designer.

However, I do not say it could not be improved in some ways. For example, it should not "forget" unseen enemy units in the fog of war, it is indeed silly when it tries to repeatedly attack enemy units defended by unseen artillery. It should only attack once and then the defending artillery should be remain visible in the fog of war for the remainder of the AI turn. It could be coded in easily, I guess. Also it should take care of its units with low strength and retreat and reinforce them instead of just pressing the attack with weakened units. But I do not think that it will ever be able to make complex decisions on how to use this or another unit as it is difficult for even humans as well.

[rezaf]

And even if we got an answere it would not make any difference as it is like that and will surely stay like that for game balancing reasons. (All the campaigns were tested with this feature on.)

Sure, I agree. Let's see how things are handled in Armageddon, that ought to give us an idea about how future PzC content might look (in other words, a sequel, if there ever is one).

Obviously I see your point here, but it leads us to the very interesting question of "what is intelligence?"

Heh, dear McGuba, there's no need to get philosophical here. I'm aware we're unlikely to have actual artificial intelligence anytime soon, but I was using the term in the sense it has come to be used in computer gaming as opposed to the literal meaning of the words. It might come as a shock to you, but baby oil isn't actually made of squashed babies either.
So, what I meant was that if the AI acts smart, this happens within the set of rules usually created by the developer, and if this fails to happen and a scenario designer must put up a smokescreen and specifically overwrite or otherwise influence the decisions of the AI, that's the scenario designers human intelligence at work.

That the PzC AI is so braindead is the case because Rudankort didn't put enough effort into this part of the game, or maybe creating a better AI was just beyond his skill level. That's no skit at Rudankort, that stuff is hard and only the very best AIs are able to put up a fight on even terms. Just look at Civ5, where the designers failed to enable the AI to move it's units in a coherent fashion - even after numerous patches and two expansions.
While it isn't possible to cover ALL the bases like in a chess brute-force AI, a computer game AI works kinda similarly. A ton of variables are evaluated and then a "decision" is made. Like, when moving an aircraft, this could be: How much fuel is left? How much ammo (can I attack at all)? Is there a suitable target in range? What are my combat odds? Any target with better odds? And so on.
An entry-level AI would feature only a very limited number of such rules, and our airplane might see it has the best odds whe making a strafing run on an infantry unit and be oblivious to the "bigger picture", like that there are two other fighters near which will get to act immediately afterwards, and if all three would attack that lone Bf109, there'd be a 87% chance of destroying it ... sounds familiar?

So, I think improving the AI of any game is very possible by providing it with a larger stack of rules. In the end, these rules come down to pre-defined scripting the AI automatically applies as needed. As opposed to the scenario designer having to check the circumstances on his own and dictate the AI behavior accordingly. Some games come with scriptable AI where you can download such rulesets and see how it's done yourself - PzC could have been MASSIVELY improved by scriptable AI as well. Once again, maybe Armageddon will show positive developments in this are, maybe not - we'll see.
_____
rezaf

[Egge]

Would it not be simplest to just let Recon units have an additional +1 to spotting?

I second that. Possibly even add +2 to spotting and leave everything else as it is. Right now, recon units are absolutely useless. It's impossible to keep them alive in the front line, you have to keep them a little behind your spearheading tanks. If you do that, however, if you keep your recon units one hex behind your tanks, your recon units cannot spot any further than your tanks can. As it is, there is no reason to use them.


Another reason why they are so weak is that ground defence is by far the most important stat and recon units have poor ground defence. Don't know how to change that easily though.

[edahl1980]

I can agree to giving recon more spotting power.

Matter of fact, i made myself a hero.
Kind of a general with a command vehicle(recon). If that unit dies, the game is over.
+2 atk
+2 def
+1 ini
+1 mov
+2 spot

The way i use other recon is i move them up, scout, move them up, scout, move them up. Then i can move the tanks without worrying about being ambushed. And after i move the recon back.

[morge4]

Well, there's OOB: Pacific in beta now, which is of course not *my* project, but I'm involved in the gxf part of this game.

In the world of Panzer Corps I'm just a modder

And I'm really enjoying beta testing OOB - Pacific. Great graphics and nice engine. Thanks Bebro!

[captainjack]

Don't know how to change that easily though.

The following instructions are repeated elsewhere in the Scenario Design forum. This process works when I do it.

The equipment editor file is in the Data folder.

FIRST - make a copy of the original Equipment_Original or similar filename.
Second - open the equipment file. Some people use Notepad, but I prefer using Excel.
Lock the headings bar.

Scroll down to a recon unit you want to use. Change the value for that unit.
Press save and press Yes to whatever it says.
Then close the file using the red box in the top right corner and don't save.
This preserves the formatting.

The changes will not take effect until you open a new game (they might not until you open a new campaign, so play around with a starting scenario to see how it works.

Be warned that it is best to change 1 thing at a time and to be very methodical and to keep that original equipment file safe and secure.

You are now on the first step of the slippery slope to becoming a modder! You can now spend endless hours reading the scenario design pages (most of which are really good).

[edahl1980]

What is OOB?

[Tarrak]

What is OOB?

OOB stands for Order of Battle: Pacific. It's a round based strategy game being developed by the the artistocrats and published by Slitherine. It's currently in Beta. You can check this Forum for further details.

[ThvN]

Now that I finally have some time to go back to modding, here's an idea that I have tested to make recon seem better: reduce spotting for tanks. ( /ducks to avoid incoming fire )

No, really, I made a simple rule for myself when assigning spotting ranges, and tanks (and other heavy armour) normally only get 1 now. Infantry gets 2, just like towed units. Recon stays at 3. Before my little hiatus, I was busy testing if I will allow some units, like 'open-topped' vehicles and light tanks (Pz I & II) to have 2 spotting, this will help keep weak units more interesting, but I haven't decided yet.

Let me say this change makes for some tense battles. Previously, all you needed was a screen of tanks to see far enough to avoid most suprises. Not anymore. You can't see anything unless you use other assets, and the AI also makes bolder attacks with its tanks, because it can't spot any supporting units.

So far, recon has become a bit more useful, but I'm not sure yet if it will work well in later war years. It might be too easy to fool the AI into attacking, which is also the reason I haven't tried the camo trait yet. I'm going to take my time, I already had to abandon a project when it turned out the AI cannot switch airplanes and started acting weird; might be a bug? Testing showed other issues: if anyone is making submarines capable of firing against airplanes, you might want to check what stat the airplane uses to defend itself with... it seems it uses its Close Defense in stead of Ground Defense?

Oh, a strange bonus byproduct of the 'less spotting for tanks' system is that a spotting hero for a tank actually becomes somewhat useful .

[captainjack]

My general preference would be for pushing up the recon move by 1 or 2, to allow a bit more actual scouting and getting into and out of tight spots, but 1 spotting for tanks does make a fair bit of sense, and it certainly would make recon more useful. I like the idea of allowing the possibility of 2 spotting for some lighter vehicles to make them a bit more useful - as it could if there were any vehicles with a reputation for unusually good visibility.

And yes (shame on me) I did - very briefly - think about suggesting 0 spotting for some of the French tanks with poor visibility and an overworked commander, and I did wonder if you could give a unit the trait to treat everything as camouflaged....

[Razz1]

What I have done is add Recon planes to the Total Realism Mod.

[Egge]

Don't know how to change that easily though.

The following instructions are repeated elsewhere in the Scenario Design forum. This process works when I do it.

The equipment editor file is in the Data folder.

FIRST - make a copy of the original Equipment_Original or similar filename.
Second - open the equipment file. Some people use Notepad, but I prefer using Excel.
Lock the headings bar.

Scroll down to a recon unit you want to use. Change the value for that unit.
Press save and press Yes to whatever it says.
Then close the file using the red box in the top right corner and don't save.
This preserves the formatting.

The changes will not take effect until you open a new game (they might not until you open a new campaign, so play around with a starting scenario to see how it works.

Be warned that it is best to change 1 thing at a time and to be very methodical and to keep that original equipment file safe and secure.

You are now on the first step of the slippery slope to becoming a modder! You can now spend endless hours reading the scenario design pages (most of which are really good).

I know how to change unit stats. What I wanted to say was that ground defence is arguably the single most important unit stat. If it were less important Recon units with their low ground defence might not suck as much as they do now. However, what I don't know is how to change the game in such a way that ground defence becomes less important.

[captainjack]

how to change the game in such a way that ground defence becomes less important.

Apologies for misunderstanding what you were asking. It sounds like this could be something to raise on the recent Panzer Corps 2 wish list thread.

[Egge]

no harm done, captainjack. Could have made my point clearer.

[BiteNibbleChomp]

( /ducks to avoid incoming fire )

Would that happen to be coming from a WWI 6-range artillery? Better recon could be used in such a battle, thats why I provided the zeppelin.

- BNC

[ThvN]

( /ducks to avoid incoming fire )

Would that happen to be coming from a WWI 6-range artillery? Better recon could be used in such a battle, thats why I provided the zeppelin.

- BNC

With the weather as bad as I encountered I still prefer to use it as an umbrella. One thing I am wondering: If (when) you continue with your WW1 Mod, how much range is this going to get?

[proline]

Now that I finally have some time to go back to modding, here's an idea that I have tested to make recon seem better: reduce spotting for tanks. ( /ducks to avoid incoming fire )

No, really, I made a simple rule for myself when assigning spotting ranges, and tanks (and other heavy armour) normally only get 1 now. Infantry gets 2, just like towed units. Recon stays at 3. Before my little hiatus, I was busy testing if I will allow some units, like 'open-topped' vehicles and light tanks (Pz I & II) to have 2 spotting, this will help keep weak units more interesting, but I haven't decided yet.

Let me say this change makes for some tense battles. Previously, all you needed was a screen of tanks to see far enough to avoid most suprises. Not anymore. You can't see anything unless you use other assets, and the AI also makes bolder attacks with its tanks, because it can't spot any supporting units.

So far, recon has become a bit more useful, but I'm not sure yet if it will work well in later war years. It might be too easy to fool the AI into attacking, which is also the reason I haven't tried the camo trait yet. I'm going to take my time, I already had to abandon a project when it turned out the AI cannot switch airplanes and started acting weird; might be a bug? Testing showed other issues: if anyone is making submarines capable of firing against airplanes, you might want to check what stat the airplane uses to defend itself with... it seems it uses its Close Defense in stead of Ground Defense?

Oh, a strange bonus byproduct of the 'less spotting for tanks' system is that a spotting hero for a tank actually becomes somewhat useful .

Thanks for exploring this! I do wonder how the AI would handle it though- to spot defensive artillery they'd need nearby recon, and they aren't that bright...

This makes me wonder- back in PGII or something like that wasn't there a rule that if a unit attacks all infantry in range will shoot at it? That meant that defensive artillery fire didn't necessarily come from right beside the unit being attacked, and perhaps made scouting further back more valuable. Of course such a chance would break the current balance of PC.