AIs and the Art of War

Will the bandit line also get a reduction? They are part of the crime problem and if you weaken the crime fighters you should also need to weaken the criminals. They are also not military trained.

Edit thinking a bit more on this

Dancing Hoskuld said:

As we have discussed elsewhere - depending on the combat type or unit type - units should only get a maximum exp from Military Instructors just because they aren't military.

I would say the first 3-6 Military Instructors give full exp to all units. More than that and they only give exp to military units.

Great Artists would give exp to Story Teller line.
Great Doctors to healer line.
Great Engineers to workers.
Great Priests to inquisitors (and missionaries?)
Great Spies to espionage units
Great Merchants to caravans and trade ships (increases the returns from those units)
Great Statesmen to diplomats

and so on.

Click to expand...

To make it a bit easier to implement

GG's can become Instructors (max 3-6) or Military Instructors. Instructors give exp to all units Military Instructors give exp only to military units. Splitting it into two specialists just makes it easier to see what is going on. We even have art we could use - the green general for Military Instructor and the brown one for Instructor. And if SO was hoping to use the brown one as a post build trainer of units then the drill sergeant might do for that and looks more right for the function.

This thread is filled with good ideas!

But speaking of specialists, and bringing things back to part of my initial premise that the crime from slaves is part of what is then forcing the AI to build Town Watch and getting them stuck in a vicious cycle, are there any limits at the moment on how many slaves a city can have? If not, do you think one might help?

Arkenor said:

This thread is filled with good ideas!

But speaking of specialists, and bringing things back to part of my initial premise that the crime from slaves is part of what is then forcing the AI to build Town Watch and getting them stuck in a vicious cycle, are there any limits at the moment on how many slaves a city can have? If not, do you think one might help?

Click to expand...

Of the two types of slaves, you can have 12 population as slaves and unlimited settled slaves. This will be slightly different in the mod mod I will be working on to specialize the slaves a bit.

I put the crime on slaves even tough I think it needs to be a value that changes with technology. At first there just is no scope for slaves to do more criminal activity than is already in the system but later on as the people become richer there does become the scope for slaves to do crime above what the normal populace do.

I did suggest that I remove crime for now but I think Koshling wanted to see what was actually happening so that he could adjust the AI to cope with it. That would be the best solution

Dancing Hoskuld said:

Will the bandit line also get a reduction? They are part of the crime problem and if you weaken the crime fighters you should also need to weaken the criminals. They are also not military trained.

Click to expand...

They are already weaker than their contemporaries (thieves at 1 when you have clubmen, rogues at 2 when you have AtlAtls etc.). The town watch line are much closer to the military in strength than the bandits are, so UI think this is already there.

Dancing Hoskuld said:

Of the two types of slaves, you can have 12 population as slaves and unlimited settled slaves. This will be slightly different in the mod mod I will be working on to specialize the slaves a bit.

I put the crime on slaves even tough I think it needs to be a value that changes with technology. At first there just is no scope for slaves to do more criminal activity than is already in the system but later on as the people become richer there does become the scope for slaves to do crime above what the normal populace do.

I did suggest that I remove crime for now but I think Koshling wanted to see what was actually happening so that he could adjust the AI to cope with it. That would be the best solution

Click to expand...

Yes, I don't want us to have a knee-jerk reaction here. I'll get round to adding some AI for properties on specialists in the next few days,

Arkenor said:

The building is called Civic (Banditry) 1. It says it autobuilds with the Banditry civic. Though oddly I can't see it in any of the building tabs, so either it didn't actually build, or its invisible. I'll build something and find out..... OK, the promotion doesn't seem to be being applied, so I guess the building is not actually being built. Which is actually a bit of a relief, for the purposes of my current game.

I've attached pictures of the building, and the promotion it desires to apply.

There are other defensive units beyond archers. While the City Defense promotion makes for a good all-round defense, I've always preferred a rounded mix of units. I always have a spearman in case of cavalry, and anti-melee melee, such as axemen. Importantly, these units are also useful for counter-attacking and other purposes beyond defending a city, leaving you with more flexibility. Rather than starting any other units with built-in city defending bonuses, merely giving them access to the City Defense promotion line would probably be sufficient. (Assuming the AI understands that sort of thing). That particularly makes sense for spear/pikemen.

With almost all units having a strong bonus against at least one other unit type, it's not necessary, and possibly counter-productive, to focus on City Defense bonuses.

Click to expand...

This must be extremely new. I can see how it could get out of hand with every unit getting it. However I also see it as fitting with the intent of the civic. I mean its only the 2nd civic in that category and if you were running banditry wouldn't all military units be bandits? To me it make sense even if it would be horrible to manage.

ls612 said:

I agree with Arkenor, currently the TW/Guard/City Guard line has a strength progression of (5/9/19), perhaps changing it to (4/7/15) would fix the issue. Does anyone object to that?

Click to expand...

Sounds fine to me.

Thunderbrd said:

@ls612: If I work up the Streetfighter buttons do you still have the shell of the xml you said you'd done for that set of promotions? (Probably would be needing some updates.) If not, I suppose I could go back and look at all the first pages of your promotions thread to find reference to the stats we agreed on for those.

Click to expand...

Is this a promotion or a unit? Because I think we have enough early units.

Hydromancerx said:

This must be extremely new. I can see how it could get out of hand with every unit getting it. However I also see it as fitting with the intent of the civic. I mean its only the 2nd civic in that category and if you were running banditry wouldn't all military units be bandits? To me it make sense even if it would be horrible to manage.

Click to expand...

Aye, it was new today or yesterday. Banditry looks pretty unusable to me now, though perhaps folks with different playstyles may be able to see the attraction. At any rates I'll not be going anywhere near it (once it starts spawning the autobuild building. At the moment it appears bugged in that regard. So my alarm is directed at what it *claims* to do, rather than what it's actually doing right now.)

Players can look after themselves (though it should be made more obvious on the main civic screen). Mostly I'm concerned as to whether the AI would decide to use it not realising the implications, then get itself in an awful mess by having all its melee, archer, and gunpowder units made while under banditry cause crime. And those units would keep causing crime for as long as they lived, I suppose, keeping that promotion when upgraded, long after the AI had turned its back on the banditry civic.

Maybe the AI is smart enough to not do itself a mischief with all that, in which case I shall retract my cries of alarm! What say you, Koshling?

H said:

Is this a promotion or a unit? Because I think we have enough early units.

Click to expand...

Promotions. Melee probably shouldn't have access to the extraordinarily powerful full City Defense promos so these would be minor city defense and also some minor city attack, further potentially increasing their city attack ability IF they have already capped out on the main City Raider line. I think it was something like +15% City Defense, +15% City Attack per step... something like that. I'll have to go back and refer to the early posts on ls612's promo thread to see how it was worked out cuz I recall I really liked how it got hammered out in the end. The only thing, as I recall, that stopped us was that I tried to give them some art and that art didn't jive well enough with the normal unit promo buttons (which I now agree with completely and would be quite happy to remake them along more coherent lines artistically.)

Well do what you will then. The AI is broken now anyway in another area but it also affects this too.

JosEPh

*reads the no kill, no capture of civies civic*

*Builds a huge force of workers and work boats then rings all the other guys cities with sleeping mode civies... shells the crippled cities/units and uses multi move military units to destroy their defenses*

As the terrorist dictator says about his new human shield wall tactics:
Stupid people are our greatest natural resource!!!

Necratoid said:

*reads the no kill, no capture of civies civic*

*Builds a huge force of workers and work boats then rings all the other guys cities with sleeping mode civies... shells the crippled cities/units and uses multi move military units to destroy their defenses*

As the terrorist dictator says about his new human shield wall tactics:
Stupid people are our greatest natural resource!!!

Click to expand...

Would not stop the city working the plots nor their units moving through the workers as if they are not there.

Arkenor said:

The banditry civic (unlocked at scavenging) now auto-places a building in your city that gives all (melee, archer, gunpowder) a promotion that raises crime +5.

Click to expand...

That sounds like one of the worst changes to the mod in a long time, making a previously fine civic completely unusable. And IIRC, the first military civic has it's own drawbacks, so now it'll take even longer to get a decent one.

JosEPh_II said:

Is everyone forgetting that crime is also on the cities workable tiles as well as the city itself? That crime diffuses from a tile to surrounding tiles, whether in the City tile or on a tile working tile. That if a for ex. an enemy Rogue (any crime spreading unit) is in your empire the tile it is on has an increase in Crime. The very basics of the TW is that it reduces crime, not only in the City but on Empire tiles. Otherwise the 2nd Crime reducing value of a TW is useless. But it isn't.

Click to expand...

But since crime only has an effect on cities, reducing it anywhere but on the city tile is inefficient, and neither the AI nor any player should be doing it.

MaXimillionZero said:

<snip>

But since crime only has an effect on cities, reducing it anywhere but on the city tile is inefficient, and neither the AI nor any player should be doing it.

Click to expand...

No that is incorrect, Crime flows from the tile of highest crime to the surrounding tiles.

Now the highest crime tile maybe the City tile, But not always. And the only way to reduce the highest tile outside of a city the fastest is the TW line. Once the Crime producing cause (unit) is removed, it does begin a slow process of dissipation. But while this make take 10's of turns the "osmosis" effect has raised the crime level in your City's tile too.

Why do you think Hydro has stated again and again, that once you build a rogue, assassin, etc. get it Out of your Empire and into the target empire asap?

Originally Posted by Arkenor
The banditry civic (unlocked at scavenging) now auto-places a building in your city that gives all (melee, archer, gunpowder) a promotion that raises crime +5.

Click to expand...

I do agree with you MaXimillionZero on this. If this is the case.

JosEPh

JosEPh_II said:

No that is incorrect, Crime flows from the tile of highest crime to the surrounding tiles.

Click to expand...

I know it flows in such a fashion, but only the crime on the actual city tile is what counts, and I'm under the impression that having a crime-reducing unit directly on the city tile helps the city more than having it on any other tile nearby. If it doesn't, then the system is quite strangely designed indeed.

Are you purposely misunderstanding what I said? The tile whether City Main tile or Not with the Highest Crime level will be seen by the AI as the tile that needs to be addressed.

Of course having a crime fighting unit in the city helps the city tile. But if the highest Crime Producing tile is not the main city tile, having a TW in the city reduces the flow from the Highest tile but does not stop the flow into the city. The AI only "sees" that the sw forested hill tile from the main city tile is the Highest Crime producing tile and will act upon reducing that tiles Crime level.

It's not strangely designed. It's a matter of saturation and addressing the highest point of saturation.

We've (AI and Player) been only given 2 means of addressing tiles outside of the main city tile. Get rid of the source of Crime generation on the tile and then place a unit that facilitates the reduction faster. The main city tile has a myriad means thru bldgs and units to address crime on That main city tile. It's osmosis and reverse osmosis. 99% of the players only worry about the Main City tile. But the AI only sees the point of highest concentration and works from there.

JosEPh

But since crime has no effect anywhere other than city tiles, the AI should focus its anti-crime units on those. Getting rid on crime-generating units elsewhere would of course be useful, but there's no need to move anti-crime units outside of cities.

MaXimillionZero said:

I know it flows in such a fashion, but only the crime on the actual city tile is what counts, and I'm under the impression that having a crime-reducing unit directly on the city tile helps the city more than having it on any other tile nearby. If it doesn't, then the system is quite strangely designed indeed.

Click to expand...

This is essentially correct. You **can** reduce crime by stationing crime reduction units on non-city tiles to reduce the back-flows to nearby cities, but stationing the same unit in eh city will always have a greater effect since it is only in the city that the build up of crime has an actual effect. The only time it **might** be sensible to station a crime fighter outside a city (for crime fighting purposes at least) is if two cities are fairly close together and stationing one mid way between to stifle the flow between them is going to help both cities with one unit.

Arkenor said:

Aye, it was new today or yesterday. Banditry looks pretty unusable to me now, though perhaps folks with different playstyles may be able to see the attraction. At any rates I'll not be going anywhere near it (once it starts spawning the autobuild building. At the moment it appears bugged in that regard. So my alarm is directed at what it *claims* to do, rather than what it's actually doing right now.)

Players can look after themselves (though it should be made more obvious on the main civic screen). Mostly I'm concerned as to whether the AI would decide to use it not realising the implications, then get itself in an awful mess by having all its melee, archer, and gunpowder units made while under banditry cause crime. And those units would keep causing crime for as long as they lived, I suppose, keeping that promotion when upgraded, long after the AI had turned its back on the banditry civic.

Maybe the AI is smart enough to not do itself a mischief with all that, in which case I shall retract my cries of alarm! What say you, Koshling?

Click to expand...

No, this is a total AI-killer I suspect (once it works). The AI will react to the crime it results in, but won't spot that the problem will occur in advance when it evaluates the civic. The (AI evaluation) problem here is 3-fold:

1) Auto-build buildings are fairly new and both civic and tech triggered auto-builds of NEGATIVE buildings are pretty much ignored - the building evaluator for both techs and civics works by assessing the new buildings that will be made available, but (currently) assumes the player will have a choice about whether to ACTUALLY build them, so it ignores negative values (assumes it won't build them) when evaluating the civic/tech. This is basically the AI being behind the curve on the auto-build mechanic, which was never much used until recently, and not used for 'bad' things at all until very recently.

2) The AI doesn't really understand the concept of 'bad' promotions, and as such once a unit gets a promotion it will never really figure in any calculation as to whether to scrap the unit (or avoid upgrading it). It will realize that such a unit is worth a bit less than another that is identical apart from not having the promotion concerned, and hence probably expose them to slightly more battle-risk, so this might whittle them down eventually. The AI most certainly won't consider scrapping a unit in order to reduce crime.

3) The AI has a TERRIBLE time with crime from units that do not have the pillager AI, since only this AI has the smarts to try to keep the unit away from its own cities. Having it on city defense units is especially troublesome since the AI will always try to keep those units in and around its cities

This is another example (an especially big one as it happens) of how the rate of asset change in C2C just means a continuous backlog of AI work, that removes time from trying to actually make real underlying improvements to the AI. In this case I STRONGLY recommend either removing this feature, as I see this as a definite case of gameplay trumping reality.

As a slight aside what interest would there be in a mod-mod that freezes the assets pretty much totally (apart from ongoing tweaks to existing values) at some point (say V29, but potentially much earlier) and just works to improve the AI within that framework so that all development time can go on actual improvements rather than tweaks to catch up with new mechanics or new uses of mechanics?

Koshling said:

No, this is a total AI-killer I suspect (once it works).

3) The AI has a TERRIBLE time with crime from units that do not have the pillager AI, since only this AI has the smarts to try to keep the unit away from its own cities. Having it on city defense units is especially troublesome since the AI will always try to keep those units in and around its cities

This is another example (an especially big one as it happens) of how the rate of asset change in C2C just means a continuous backlog of AI work, that removes time from trying to actually make real underlying improvements to the AI. In this case I STRONGLY recommend either removing this feature, as I see this as a definite case of gameplay trumping reality.

As a slight aside what interest would there be in a mod-mod that freezes the assets pretty much totally (apart from ongoing tweaks to existing values) at some point (say V29, but potentially much earlier) and just works to improve the AI within that framework so that all development time can go on actual improvements rather than tweaks to catch up with new mechanics or new uses of mechanics?

Click to expand...

So COMPLETELY remove the crime from C2C, is that what your saying??

strategyonly said:

So COMPLETELY remove the crime from C2C, is that what your saying??

Click to expand...

No, remove crime from free promotions given by autobuild buildings indiscriminately to all units is all I'm saying - specific case currently being the auto-build building that changing to civic banditry causes to be built, which adds crime production to almost all units.

JosEPh_II said:

Are you purposely misunderstanding what I said? The tile whether City Main tile or Not with the Highest Crime level will be seen by the AI as the tile that needs to be addressed.

Of course having a crime fighting unit in the city helps the city tile. But if the highest Crime Producing tile is not the main city tile, having a TW in the city reduces the flow from the Highest tile but does not stop the flow into the city. The AI only "sees" that the sw forested hill tile from the main city tile is the Highest Crime producing tile and will act upon reducing that tiles Crime level.

It's not strangely designed. It's a matter of saturation and addressing the highest point of saturation.

We've (AI and Player) been only given 2 means of addressing tiles outside of the main city tile. Get rid of the source of Crime generation on the tile and then place a unit that facilitates the reduction faster. The main city tile has a myriad means thru bldgs and units to address crime on That main city tile. It's osmosis and reverse osmosis. 99% of the players only worry about the Main City tile. But the AI only sees the point of highest concentration and works from there.

JosEPh

Click to expand...

You are indeed correct, crime does flow out from Cities into adjacent tiles if there is a large difference in their crime levels and the City is gaining crime. However, it isn't a matter of the city getting 'saturated', it is a matter of propagation along routes primarily. If the city has 100 crime and the neighbor tile has 0 crime that crime from the city will flow to the neighbor tile, even if the city isn't near it's cap.

The same mechanism affects crime fighting. Let's say that you have 5 TWs on a tile next to the city. The net effect of the TWs on that tile would be -50 crime, but their effect on the city would be considerably less than that (the exact number involves integral calculus, so I'm not going to calculate the exact number now). So since there is no limit to how many units you can keep in a city unless that tile is strategically important the TWs would be far better suited to being in the city than on a neighboring tile.