So How has the AI improved ??

[amritochates]

I am as my join date suggests a long time lurker, fairly recent member and dedicated fan of the TW series, well the last part isn't totally accurate and is valid only until MTW which is why this query is being posted.

Needless to say RTW was in most respects a complete disappointment for me, especially with regards to both the strategic and the tactical AI. In fact in my opinion the overall game was sub-standard enough to put off my purchase of MTW II till the release of the initial set of patches. Now that I am contemplating purchasing the game I would like your input on certain matters.

A continuous perusal of the forum seems to indicate a significant improvement in the AI, however that by itself is insufficient so I intend to ask a specific set of queries to judge the actual level of improvement:


Firstly, one of the major problems with RTW and BI was the inability of the AI in a majority of situations to maintain a battle line. Even with Darthmod formations the AI would hold the battle line only until it was about 100 mts away from my battle line- at which point it would decide that to utilise tactics from the Homeric age and would then engage my battle line one individual unit at a time deliberately destroying the cohesion of its battle line.Has this major defect been rectified, because if not then we can end this thread right at this point for without the AI attacking as a cohesive whole there is no point in father inquiring about any subsequent improvements.

Secondly, is the AI able to deal with severely depleted units or does one still face 20 unit AI stacks that have a roster of less than 200 men. Infact does the AI utilise any form of automerge or does it ever retrain its depleted units?]

Thirdly, does the AI always aim to secure terrain and height advantages as It did in MTW and does it seek to consistently outflank my battle line?]

Finally, is the AI able to co-ordinate multiple stack armies to present me in a position of numerical inferiority?]

As stated above your response on the following points will be appreciated, so I wish to thank all those who contribute to my queries for their time and input.
~:thumb:

[Yun Dog]

I am as my join date suggests a long time lurker, fairly recent member and dedicated fan of the TW series, well the last part isn't totally accurate and is valid only until MTW which is why this query is being posted.

Needless to say RTW was in most respects a complete disappointment for me, especially with regards to both the strategic and the tactical AI. In fact in my opinion the overall game was sub-standard enough to put off my purchase of MTW II till the release of the initial set of patches. Now that I am contemplating purchasing the game I would like your input on certain matters.

A continuous perusal of the forum seems to indicate a significant improvement in the AI, however that by itself is insufficient so I intend to ask a specific set of queries to judge the actual level of improvement:


Firstly, one of the major problems with RTW and BI was the inability of the AI in a majority of situations to maintain a battle line. Even with Darthmod formations the AI would hold the battle line only until it was about 100 mts away from my battle line- at which point it would decide that to utilise tactics from the Homeric age and would then engage my battle line one individual unit at a time deliberately destroying the cohesion of its battle line.Has this major defect been rectified, because if not then we can end this thread right at this point for without the AI attacking as a cohesive whole there is no point in father inquiring about any subsequent improvements.

Secondly, is the AI able to deal with severely depleted units or does one still face 20 unit AI stacks that have a roster of less than 200 men. Infact does the AI utilise any form of automerge or does it ever retrain its depleted units?]

Thirdly, does the AI always aim to secure terrain and height advantages as It did in MTW and does it seek to consistently outflank my battle line?]

Finally, is the AI able to co-ordinate multiple stack armies to present me in a position of numerical inferiority?]

As stated above your response on the following points will be appreciated, so I wish to thank all those who contribute to my queries for their time and input.
~:thumb:

Some may disagree with me but M2TW is basically RTW in a new dress

1) As far as Im concerned this is how the AI attacks -

a) frontal charges in with his general (alone) into your infantry line and archer fire - his general dies

b) some of his units that arnt just standing around facing the wrong direction still getting shot to pieces may attempt to rush you in a piece meal fashion :laugh4: :laugh4:

2) you mean you dont like being charged by a unit of 2 blokes and thier half lame dog - remarkably hard to kill those men youll find
nothings changed there


3) outflanking and army positioning ala MTW :laugh4: :laugh4: :laugh4: - :no:

4) If you played MTW you will find the strategic and tactical AI of M2TW like playing against an infant (and not one of those genius chess champion ones - one of the crying for a dummy tantrem ones - maybe THIS time that 1 stack naval invasion will work? ) :laugh4: :laugh4: :laugh4:

add to this it doesnt defend its castles/citys and most of its armies circle aimlessly looking for opportunistic targets (your citys) to seige while your armies roll its nation into the earth

Im going to stop here... but I could go on and on

they say a pictures worth a thousand words - so here the picture of the great Timurid invasion

http://members.westnet.com.au/rogun/screenies/broken.jpg

[sapi]

I am as my join date suggests a long time lurker, fairly recent member and dedicated fan of the TW series, well the last part isn't totally accurate and is valid only until MTW which is why this query is being posted.

Needless to say RTW was in most respects a complete disappointment for me, especially with regards to both the strategic and the tactical AI. In fact in my opinion the overall game was sub-standard enough to put off my purchase of MTW II till the release of the initial set of patches. Now that I am contemplating purchasing the game I would like your input on certain matters.

A continuous perusal of the forum seems to indicate a significant improvement in the AI, however that by itself is insufficient so I intend to ask a specific set of queries to judge the actual level of improvement:


Firstly, one of the major problems with RTW and BI was the inability of the AI in a majority of situations to maintain a battle line. Even with Darthmod formations the AI would hold the battle line only until it was about 100 mts away from my battle line- at which point it would decide that to utilise tactics from the Homeric age and would then engage my battle line one individual unit at a time deliberately destroying the cohesion of its battle line.Has this major defect been rectified, because if not then we can end this thread right at this point for without the AI attacking as a cohesive whole there is no point in father inquiring about any subsequent improvements.It's much better, imo, but there are still some issues (in particular the AI's actions when badly outnumbered or simply low on men)


Secondly, is the AI able to deal with severely depleted units or does one still face 20 unit AI stacks that have a roster of less than 200 men. Infact does the AI utilise any form of automerge or does it ever retrain its depleted units?]I've never noticed a stack full of depleted units; although that's far from a decisive answer to that question.


Thirdly, does the AI always aim to secure terrain and height advantages as It did in MTW and does it seek to consistently outflank my battle line?]They will deploy in the best possible position for the most part, but won't get engaged in a game of tactical merry-go-round in battle


Finally, is the AI able to co-ordinate multiple stack armies to present me in a position of numerical inferiority?]It will take advantage of such positions, but I've never seen the AI attempt to create them.

[Didz]

I suspect that the answer is that all of the the faults you perceived can still be found in MTW2. However, perhaps it would be wise to form your own judgement. Have a look at the battle report I published for the Battle of Nottingham and see if the performance of the AI suggests any improvement has occured.
https://forums.totalwar.org/vb/showpost.php?p=1552251&postcount=19
This battle was fought after installation of the 1.2 MOD on Medium Difficulty setting.

There are other battle reports in this thread and so overall it should give you a decent idea of how the AI performs both in campaign and battle. I make no excuses for how I perform however, I accept that I am no Alexander.

[econ21]

I think Sapi's answers match my casual observation. I find both the tactical and strategic AI much improved on RTW, although the strategic AI needs the most work IMO.

However, I do have a suspicion that the AI deliberately keeps several stacks together for mutual defence; and I have also found it less likely to attack you piecemeal, one stack at a time in the same turn.

Comparing the behaviour of the BI hordes and the M2TW Mongols/Timurids may provide the best test of this as both have multiple AI stacks. Unfortunately, my experience of the Mongols/Timurids is limited.

[Kobal2fr]

4) If you played MTW you will find the strategic and tactical AI of M2TW like playing against an infant

You mean, the same MTW where huge armies zig-zag all around so your archers can have more time to have fun, and where cavalry cleverly feints a charge on a spearwall, then retreats a bit, then charge again in a huge blob that gets instantly routed because said archers completely depleted them, with only the jedi general destroying half your spearmen ? The selfsame MTW that has huge armies redeploying entirely because a lone cavalryman has farted on your flanks ? The brilliant and clever strategic MTW that has AIs with thousands of florins which keep pumping out full stacks of peasants and ballistas no matter what you do or how many armies of theirs you trounce, how many provinces you utterly destroy or how much of their trade you disrupt ? That MTW ?

Just checking, you know. There might be another one I haven't played :grin:

Aaaah, nostalgia, how thy rose-tinted spectacles deceive thee.

[Noir]

Originally posted by cobal2fr
You mean, the same MTW where huge armies zig-zag all around so your archers can have more time to have fun, and where cavalry cleverly feints a charge on a spearwall, then retreats a bit, then charge again in a huge blob that gets instantly routed because said archers completely depleted them, with only the jedi general destroying half your spearmen ? The selfsame MTW that has huge armies redeploying entirely because a lone cavalryman has farted on your flanks ? The brilliant and clever strategic MTW that has AIs with thousands of florins which keep pumping out full stacks of peasants and ballistas no matter what you do or how many armies of theirs you trounce, how many provinces you utterly destroy or how much of their trade you disrupt ? That MTW ?

While all this is true it has for the most part hardly to do with the competence of the AI - it largely relates to the way the campaign is designed to play and not how well the battle/strategic AI operates.

Also many of those factors are affected by the maps (and their huge hills that a human opponent would refuse to play), and by the "agressive defening" strategy where the player destroys the AI's armies just by sitting in a very defensible province.

It follows that IMO they hardly constitute an argument over the AI competence issue.

You can change most of the things you mention with simple modding in MTW btw.

Also try playing MP (in any of the TW games) and compare the experience to the MTW AI and to the M2TW AI - you might be able to spot the difference better.

Many Thanks

Noir

[Kobal2fr]

Errr... Sorry, you've got me confused there.

How does "the AI can't attack worth a fiddlefart, oh, and broken terrain is a big no-no as it can only deal with smooth, flat, rolling plains" NOT constitute an argument over AI competence ? Or rather, how is it "not an argument over the AI competence issue" but "related to the way the campaign is designed to play" when we're talking about MTW, but "playing against the AI is like playing against a retard, and not the cute kind" when we're talking about the same kind of things, yet on a smaller scale, but happening in M2TW ?

You can mod most of these things with simple modding in M2TW too btw. There's a clearly commented and fairly straightforward .xml file if you want to code your own clever battle AI. And another for the strat AI.

[Ulstan]

While all this is true it has for the most part hardly to do with the competence of the AI

Er, what? They are all specific examples of the poor quality of the AI.

I grew sooo frustated with MTW in that the only way you would ever see an army of anything other than peasants and spearmen and seige engines was if some faction re-emerged.

A good tactical AI is of no use if it only has peasants to command.

[Noir]

Originally posted by cobal2fr
How does "the AI can't attack worth a fiddlefart, oh, and broken terrain is a big no-no as it can only deal with smooth, flat, rolling plains" NOT constitute an argument over AI competence ?

It doesn't because you should be comparing the competence of the AI when he has equal chances with the player and not when he doesn't. This clearly isn't a fair comparison.


Originally posted by cobal2fr
Or rather, how is it "not an argument over the AI competence issue" but "related to the way the campaign is designed to play" when we're talking about MTW, but "playing against the AI is like playing against a retard, and not the cute kind" when we're talking about the same kind of things, yet on a smaller scale, but happening in M2TW ?

You are right it does play a part in judging both. However what you've missed is that the point i am trying to make is that you should judge the AI from custom battles in relatively uqual potential armies and maps that do not give massive advantages to the Ai or the player and not from how the AI got beaten by army X in campaign Y where many other (campaign related) factors come into it.


Originally posted by cobal2fr
You can mod most of these things with simple modding in M2TW too btw. There's a clearly commented and fairly straightforward .xml file if you want to code your own clever battle AI. And another for the strat AI.

Very true - however once this is done the M2TW Ai is still inferior (in custom battles with equal troops and terrain that does not provide huge advantages to either side) from the MTW AI IMO.

My comment was meant to show that that MTW, as you called it isn't as bad as it sounds from your post, in terms of SP game potential.


originally posted by Ulstan
Er, what? They are all specific examples of the poor quality of the AI.

I grew sooo frustated with MTW in that the only way you would ever see an army of anything other than peasants and spearmen and seige engines was if some faction re-emerged.

This is a hardcoded deficiency - a sort of a bug if you wish rather than competence of the AI choices. I repeat that the competence of the AI should be judged by what sort of challenge it gives to the player when he has the same chances of success and not when he doesn't.

It sounds that you haven't played a single mod of MTW - much like the RTW mods they all moded peasants and the lower siege engines out. Have you tried the game that way? Its marginaly better.

Many Thanks

Noir

[Didz]

@Noir
I'm puzzled by your response to cabal2fr.

Surely, if one tries to test the performance of the AI in a custom battle where the opposing forces are equal and the the terrain offers no potential advantages to either side then there is very little to test simply because no matter how good the AI is (or how poor) it has no opportunities to expliot and no disadvantages to minimise.

Isn't the real measure of AI performance how the AI makes best use of whatever advantages it has, and how effectively it minimises any disadvantages. After all thats what human players do, and that ought to be the standard that the AI aspires to mimic.

I'm curious about how you would judge the competency of the AI in such a situation.

[Noir]

Originally posted by Didz
Surely, if one tries to test the performance of the AI in a custom battle where the opposiing forces are equal and the the terrain offers no potential advantages to either side then there is very little to test simply because no matter how good the AI is (or how poor) it has no opportunities to expliot.

Isn't the real measure of AI performance how the AI makes best use of whatever advantages it has, and how effectively it minimises any disadvantages. After all thats what human players do, and that ought to be the standard that the AI aspired to mimic.

You actually hit the nail in the head.

having no exploits is what constitutes a good game:
the battle is then decided by how good match ups you are making, how well you plan and execute flanking maneuvers and how well you time and execute all this in a well thought battle plan that takes into account the enemy forces and any existing terrain features (a little forest, a little bump of ground etc).

All this becomes second nature to clearly see once you play MP. I have made the transition from SP to MP very recently, and i can tell you that a game that has exploits as you put it bluntly and correctly is no fun *edit*= neither can it be taken as an exhibit argument of "competence".

MP players refuse to play maps that do have "exploits" after all - isn't it unreasonable then to expect the AI to play against them and judge him as bad when he (understandably/predictably) cannot?

Many Thanks

Noir

[Kobal2fr]

You are right it does play a part in judging both. However what you've missed is that that the point i am trying to make is that you should judge the AI from custom battles in relatively uqual potential armies and maps that do not give massive advantages to the Ai or the player

Why should I ? That's not what happens in campaign battles. In campaigns, one side is usually much advantaged in numbers or quality over the other (else there wouldn't be a battle) and terrain is also usually playing on one side, wether said terrain is a mountainous retreat or a big stone wall with towers and stuff.
Custom battles on grassy plains with equal forces are all very well, but they have no "reality" whatsoever, and are an artificial setting.
Also, wether the MTW AI is or isn't better at handling those (and frankly, I'm not arguing either way. Beating MTW on the field even on VH was a cakewalk too, and that's with AI soldiers having huge combat boni) is largely irrelevant because you won't be fighting those battles in an SP campaign.


My comment was meant to show that that MTW, as you called it isn't as bad as it sounds from your post, in terms of SP game potential.


I know. I liked MTW, even though I feel the whole boardgame thing didn't work that well on such a huge scale, which is why I like M2's strat map better for that setting.
But my comment was meant to show that this M2TW isn't as bad as it sounds from old grog posts in terms of SP game potential :grin: ; and that one could focus on MTW's negatives just as much as people seem to enjoy poiting fingers at what M2 does badly and dismissing all the good stuff.

Heck, I'm an old grog and I love it.

EDIT : as an unrelated note, WHY DOES EVERYONE INSIST ON NICKING MY K ? K is a manly letter, the hardest consonnant of them all ! Only pansies and women and lily-livered hippies use Cs.

[Noir]

Originally posted by Kobal2fr
Why should I ? That's not what happens in campaign battles. In campaigns, one side is usually much advantaged in numbers or quality over the other (else there wouldn't be a battle) and terrain is also usually playing on one side, wether said terrain is a mountainous retreat or a big stone wall with towers and stuff.
Custom battles on grassy plains with equal forces are all very well, but they have no "reality" whatsoever, and are an artificial setting.

You should, if you wish to express an objective opinion on the AI performance. You should be basing that opinion then in objective criteria - ie you should be making conditions that do not favor any result and then see what happens.

Unfortunately many things in all TW games are done for the enjoyment of the player only.

Tech trees and the like are killers for the AI - however they are followed upon because the player would like them. In short the game follows what is "fun" and not what is "fun & what helps the AI". In the end the game lacks challenge because of this, as well as because of the lack of balance - but that's another story.


Originally posted by Kobal2fr
But my comment was meant to show that this M2TW isn't as bad as it sounds from old grog posts in terms of SP game potential ; and that one could focus on MTW's negatives just as much as people seem to enjoy poiting fingers at what M2 does badly and dismissing all the good stuff.

I see the point, but there's a misunderstanding: people might be posting because the game falls short of their expectations or of the previous performance - it doesn't mean that they don't want to like it or that they enjoy "bashing" it. In all probability (at least some) are posting exactly because they are attached to it - despite it letting them down.

Many Thanks

Noir

[crpcarrot]

Noir

based on your argument i would prefer an unbalanced AI. having a 100% level playing field (inall aspects) is well and good in MP but it doesn nothing for SP. Of course players take advantage on opotunities but the AI i would like to play against would be able to recognise when i am doing it and try to counter it. not enhance my advantage by doing exactly what i want it to. and it should also be able to recongnise when it has these situations to exploit and exploit them blatantly. it would make the game more interesting.

[B-DogKY]

I have to say, after reading all these nostalgic comments about how MTW was soooooo much better than RTW et al, I dusted off my old copy of MTW and loaded it up for another go. UGHHHHHH. I instantly remembered why I loved RTW almost from the first battle I fought. After chasing the AI army around the map for the entire time (yes time actually ran out with no fighting at all...) I was about ready to quit, but I thought I'd play on for a bit, maybe it gets better.... The next battle involved 5 BG units for the AI attacking vs 1 BG unit, 10 SPEARS, and 4 light cav for me defending. I won, with 5 spearman and my general left. I thought spears were good against cav? Apparently not..... That was it for me, I unloaded the game posthaste.

For me, a good tactical AI is NOT characterized by the ability to run around the battlefield. I understand making use of terrain and all, but in real life, an inferior army would do it's best to outmanuver a superior army, but if brought to battle, they did their best, or routed wildly beyond the hope of their commanders to command them. One thing I think that is overlooked in all TW games as far as tactics are concerned is the fact that the defending army HAS to fight this battle. They have deployed on this battlefield for any one of a myriad of reasons, but no matter which reason, if the battle does not take place, they have not improved their strategic position, so playing ring-around-the-rosie or waiting for the timer to run out is not an option.

For example, the last real Lancastrian army entered England. When Edward brought his army to face them, Lancaster decided to make for Wales to link up with Jasper Tudor's army. There was a desperate race to cross the Severn. Since Gloucester was closed to them, the Lancastrians made for the crossing at Tewkesbury. They covered 24 miles in 15 hours and got there first, but Edward had done the impossible, in 12 hours his army had covered 35 miles and were within 5 miles of the Lancaster army. Knowing they could not cross with Edward so close, Lancaster turned to fight. Also knowing they cannot cross, Edward encamps his army for the night.

Why does Lancaster not slip across now, given the opportunity? Simply because they will not improve their strategic position. Edward's army is faster and Lancaster now needs to win this battle to "get him off their backs". Whether or not they believe they can win is immaterial (they do not, by the way), they are simply out of options. When the battle starts on the morrow, if Edward outflanks them, they cannot play ring-around-the-rosie, tiring their men to no good purpose, and in fact probably giving Edward possesion of the crossing. Even if no fight took place, their strategic position just got worse!! They would simply adjust their lines to meet the flanking movement, and fight. They absolutely HAVE to bloody Edward's army and slow him down, simply remaining in possesion of the battlefield will not suffice.

Now, all these strategic concerns cannot be taken into consideration as TW exists now, but the one major change they made to RTW was that when an army comes to the battlefield it mostly (I'd say over 90%) COMES TO FIGHT. They didn't come to dance, or withdraw before a shot is fired. This is a VAST improvement over MTW, and one that makes MTW, IMHO, not worth playing anymore. I have had 2 battles I remember to this day from TW and they were both in BI (although I did love watching the sword saint destroy an entire unit in STW, as a warmup....).

Anyway, I ranted a bit, but I said what I wanted. I realize some will look at my post count and think I'm a bit cheeky, but I have played TW for years and I have read the Guild for years, so I do actually feel qualified to make these statements.

[Tyrac]

While I am not going to go out and say the AI is awesome or perfect I do remember MTW and RTW pretty well.

I beat the current AI unless the battle is strongly in the AI's favor
..... or I am very drunk.

However I utterly smashed the AI in RTW and MTW. I would win fights that were 3 to 1 odds with almost no losses all the time. It is the main reason why I drifted away from them.

Also wasn't the AI given large bonuses for difficulty level back then that now it does not receive? Perhaps that might account for a great deal of the memories of it being "so much" better.

It was NOT better. It was easier to trick. It could not handle "ranged" combat. It could not deal with terrain at all.

I remember in MTW just making a line of ranged troops and routing multiple enemy stacks. That would not happen now.

Give the current AI stat bonuses based on difficulty level and it would kick your ass.

[Noir]

originaly posted by cpcarrot
...based on your argument i would prefer an unbalanced AI. having a 100% level playing field (inall aspects) is well and good in MP but it doesn nothing for SP. Of course players take advantage on opotunities but the AI i would like to play against would be able to recognise when i am doing it and try to counter it. not enhance my advantage by doing exactly what i want it to. and it should also be able to recongnise when it has these situations to exploit and exploit them blatantly. it would make the game more interesting.

True yet, can it be achieved? In the Collosseum, there was an interesting discourse with Loudens some weeks ago, relative to the AI in the new and old engine. In short he seems to claim that the AI lost abilities in th engine transition, as opposed to my claim that he has them but they won't manifest due to the engine change.

In any case it seems that it was clear to both of us that the tactical AI isn't as challenging as it was.

I still enjoy playing SP and do so often - dont get me wrong. However one thing that puts me off i the unrealistic expectations that the game design imposes on the AI. What you also state there, are unrealistic expectations - and CA is putting up with them in the campaign design because the SP players want varied maps, varied units and tiled campaign map. The AI can't handle any of them neither are to his benefit, in the game.

I also love varied maps and still play SP much more than MP - however many of them in MTW for example are plain cheats - the "hills" are so steep and long slopes that when the AI attacks is simply taken out by arrows as Kobal2fr is stating - he has no chance - but if you follow his treasury you'll see that he's attacking because he's on the red - again the campaign plays against him in the financies as is.

Now if you make the slopes instead of say 45 degrees angle something more like 20 degrees angle with multiple angles of approach in the horizontal direction - then the "overdefence" advantage/exploit becomes more equal and the battle is more challenging. If you round the map edges units cannot be trapped as easily. If you follow the AI competence in making tech trees instead of human preference then you'll get full AI stacks and logical attacks. The devil is in the details.

Does all that make nothing for SP? - with its allegedly labeled "boring" campaigns even from its most dedicated fans?

Many Thanks

Noir

[TinCow]

My general opinion of M2TW 1.2 AI:

The tactical AI is blunt and unimaginative, but it doesn't feel buggy to me. It favors direct attacks with solid lines, missile duels if it thinks it can win them, and flanking attacks with cavalry. These are good enough to defeat an unskilled player, but most people with a bit of experience can defeat the AI without any serious problems.

The AI can win battles with it's blunt and unimaginative tactics when it has a sufficient advantage in men. However, the strategic AI is still relatively poor and does not produce those kinds of situations to allow the tactical AI to gain the advantage it needs.

All in all, even on VH/VH, most players will have no problems winning the game with any faction (in vanilla). However, the AI is still likely to win a couple battles here and there and the loss of a settlement or two during the course of a campaign is to be expected. This is in contrast with RTW, where loss of a settlement was essentially unheard of.

[Didz]

You actually hit the nail in the head.
Not sure about that, I suspect that you merely misread my post and imagined a nail which wasn't intended to exist.

having no exploits is what constitutes a good game:the battle is then decided by how good match ups you are making, how well you plan and execute flanking maneuvers and how well you time and execute all this in a well thought battle plan that takes into account the enemy forces and any existing terrain features (a little forest, a little bump of ground etc).
If there were no potential advantages to exploit then neither the AI nor the player can gain any advantage and the result of the battle would be determined purely by luck, and not tactical judgement. More importantly, as any AI is basically a mathematical routine designed to assess relative risk and identify potential benefits it can exploit, placing it in a situation where there are none is merely going to test its defaul sub-routine to 'do-something' when there are no intelligent options available.

This was why I was curious as to how you intended to assess the performance of the AI when, in effect, you were planning to deny it any opportunity to prove itself.

However, you appear to contradict yourself anyway, because you go on to mention that in your scenario there will be 'little forests and little bumps' which you would expect the AI to expliot. In other words there would be terrain advantages to be explioted and so you clearly didn't mean that the AI should be tested in a custom battle without any at all.

That being the case you have answered my question, in that I can see now how one could judge the AI's ability to exploit the potential advantages you have allowed it.


MP players refuse to play maps that do have "exploits" after all - isn't it unreasonable then to expect the AI to play against them and judge him as bad when he (understandably/predictably) cannot?
In my experience MP players will use every trick in the book to achieve just the opposite. Their aim usually is to ensure that they have all the advantages and you have as few as possible. In fact I used to find it extremely difficult to achieve anything like a balanced game when I used to play MP. Perhaps, things have changed but it sounds a bit out of character to me.

However, what players do or do not do in the MP game hardly has any relevance to the performance of the AI. The AI has to be able to cope, and provide a challenge, regardless of whether the playing field is level or not, and more importantly ought to be coded to try and ensure that whenever possible the situation is not only not level but heavily sloped in its own favour. Thats what a human player will be seeking and so the AI must be able to compete with its human opponent.

This again comes back to the question of how one judges the effectiveness of the AI, because in reality it encompasses much more than its ability to micro-manage the units in a battle. It also has to manage its cities effectively, handle diplomatic situations to its own advantage, manage its trade and economy, work towards achieving its specific faction goals, assemble effective armies to counter the strengths of the specific enemy factions it expects to meet and deliver its armies to the battlefield of its its choice preferably securing advantages in both terrain and numbers over its human opponents.

Only then when it has achieved all this and more does the issue of how it micro-manages the tactical employment of its troops become an issue.

[Noir]

Originally posted by Didz
Not sure about that, I suspect that you merely misread my post and imagined a nail which wasn't intended to exist.

I didn't intent to "use" your post to the benefit of my statements - i honestly think that it touched the heart of the argument.


Originally posted by Didz
If there were no potential advantages to exploit then neither the AI nor the player can gain any advantage and the result of the battle would be determined purely by luck, and not tactical judgement. More importantly, as any AI is basically a mathematical routine designed to assess relative risk and identify potential benefits it can exploit, placing it in a situation where there are none is merely going to test its defaul sub-routine to 'do-something' when there are no intelligent options available.

Pardon me for saying so, but you have no clue what you are talking about, this paragraph is just wrong, and shows IMO a poor understanding of the game.

There are two tactical components in the battle gameplay; these are in use whether there are territorial (or other) advantages or not: match-ups & flanking. The AI is aware of both of them in TW since time immemorial (MTW) and uses them. Traditionally he is better in match ups as he can use all his units at the same time, in MTW he exhibited also the ability to flank decently with cavalry. These make up for "intelligent" decisions no matter the terrain or the army composition.

If you have two comlpetely same armies in an entirely flat map the battle is not 100% of randomness, but 100% of match up and flanking maneuvers (and so skill, timing, reassessment and good judgement); this is precisely because there are no initial advantages and in fact in VI, MP tournaments were played in steppe maps. Despite no advantages initially, little enequalities that present themselves gradually as the battle progresses from the skirmish phase to the melee, to the finishing. Its up to the player to use them as they come and press them relative to its position at every moment of the battle, to achieve victory.

A featureless map is admiteddly boring, however - so terrain features can spice up the game as long as they are accessible and can be feasibly claimed by all sides involved. If they are not so, then the "advantage" turns into an "exploit" - and you see the edge camping and aggressive defending that people do in the SP game concsiously or unconcsiously.

This should answer the point where i "contradict" myself:

Originally posted by Didz
However, you appear to contradict yourself anyway, because you go on to mention that in your scenario there will be 'little forests and little bumps' which you would expect the AI to expliot. In other words there would be terrain advantages to be explioted and so you clearly didn't mean that the AI should be tested in a custom battle without any at all.

You don't seem to differentiate between advantages and "advantages" (a little bump agains a hill), which does happen alot - but judging from your previous paragraph i understand; by the way there are excellent guides around for how the game works: froggbeastegg's unit guides are some of them.

The point i am trying to make about the maps is that no "AI" then would be able to win (against an opponent defending a hill with reltively equal forces), no matter how much SP players want him to; since no player can win it also - statistically speaking.


Originally posted by Didz
In my experience MP players will use every trick in the book to achieve just the opposite. Their aim usually is to ensure that they have all the advantages and you have as few as possible. In fact I used to find it extremely difficult to achieve anything like a balanced game when I used to play MP. Perhaps, things have changed but it sounds a bit out of character to me.

Sorry, but the MP community is the only sort of players i know that do care about balance of maps/units/unit types (to varying degrees admitedly) and that's because its no fun for them to beat up an opponent when they defend a huge hill - so if you mean that they go for that sort of advantages to my experience they don't and they will (bitterly) complain if you do or drop out of your game.


Originally posted by Didz
This again comes back to the question of how one judges the effectiveness of the AI, because in reality it encompasses much more than its ability to micro-manage the units in a battle. It also has to manage its cities effectively, handle diplomatic situations to its own advantage, manage its trade and economy, work towards achieving its specific faction goals, assemble effective armies to counter the strengths of the specific enemy factions it expects to meet and deliver its armies to the battlefield of its its choice preferably securing advantages in both terrain and numbers over its human opponents.

True yet there are two Ai's to the best of my knowledge in teh game that operate independently on the fields you mention. If you notice above i mention that game design works against the way the strategic AI does things (against his routines) to the best of my knowledge because, features are included for the player (and his "enjoyment").

Many Thanks

Noir

[Husar]

Well, if I think of Mongol and Timurid stacks with all those experience advantages, they can be quite tough to beat, depending on where and with what you face them. I remember losing quite a few turkish cavalry armies to them in my first turkish campaign. Of course they do have all thos advantages and very strong units, but like many said, sometimes it's just that the build only peasants to face my guys in full plate what makes them weak.

I think it shouldn't be all that hard to make them recruit units from further up the tech tree. You could make them save some money for building improvements every turn(so they get access to better units in the first place instead of going bankrupt from peasant upkeep) and then you could change the build preferences to favour not some relation between stats and upkeep or so, but to favour the newest available unit(if they are sorted by availability in the export_descr_buildings file, that should be very easy to program even). Now, there's some work for you CA.:whip: ~;)

I'm having my hopes up for kingdoms. And on a sidenote, I notice the AI doesn't recruit knights all that often, they seem to prefer the dismounted version, which leads me to believe they actually do look at the stats(since I improved attack of twohanders, they actually recruit these in higher numbers as well) so erm on one hand it may explain why they love artillery(63 attack...) and on the other hand this needs a fix, I'll just be blunt here and say the recruitment routines are outdated or maybe someone was just lazy or whatever.:sweatdrop:

[econ21]

I wonder what would be a good test of the battlefield AI in the various games? I think some custom battles would be useful. Doing some small scale custom battles against the AI for a RTR PBM really brought home to me the limitations of the RTW battlefield AI that had been hidden in a lot of campaign play. (The biggest problem I noticed was a "blobbing" effect, whereby the AI would pile in on a few of my units. This is wasteful, as only the front of the blob can fight - the rest of the AI units are wasted milling around at the back of the blob. By contrast, the rest of my force is free to flank and destroy the blob. Basically, the AI would not attack on a broad front and tie down my force. In M2TW campaign battles, I don't observe this - often the whole of my line is engaged and so the AI can often bring superior numbers or quality to bear.)

If I were to do a test, I would try to choose a map with a gentle slope and defend it with a small force - say 5 spears, backed by 3 archers and a couple of flanking cavalry, with a general at the rear. Let the AI attack with a 50% bigger force.

Then maybe reverse the situation.

I suspect STW and MTW would give similar results. M2TW would be better than RTW. How M2TW would compare with STW/MTW is what is in dispute. I think it would be comparable.

Is such an exercise worth doing? I might be tempted, but if people would just dismiss it, I wouldn't give it any more thought.


Well, if I think of Mongol and Timurid stacks with all those experience advantages, they can be quite tough to beat, depending on where and with what you face them.

Yes, the Mongols/Timurids are very souped up. But when I referred to them I was thinking of the strategic AI: do the stacks tend to work together for mutual protection in defence and to overwhelm in the attack?

In BI, I think the horde stacks generally did not work together. You could often be attacked by one stack at a time in the same turn. [With the STW/MTW risk style map, this issue of coordination did not arise - the AI would be able to bring its numbers to bear, although typically that meant exhausting multi-wave battles.]

In M2TW, I get the impression AI stacks do work together more but I have not got enough experience to be sure. Fighting the Mongols and Timurids would be a good test, because they have multiple stacks and so you could see whether the AI uses them for mutual support.

[Tambarskjelve]

The only evidence i have seen of AI stack work together, was in my Polish campaign, where the Timurids came westwards from the steppes.
My previously hidden cavalry attacked a lone Tim stack, inflicted heavy casualties and retreated.
The next turn, all the timurid stacks drew up in a tight cluster, never to send out "loners" again.
Might be coincidence, though.

[econ21]

The next turn, all the timurid stacks drew up in a tight cluster, never to send out "loners" again.

Yes, the first wave Mongols "clustered" quite efficiently in our HRE PBM over in the Throne Room. I was thinking I could pick them off one, or at most two, a time as with the BI hordes but they made that hard. (Making many of their generals night fighters also helped.)

[Husar]

I think the cluster a bit too much and thus don't get to conquer anything until about 100 years after their appearance. Seems like every time reinforcements arrive they move back to meet them etc. And sometimes they almost get stuck walking around in the mountains to the east, walking abck and forth until years later they finally start to capture one of their target cities and spread like they should.

[Didz]

Pardon me for saying so, but you have no clue what you are talking about, this paragraph is just wrong, and shows IMO a poor understanding of the game.
I shall ignore this obvious insult and merely deal with the facts as you choose to present them.


There are two tactical components in the battle gameplay; these are in use whether there are territorial (or other) advantages or not: match-ups & flanking.
Actually, that’s three but who’s counting:

I agree the three tactical components are:
- Effective exploitation of terrain.
- Effective exploitation of troop qualities.
- Effective manoeuvre

However, if the battlefield situation is completely balanced as you originally suggested then the AI has absolutely nothing to work with and so the result is purely random and not a test of the AI at all.

At best all you are really testing is reaction not intelligence. To understand this you need to consider what the AI routine should be doing. For example, it should be conducting a mathematical assessment of the risks associated with the various options it has and looking for the one which gives it the most chance of success.

So it should be analysing the terrain within reach to see if it can gain any advantage from it, it should be matching its units against its opponents units such as it gains maximum advantage from each conflict and it should be manoeuvring to place its units to gain maximum position benefits.

However, assuming that everything is equal as you suggest then the net result of this analysis will be zero. Any terrain advantage it can gain will also available to its opponent, any beneficial troop match it makes will expose one of its other units to a similar counter threat of equal disadvantage and everyone knows that flanks attacks only work if you either have enough troops to pin your opponent in place or you can catch them off guard.

Therefore, having concluded just like the computer in the film ‘Wargames’ playing ‘Tic Tac Toe’ that nobody can be assured of victory the AI should either withdraw and avoid battle until it has a calculable advantage or it will perform a random default ‘do something’ action which cannot be justified by its assessment.

So, my issue remains that if the playing field is totally level, what exactly can you be testing in the AI routine other than the fact that it has a default ‘do something’ action.

Despite no advantages initially, little inequalities that present themselves gradually as the battle progresses from the skirmish phase to the melee, to the finishing.
And those little inequalities come down to luck, which was the point I was making.

Obviously, as luck begins to affect the balance of forces the AI then has calculable advantages and disadvantages to deal with. However, by this point it has already committed itself to some random action based on its ‘do something’ sub-routine which may already have placed it in a bad situation.

A good human player would never deliberately commit his army to a battle where the outcome was simply down to luck and so testing the AI on how well it reacts to a lucky break is actually testing it for something which by rights we would not want it to rely upon anyway.

The whole issue with the AI at present is that it does not present sufficient challenge and one of the main reasons it doesn’t is because it fails to engineer situations which give it an advantage. Worse still it seems willing to engage in battles and sieges where it is at a disadvantage and is therefore relying purely on luck for victory.


You don't seem to differentiate between advantages and "advantages" (a little bump against a hill), which does happen a lot
There should be no difference. If a little bump is all that’s available then the AI should be as keen to exploit it as if it was a mountain.


The point I am trying to make about the maps is that no "AI" then would be able to win (against an opponent defending a hill with relatively equal forces), no matter how much SP players want him to; since no player can win it also - statistically speaking.
Well actually point is that if the AI attempted to attack a human opponent defending a hill with relatively equal forces to its own then it has already failed its intelligence test before the battle even begins.


Sorry, but the MP community is the only sort of players I know that do care about balance of maps/units/unit types (to varying degrees admittedly and that's because its no fun for them to beat up an opponent when they defend a huge hill - so if you mean that they go for that sort of advantages to my experience they don't and they will (bitterly) complain if you do or drop out of your game.
Then your very lucky to belong to such a community, as everyone I’ve ever been involved with will use every trick in book to make sure the odds are stacked in their favour before the battle even starts, Including hacks and cheats if they can get hold of them.

[Kobal2fr]

If I were to do a test, I would try to choose a map with a gentle slope and defend it with a small force - say 5 spears, backed by 3 archers and a couple of flanking cavalry, with a general at the rear. Let the AI attack with a 50% bigger force.

Then maybe reverse the situation.

I suspect STW and MTW would give similar results

I suspect M2 would come out on top of such a test. Now, I've never read Sun Tsu's Art of War, so I can only guess that his answer to the question "What to do when facing a well-entrenched, missile heavy force on high ground ?" is "endeavour not to". However, I'm fairly certain that if nudged enough old man Tsu (or is that old man Sun ?) would come up with something better than "prance back and forth like a pillock, under fire the whole time".
At least the M2 AI knows enough to bumrush its way out of situations like that. Heavy missile superiority is not that much of an advantage in M2, because the AI will close as fast as possible. MTW will try to get "fancy". And die.

Frankly, I'd be very interested in the results of such a test, and it would certainly go a long way to putting an end to all this ... rethoric :grin:


You should, if you wish to express an objective opinion on the AI performance. You should be basing that opinion then in objective criteria - ie you should be making conditions that do not favor any result and then see what happens.


Nope. Cause it's never gonna happen in the field. I want to know it can deal with unbalanced forces, unbalanced terrain, unbalanced numbers, going either for or against it.
I want the AI to be able to exploit all the advantages terrain and match-up provides it, I want it to be underhanded and cunning, I want it to punish every mistake, and I want it to be able to minimize the effect of MY advantages.

If it's outnumbered, I want it to go for my general first, and with everything it's got. If it's outnumbering me, I want it to make me feel it and never give me a breather, etc... THAT's good AI.

If the AI can severly beat me on an open plain, with exactly the same forces on both sides, then it's not really good AI, it's cheaty AI (Let me entertain the thoughts than I'm not such a horrible general that I wouldn't know about flanking with cav, flanking with archers, keeping reserves etc... On the whole, there's not a lot of room for creative generalship on a flat, featureless plain. In the end it all comes down to who fails his morale roll first.)

The big letdown for me in M2TW is not the battlefield, but the strategic AI - it simply doesn't know how to bring the full economical force of its empire to bear, and enters too many fights it cannot win. And that's not desperation either - I've been attacked by "number 1 everywhere" Milanese, while I was scraping for militia myself, but they still insisted on sieging my castles with forces proportionate to the defenders. Needless to say, the sieges weren't a great success, when frankly with the armies Milan fielded at the time, and the state of my own forces, they could have been in Rennes by Christmas.

By comparison, and I agree with TinCow here, the battle AI is rustic and unimaginative, but it's functionnal enough (sieges excluded. By the gods, the AI wouldn't take a sand castle with the whole US Army. It does better when it fields proper siege engines though, but when it only has rams and towers... yeh gods :skull:).

[Noir]

Originally posted by Didz
I shall ignore this obvious insult and merely deal with the facts as you choose to present them.

It wasn't meant as an insult - i apologise if it came about as such.


I agree the three tactical components are:
- Effective exploitation of terrain.
- Effective exploitation of troop qualities.
- Effective manoeuvre

Partially Agreed. Your second point i term match ups that is straight engagement of units - flanking i term side/rear engagement of units that are locked in battle with reserve units/detachments: this is usually to offset unfavourable matching in trying to turn the tables.

Your first point i included and i commented that is acceptable IMO only to the point that the battle remains equally fair to both sides - i doubt that its a gameplay component as it influences the two others rather than define them (much like the weather) - however that's a matter of definition, i guess.


Originally posted by Didz
However, if the battlefield situation is completely balanced as you originally suggested then the AI has absolutely nothing to work with and so the result is purely random and not a test of the AI at all.

At best all you are really testing is reaction not intelligence. To understand this you need to consider what the AI routine should be doing. For example, it should be conducting a mathematical assessment of the risks associated with the various options it has and looking for the one which gives it the most chance of success.

So it should be analysing the terrain within reach to see if it can gain any advantage from it, it should be matching its units against its opponents units such as it gains maximum advantage from each conflict and it should be manoeuvring to place its units to gain maximum position benefits.

However, assuming that everything is equal as you suggest then the net result of this analysis will be zero. Any terrain advantage it can gain will also available to its opponent, any beneficial troop match it makes will expose one of its other units to a similar counter threat of equal disadvantage and everyone knows that flanks attacks only work if you either have enough troops to pin your opponent in place or you can catch them off guard.

Therefore, having concluded just like the computer in the film ‘Wargames’ playing ‘Tic Tac Toe’ that nobody can be assured of victory the AI should either withdraw and avoid battle until it has a calculable advantage or it will perform a random default ‘do something’ action which cannot be justified by its assessment.

So, my issue remains that if the playing field is totally level, what exactly can you be testing in the AI routine other than the fact that it has a default ‘do something’ action.

And those little inequalities come down to luck, which was the point I was making.

Obviously, as luck begins to affect the balance of forces the AI then has calculable advantages and disadvantages to deal with. However, by this point it has already committed itself to some random action based on its ‘do something’ sub-routine which may already have placed it in a bad situation.

A good human player would never deliberately commit his army to a battle where the outcome was simply down to luck and so testing the AI on how well it reacts to a lucky break is actually testing it for something which by rights we would not want it to rely upon anyway.

The whole issue with the AI at present is that it does not present sufficient challenge and one of the main reasons it doesn’t is because it fails to engineer situations which give it an advantage. Worse still it seems willing to engage in battles and sieges where it is at a disadvantage and is therefore relying purely on luck for victory.

Disagree - "Luck" or randomness in the outcome of a TW battle isn't large enough when the balance of forces is equal to determine the result. It can be made to if the exact same moves are made yes - but the exact same moves aren't made ever. Statistically it sin't a possibility as the system is dynamic and not deterministic.

What comes across from your post is that a game that's "on the line" is decided by "the hand of god" in TW. It isn't.

Maneuvering around the enemy in order to skirmish, attack & flank, still happens whether there is unequal terrain or not. The relative positions of units themselves provide for advantages and disadvantages during the game that have nothing to do with "luck". The AI instinctively aknowledges good/bad match ups and moves around in order to achieve the good and avoid the bad. That is part of his "intelligence" - and his competence in that should be judged under equal conditions - not when it is being in a disadvantage (say his match ups will fail because the player holds the high ground or because the player will counter them with reserves or because the player's units are stronger).

In skirmishing its the same thing - you are being outshooted (or not) because the opponent (player or AI) manages his shooters at better positions or more effectively (or not). There is an amount of luck/randomness in that you can't predict all the situational factors at all times as the game is dynamic and the accumulation of small things can have an effect in the overall result - but the player can be in charge of these (if he has mastered the interface) - they are not (lead to) random results, they depend on the player's/Ai's decisions/actions.

In fact part of being a good player is managing the randomness in the game - in order to overcome the disadvatages - press the advantages and achieve results.

If luck could purely determine the outcome of equal power armies clashing in flat terrain then the game has nothing to do with tactics altogether - let alone a tactical AI.

Advantageous match ups can be achieved in such a situation by various tactics: for example by unbalancing a flank and luring the opponent (AI or not) to attack there; then you rely on the strong flank to rout the rest of the enemy and the weak flank to hold the assault while the victorious flank units come back to finish them off. Success depends upon the judgement of the relative positions and of the relative strength as well as in the execution of the maneuvers.

This is only one of many strategies - it largely depends on the army composition and playing style of the player and that the composition dictates.

There are indeed two qualities of the tactical AI to be checked as you say: the overall strategic decisions and army control. These two intermingle though especially the more closer two armies come to engage as well after they do engage - then the situation is continuously changing and so strategic reassessment and what you call "reactions" happen at the same time and influence so much that are inseparable (until the end game where you regroup).

Initial assessment and plans govern a battle the more unequal the initial situation is (as you imply) - they don't though if it is equal - continuous assessment is needed in that case and skill (army cotrol) also plays a significant part. That is you have to be good at assessing, deciding and executing a plan and if things change follow them by adapting to the course of the game.

That ensures tactical depth (devise ways to tip the situation - grasp and accumulate little advantages - adapt to the opponent) - while in the case of unequality tactical depth is low (just use the advantage whether that is terrain/army/whatever-is-to-your-favor to win).

The AI is not advanced IMO enough to plan for such things in most cases as you mention (create situations) (they are embedded in the default formations he assumes, i guess) - but he's surely advanced enough to exploit holes and respond to changing density in formations broadly. He's also aware of the balance of forces generally in what when he has an advantage in melee he goes for head-on attacks - so he assesses and creates opportunities to some - lesser - degree.


Originally posted by Didz
There should be no difference. If a little bump is all that’s available then the AI should be as keen to exploit it as if it was a mountain.

The AI may or may not exploit it depending on whether he attacks/defends, based on the relative potential of the armies and other factors IMO - however its the potential of the bump or hill to work for or against the AI or a player that i am talking about.

If you play defense on the hill, then the AI or anyone else will lose. If he's called a "bad AI" because he "can't win" that one - then there's something wrong IMO.

In fact in SP, battles against the AI defending a hill are won because the AI will lose his position when the player advances from another side. This is achieved by exploiting the "refacing" routine of the AI. Is the AI bad or good in that case? In your opinion is obviously bad. In my opinion its beyond his capabilities so i storm the hill from the side he defends head on in SP; i might not always win and i might have a lot of casualties but i get a much more challenging and fun battle than i would if i had bypassed him, forced him to reface and lose the high ground in the process and fought the battle in equal or even higher ground.


Originally posted by Didz
Well actually point is that if the AI attempted to attack a human opponent defending a hill with relatively equal forces to its own then it has already failed its intelligence test before the battle even begins.

Very debatable and sort of unrealistic for the AI as it stands. I stated previously that in many cases the AI attacks because he's on the red - or other times because of hardcore limitations that tell him to "attack" in order to fulfil the "challenge" requirement from the part of the developer and the "TW" title of the games. In fact even the player could be (rarely because the player is richer most of the time) forced in such an attack if he is desperate to achieve a conquest result due to pressure - its not correct that you label it a bad move a-priori.

The developers IMO should have been designing the game around what the AI can or can't do well instead. They don't though, because they'll lose customers (fewer options) and the result is that the game lacks challenge.

Then many expect the AI to be able to do what they do in SP, that is exploit this or that, when he can't - exploiting single weaknesses/deficiencies isn't how AI's operate.

Good game design should ensure that these deficiencies are out of the game -then you and i have less to exploit and the AI has more equal chances and challenging games are a possibility again. All mods work towards that goal to a higher/lower degree - that is tune the game with how the AI works.

The TW AI was based originally around certain principles (such as a strong RPS) that the developer is slowly abandoning or overblending - but there is little evidence that the AI is adapting to the new conditions and game workings or that he can be made to do so at the moment for that matter.


Originally posted by Didz
Then your very lucky to belong to such a community, as everyone I’ve ever been involved with will use every trick in book to make sure the odds are stacked in their favour before the battle even starts, Including hacks and cheats if they can get hold of them.

There are good and bad apples everywhere, i guess - and again it has less to do with luck than you imply IMO - one can choose to play with people that he feels fiendly with and respect and not with people that he doesn't. I am honestly sorry if all you came accross is people that do all this.

If you read the MP forums here and elsewhere you'll find out that balance is the primary element of discussion whether talking about units-maps-gameplay-types of units-unit pricing.

MP is affected by that much more than SP - however SP is also affected - how many of the battles of a campaign are not a chore task and how well one remembers the rest of them?

Many Thanks

Noir

[Noir]

Oroginally posted by Kobal2fr
Nope. Cause it's never gonna happen in the field. I want to know it can deal with unbalanced forces, unbalanced terrain, unbalanced numbers, going either for or against it.
I want the AI to be able to exploit all the advantages terrain and match-up provides it, I want it to be underhanded and cunning, I want it to punish every mistake, and I want it to be able to minimize the effect of MY advantages.

If it's outnumbered, I want it to go for my general first, and with everything it's got. If it's outnumbering me, I want it to make me feel it and never give me a breather, etc... THAT's good AI.

Perhaps - but it seems to me that the AI operates upon certain routines that are triggered after some assessment rather than being able to identify singularities and exploit them as you (and many others apparently) expect.

As good as your propositions may sound - they also sound (to me) equally unrealistic (at this stage at least). One of my points is that AI performance involves the design of the campaign&battle game as well as the AI competence. For the truth of it - notice that mods follow the way the AI operates and go to painstaiking efforts to restore balance of various kinds (unit balance - army balance - faction balance - etc). It obviously helps the game being more challenging.

However everyone seems to expect that the AI will beat them at their own game of exploiting this or that: fair enough. You might be waiting more than you think though from my perspective.

There is a way to get all that your heart is longing for, from TW battles though, instead of posting in the Citadel about it : get online - i'm sure plenty of people will be delighted to play with you and perhaps you'll see then after some tens of battles, what sort of "generalship" can be achieved in flat terrain.

Many Thanks

Noir

[Didz]

What comes across from your post is that a game that's "on the line" is decided by "the hand of god" in TW. It isn't.
Nope! its determined by the combat sub-routines random variable generator, God has nothing to do with it.

The point I was trying to make, and have been since the start of this debate, is quite a simple one. Namely that if you present the AI with a totally balanced scenario where neither it nor its opponent has any advantages then any action the AI takes cannot be a test of its performance because any risk assessment it performed would be unable to establish a preferred reaction.

I realise that I have failed miserably to get this point across but I shall have one more attempt before I give up.

Lets look at an extreme example:

Two opposing but identical spear units facing each other across a totally flat peice of terrain. No wind, no sun in eithers eyes, absolutely identical conditions and stats. The only difference between each is the colour of their uniforms. The AI assesses the situation and comes up totally blank. There are no advantages to be gained by moving, there are no disadvantages either. If it attempts to move to a flank the enemy will change frontage. Worse case scenario is that if it moves it loses unit cohension therefore it stands and does nothing. Therefore, its default 'do something' sub-routine kicks in and forces it to perform a random action. It advances, the result could go either way, but the result was not initiated by the AI but by the default 'do something' routine which overrode the AI to make it attack.

Lets look at two other examples:

CHESS PROGRAMS: Chess programmes probably have the most sophisticated AI routines in the gaming industry, some can even beat grand masters. But no chess programme AI routine can generate an opening move, simply because at the start of a chess game there are no advantages or disadvantages to compare. The board is totally balanced and so if given the problem an AI routine will go into introverted analysis for ever in the hope of finding one. To avoid this Chess programmes are given a database of standard opening moves, sometimes five or more moves in a sequence, which the program will select from at random as a 'do something' strategy. So, in the first few moves of most Chess games against a computer you are not actually playing the AI at all by a random move sequence hard wired into it.

and finally an example from an AI used in real warfare:

Onboard ship to ship defence systems are heavily control by computerised auto threat management and response computers, with highly sophisticated AI systems. However, even they cannot make decisions based upon a totally balanced scenario. This situation was never considered likely to happen in real warfare and so was initially overlooked, until a ship found itself under attack from two identical missiles fired from exactly the same distance but from different directions. The AI routine went into 'assessment mode' looking for any factor which would help it decide which missile to deal with first and continued to monitor the alternative threats but do nothing until both missiles hit the ship. Now all such battle systems have built in 'do something' sub-routines which will kick in and override the AI to force the ships system to attack one or other of the threats at random.

That was really the only point I was trying to make, and I could not see how one would judge the performance of an AI in a situation where effectively it isn't at the helm but has been overridden.

It not an important point, but if one is then going to judge the AI on what happens its worth noting that in fact it probably isn't responsible for the end result. Its actually the 'do something' sub-routine that your testing.

However everyone seems to expect that the AI will beat them at their own game of exploiting this or that: fair enough. You might be waiting more than you think though from my perspective.
Certainly the best AI would be one that monitored both its own and its opponents performance and adjusted its own strategies to compensate.

In other words an AI which learns from its own mistakes and adjusts its own playing strategies to expliot not just the immediate situation it is in, but the long term behaviour of its opponents.

There have been some attempts at this sort of AI before but so far its yet to become common in gaming.

[TevashSzat]

The ai imo has improved since RTW. That being said, it still has much faults. Personally, I think the battle ai is half decent like it wont do anything that incredibly stupid, but any decent player can easily win outnumbered. The bigger problem is the campaign map ai. They just have a horrible sense of economics and somehow thinks that they must spend every penny they have every turn or someone will come out and take it. What ends up happening is that they spend a huge amount of money creating this huge militia armies whose upkeep doesn't allow them to upgrade their cities hence forcing them to only rely on militias until late in the game.

[Noir]

Originally posted by Didz
Two opposing but identical spear units facing each other across a totally flat peice of terrain. No wind, no sun in eithers eyes, absolutely identical conditions and stats. The only difference between each is the colour of their uniforms. The AI assesses the situation and comes up totally blank. There are no advantages to be gained by moving, there are no disadvantages either. If it attempts to move to a flank the enemy will change frontage. Worse case scenario is that if it moves it loses unit cohension therefore it stands and does nothing. Therefore, its default 'do something' sub-routine kicks in and forces it to perform a random action. It advances, the result could go either way, but the result was not initiated by the AI but by the default 'do something' routine which overrode the AI to make it attack.

I've run several such situations with the aim to "test" unit performance in melee. There are things that do make a difference in such a situation and the AI seemed always aware of most of them and used them :

1. Attack the opponent partly from the flank/rear: the AI marches diagonally to the enemy unit aiming to "hit" partially the flank in the charge. This little advantage in first kills with the charge may be the deciding point. The AI acknowledges it and uses it.

2. Hitting a full face of the enemy unit upon charging: if you manage to hit with all the front of your unit simultaneously the flank of an enemy unit at the charge, the charge will be devastating - if a few only men of yours do so, then the charge is not as effective. The AI seems fully aware of this and uses it in such confrontations.

3. Hold formation/engage at will: hold a charge with hold formation/engage at will after charge is over. The Ai doesn't seem particularly inept at doing this. I know though that for example if you pit cavalry that can do wedge he'll use that because its more effective on the charge. The AI uses the special abilities but not the melee modes.

4. Formation width during melee : it does affect a lot especially relative to how an opponent fights, for example : if the X spears are in hold formation in a medium size breadth formation, then the Y spears can be employed at an engage at will wider formation to catch both flanks, and "eat" the enemy formation from the sides.

Assuming a very small slope exists somewhere:
5. The AI will pace the unit he controls and move diagonally towards you along the slope so when you engage he has a small high ground advantage.

This is AI competence purely IMO - there is no "do-nothing" routine kikcing in IMO because there's plenty to consider. Moving around may give a potential advantage by the relative positions of the units ie how you approach the enemy unit - how much you and they are fatigued - how you fight them in melee upon engagement (you can also change the formation breadth during melee, by dragging it longer and click immediately attack, more of your guys will line on the flanks of the enemy then. Depending how the opponent fights - ie are they spears or swords? - it might give an advantage in kills or serve a purpose (hold up/pin more than one units)).

*edit*= Your example is also the only occasion where the gameplay is left essentially with a single match up (no better or worse match ups no match up component) and only some partial flanking (and not with a reserve unit) - that is it is dramatically restricted in possibilities as it has been reduced in gameplay components. Therefore it is inadequate IMO and for the truth of it try to imagine how much more complex the situation becomes if you add to both sides a sword unit. That is: a sword and a spear vs a sword. and a spear.

Will in the range of the possibilities of this new encounter the AI be tested or not? Now there is really plenty to do: pin the sword with your spear and aim for your sword to kill the enemy spear quicker. Double the enemy sword or spear. Let the enemy double your sword or spear with the aim to flank with the other. Pin both the enemy sword and spear with one of your units and take them peacemeal by flanking with your second unit. Pin the enemy spear with your spear and get the enemy sword by a better charge.

The Ai will not just "stay there" - even if he has no initial plan and he is adapting to your movements. Do nothing is not an option he'll take his chaces when the opportunity comes to defeat you.

The situation will become all the more complex if you add a horse and even more if you add a shooter. Equal forces - flat ground but immenselly more complexity of possibilities to test the opponents than your example.

It is in the exploring and executing of these possibilities that the AI is tested. Now try the same experiment adding a few upgrades to your units or putting them in a hill.

The odds are tipped to one's favor - unless the payer does a major/singularity mistake (giving up the high ground) there's no match up that he'll lose (due to the upgrades or the high ground) and flanking cannot be done effectively as the opponent is slower than you (he has to move uphill to flank you) nor does it have the same effect (if your units have upgrades). So which exactly quality of the AI is tested there?

The same goes if one has reserves and the AI has not.

Many Thanks

Noir

[Didz]

This is AI competence purely IMO - there is no "do-nothing" routine kikcing in IMO because there's plenty to consider even if the ground is flat.
Sorry, to be pedantic about this but there must be.

In fact, I would bet on the fact that some of the examples you just listed are probably the default 'do something' options hard wired into the AI.

The bottom line here is that a computer program cannot make a choice or exercise a reasoned judgement, it can only calculate mathematical odds. Therefore, if you present an AI with a scenario which is totally balanced it cannot calculate a solution. In the MOD scenario I mentioned the navy actually thought that the program had crashed or frozen and wasted months looking for a program bug. It was only after careful analysis that it was realised exactly what had happened. The video footage actually shows the missile defence systems controlled by the AI dithering between one threat and the other as it AI system desperately tried to calculate which threat was the greater during the few seconds prior to impact.

Presented with a perfectly balanced situation in a TW game the AI will do exactly the same thing, and we even have some evidence of this in the way troops dither about in battles and armies do in campaigns.

The only way to break this effect is to overide the AI and give it a direct arbitary action, which says if the threat assessment is balanced 'do this'. Therefore, we need to be careful when we set the AI tests that we are actually testing the AI, and not this 'do something' routine.

I would even go so far as to suggest that some of the passive AI complaints might be the result of the AI's inability to calculate a solution. I certainly don't see any evidence to support your statement that the 'AI will not just stay there' as I've explioted this very failing too often in both battles and seiges to inflict critical missile casualties on the AI controlled army. They do absolutely just stay there and let you shoot them, in fact, interestingly I've noticed that on some occassions they will let you empty your entire ammunition supply into them and then when you begin to withdraw your missile troops will actually try to advance. Suggesting to me that the detection of the movement to the rear is just enough to sway the AI routine into making a decision to attack.

[Noir]

Originally posted by Didz
The only way to break this effect is to overide the AI and give it a direct arbitary action, which says if the threat assessment is balanced 'do this'. Therefore, we need to be careful when we set the AI tests that we are actually testing the AI, and not this 'do something' routine.

The AI is an aggregation of the do-this or do-that routines in all probability - what to test is how all these perform in unison.

Your argument concerns only the initial conditions and not the course of a battle that can change them. The AI responds to that or even takes the initiative. Also you forget that when the AI "attacks" by default he will come at you and try to break you with his forces whether equal or not. If he is defending he will try to resist your assault whether he has equal forces or not.

Having an aim he takes an action - so the "do-nothing" option doesn't sound convincing for the TW Ai to me.

(PS please read the edit of the above post i was writing while you posted - thanks.)

PPS It might be a be-still is better than moving option (while in defence for example), however you seem to claim that the AI breaks down upon being confronted with equal forces. I sincerely doubt that - but its easy to check: just make up two identical 20 units armies in custom and pit them in flat ground. I bet that if the AI is on the offensive he'll come at you.

The "passive" AI - was a new bug in TW to the best of my knowledge - i haven't obsrved it in any of the previous games and certainly not in the two older ones. You seem to claim that it is an inherent property of the AI. It doesn't sound right to me.

Many Thanks

Noir

[Didz]

Your argument concerns only the initial conditions and not the course of a battle that can change them.
Not really, obviously the initial assessment is the key one in that it ought to provide the AI with a clear understanding of its strategic or tactical position and indicate the nature of its most promising solution. But one would hope that the AI is constantly re-evaluating its situation during play so that it can modify its initial decisions.

In fact, I'm sure it does this becuase as I've already said it is possible to play the AI for a sucker by feeding or denying it opportunities during play. Thus you can create false opportunities to draw it into traps and you can deny it opportunities to render it passive.


Having an aim he takes an action - so the "do-nothing" option doesn't sound convincing for the TW Ai to me.
And yet it is possible to acheive and therefore must be an option the AI considers valid under the right circumstances.


however you seem to claim that the AI breaks down upon being confronted with equal forces.
No, what I said is that no AI can calculate a solution if its assessment is faced with a totally balanced situation. It hasn't broken down, it merely has nothing to work with, thats why all AI routines have override routines that trigger a 'do something' action when the AI cannot make a decision.


I sincerely doubt that - but its easy to check: just make up two identical 20 units armies in custom and pit them in flat ground. I bet that if the AI is on the offensive he'll come at you.
Only becuase that is the most like 'do something' action programmed for that situation.


The "passive" AI - was a new bug in TW to the best of my knowledge - i haven't obsrved it in any of the previous games and certainly not in the two older ones. You seem to claim that it is an inherent property of the AI. It doesn't sound right to me.
Nope, that situation has been in existence since day 1, I've used it in STW, RTW, MTW and I'm still using it in MTW2. Arbalesta's were brilliant at explioting the passive AI in MTW for example, as were muskets in STW.

However, its not a bug and its wrong to call it one. In most instances it occurs because the AI has calculated that an attack cannot succeed and yet the morale state of its units does not justify a withdrawal. To overcome this CA have introduced some additional 'do something' triggers which seem to kick in if the AI units are suffering excessive missile casualties.

This 'do something' routine overrides the AI and forces the AI controlled units to make a pointless attack on the missile unit inflicting the most casualties. The result leads to the unit chasing the missile troops until the action is overridden by a fresh AI assessment that points out the futility and risk of the action and causes them to break off and return to their original position.

I exploit this conflict between the AI and the 'do something' routine a lot during sally battles as the AI recognizes the threat posed by the city wall defences but is constantly overridden by the 'do something' routine when hit by sallying missile troops. Thus you can kite the AI units into range of the wall defences, wait for the AI routine to kick in and order them to withdraw, then shoot them in the back with the sallying missile troops as they fall back triggering the 'Do something' routine to override the AI and order them to attack and repeat the process until you run out of arrows.

[Didz]

A simple test to prove the existence of the 'Do Something' routine.

Just to prove the existence of the 'do something' sub-routine here is a very simple test.

Step 1. A level playing field leaving the AI to decide.
https://img182.imageshack.us/img182/8758/ai1kj2.jpg (https://imageshack.us)
By Didz (http://profile.imageshack.us/user/Didz) at 1969-12-31
Two identical units of spearmen face each other over a totally level peice of ground. The AI commanding the Scots has been told it is on the defensive and so the 'do something' routine has not overridden the AI to force it to attack. Consequently the AI does nothing as it cannot calculate any solution.
I left these units staring at each other for 15 minutes before taking this picture nothing moved.

Step 2: A level playing field but letting the 'Do Something' routine override the AI.
https://img182.imageshack.us/img182/9637/ai2mz0.jpg (https://imageshack.us)
By Didz (http://profile.imageshack.us/user/Didz) at 1969-12-31
A second battle with exactly the same set-up, except that this time the AI has been told that it is the Attacker. Therefore, when the battle engine detects that the AI is unable to calculate a solution the 'Do Something' routine kicks in to override the AI and force it to attack. The affect was instantaneous with the Scots advancing as soon as the battle started.

Step 3: Proving that the AI does exist.
https://img257.imageshack.us/img257/1062/ai3pj9.jpg (https://imageshack.us)
By Didz (http://profile.imageshack.us/user/Didz) at 1969-12-31
Finally, just to prove that the AI does exist and will calculate a solution if it has the potential to do so. The third test puts the AI back on the defensive but this time gives it more troops and thus a numerical advantage. As you can see in this final test despite being on the defensive as in Step 1 the AI has calculated the odds realised it has an advantage and has ordered an attack.

Conclusion
As predicted, if the AI is presented with a totally balanced scenario and left to its own devices it will do nothing. However, if the 'Do Something' routine is told it is the attacker, it will override the AI and force the AI to attack even though it still has no advantage. Finally, if given an advantage the AI will move to expliot that advantage even though it is contrary to its mission.

[redriver]

I am as my join date suggests a long time lurker, fairly recent member and dedicated fan of the TW series, well the last part isn't totally accurate and is valid only until MTW which is why this query is being posted.

Needless to say RTW was in most respects a complete disappointment for me, especially with regards to both the strategic and the tactical AI. In fact in my opinion the overall game was sub-standard enough to put off my purchase of MTW II till the release of the initial set of patches. Now that I am contemplating purchasing the game I would like your input on certain matters.

A continuous perusal of the forum seems to indicate a significant improvement in the AI, however that by itself is insufficient so I intend to ask a specific set of queries to judge the actual level of improvement:


Firstly, one of the major problems with RTW and BI was the inability of the AI in a majority of situations to maintain a battle line. Even with Darthmod formations the AI would hold the battle line only until it was about 100 mts away from my battle line- at which point it would decide that to utilise tactics from the Homeric age and would then engage my battle line one individual unit at a time deliberately destroying the cohesion of its battle line.Has this major defect been rectified, because if not then we can end this thread right at this point for without the AI attacking as a cohesive whole there is no point in father inquiring about any subsequent improvements.

False. This is not how it works in RTW 1.5

Secondly, is the AI able to deal with severely depleted units or does one still face 20 unit AI stacks that have a roster of less than 200 men. Infact does the AI utilise any form of automerge or does it ever retrain its depleted units?]

Thirdly, does the AI always aim to secure terrain and height advantages as It did in MTW and does it seek to consistently outflank my battle line?]

It will do that in RTW 1.5 dependin' on cercumstances...

Finally, is the AI able to co-ordinate multiple stack armies to present me in a position of numerical inferiority?]

As stated above your response on the following points will be appreciated, so I wish to thank all those who contribute to my queries for their time and input.
~:thumb:

don't have MTW2 so my points are regardin' RTW 1.5 instead...

[Noir]

@Didz,
i disagree with all of your points including the "illustration" (in which you've used two exactly same single units despite my argument that this detracts gameplay components); however instead of going on with another long post - i'll leave it here. I think we both expanded our arguments enough and there is no need to further post, as far as i am concerned.

BTW, our observations generally match - its the interpretation that differs (as it happens in these cases).

Thanks for the exchange of arguments and sorry if things got heated unecessarily - it was entirely my fault.

Many Thanks

Noir

[Alsn]

Didz, i think that youre way overthinking this "do something" routine that you claim exists.

From your post with your test i would draw the following conclusions:

The first example(where the AI did nothing) the AI was on the defensive. In this game(M2:TW that is) if the time runs out then the defenders win, thus if you are presented with exactly equal odds then the logical thing to do would obviously be to just wait(if the opponent doesnt respond it guarantees a victory).

If the AI is on the attack then waiting will guarantee a loss which means the only possibility for the AI is to attack(its still 50-50 if they attack).

If the AI is on the defensive but has a numerical advantage it logically concludes that that it shouldnt be able to lose and thus attacks.

I fail to see how these arent purely logical decisions and instead supposedly part of some "do something" routine you claim to exist.

[Didz]

@Alsn
Yes, thats why step 3 was important. Step 1 and 2 could easily be explained by the existence of a 'do something' trigger which says something like if your on the defensive 'do this' if your on the attack 'do that'.

What step 3 proved is that there is a evaluation routine which assesses the odds and can make a decision to attack if the odds favour such an action.

Basically, what your saying is that step 1 and step 2 could be the result of a 'do something' trigger without any AI evaluation, and I think thats possible if unlikely. However, I think it would be hard to argue that step 3 was the result of anything other than an assessment of relative strength by the AI.

The question I raised right at the start of this saga, was if the test scenario was totally balanced, how would one establish that the actions selected by the computer were the result of the AI assessment and not merely the default 'do something' routines hard coded into the battle engine.

All of this of course assumes that there is an AI routine. I agree with you that the entire behaviour of the computer might be governed by nothing more than a massive list of 'do something' triggers as you seem to suggest. It would certainly explain some of it 'dumber' actions but it seems unlikely given the wide variety of possible situations that the computer would have to be pre-programmed to react to.

Nevertheless, traits are triggerred by such a massive list of triggers, so I suppose its possible that no AI system exists.

[R'as al Ghul]

Step 3: Proving that the AI does exist.
https://img257.imageshack.us/img257/1062/ai3pj9.jpg (https://imageshack.us)
By Didz (http://profile.imageshack.us/user/Didz) at 1969-12-31
Finally, just to prove that the AI does exist and will calculate a solution if it has the potential to do so. The third test puts the AI back on the defensive but this time gives it more troops and thus a numerical advantage. As you can see in this final test despite being on the defensive as in Step 1 the AI has calculated the odds realised it has an advantage and has ordered an attack.

Conclusion
As predicted, if the AI is presented with a totally balanced scenario and left to its own devices it will do nothing. However, if the 'Do Something' routine is told it is the attacker, it will override the AI and force the AI to attack even though it still has no advantage. Finally, if given an advantage the AI will move to expliot that advantage even though it is contrary to its mission.

Hello,

sorry for coming so late into this discussion. Didz and Noir make both interesting points. I can't possibly comment on all of them but I'm inclined to agree more with Noir. Anyway, this third step that Didz posted seems odd to me. Yes, the AI has a numerical advantage and it's right to attack in that situation. What I don't understand is, why it doesn't attack with both units at the same time. Judging from the picture and my own tests the AI doesn't seem to coordinate its units. If it had walked its general unit to the back of the enemy unit it surely would've had better results in terms of casualties. Shouldn't the AI be aware of that and wasn't that the case in STW and MTW?

How did this test play out Didz? Was the general just standing there, doing nothing or did it eventually join the fight?

I've already posted this somewhere else, but imo it should be almost impossible to beat the AI if both armies are equal and on flat terrain. The unbeatable advantage of the AI should be numerical assessment of the situation (terrain, morale, weather, unit strength, etc.), its reaction time and the ability to manage 20 units simultaneously.
This poses the question: What exactly is the AI doing anyway?
I'd love to hear a response from CA to that.

R'as

[Kobal2fr]

My guess is that in the 3rd test, the AI doesn't wish to commit its general for fear of losing him to a bad die roll. After all, it doesn't need to to win : either its spears win the 1-on-1 fight and it wins, or the player wins the fight with a severely depleted unit that the general unit is sure to beat. Whereas should both units attack, and the player kills the AI captain, both AI spears could end up routing...

[Didz]

How did this test play out Didz? Was the general just standing there, doing nothing or did it eventually join the fight?
To be honest I never bothered to play these battles out to a conclusion as it wasn't necessary to establish the point I was testing. But I agree with you that whilst the AI seems to have decided that attack was a good strategy it doesn't seem to have made best tactical use of its resources.

That second unit did just stop where it is shown and watched its partner fight, whereas I'm pretty sure a human player would have moved it round the enemies flank or rear as you suggested.

Again this could be a random trigger that chooses each units tactical response and perhaps the programmer thought flanking moves inappropriate for spearmen. It depends really on where the AI breaks off in terms of its assessment, perhaps it doesn't get involved in the close in tactical stuff.

If such close in tactical decisions are determined by random triggers then that might also explain why attacks by the AI are often so fragmented. Its has certainly puzzled me in the past why the AI would calculate that an attack was a good strategy and then only send one unit to be slaughtered at a time instead of its entire battleline. Likewise the fact that it doesn;t always do it, suggests some sort of random factor is involved.

[PseRamesses]

Firstly, one of the major problems with RTW and BI was the inability of the AI in a majority of situations to maintain a battle line. Even with Darthmod formations the AI would hold the battle line only until it was about 100 mts away from my battle line- at which point it would decide that to utilise tactics from the Homeric age and would then engage my battle line one individual unit at a time deliberately destroying the cohesion of its battle line.Has this major defect been rectified, because if not then we can end this thread right at this point for without the AI attacking as a cohesive whole there is no point in father inquiring about any subsequent improvements.

Secondly, is the AI able to deal with severely depleted units or does one still face 20 unit AI stacks that have a roster of less than 200 men. Infact does the AI utilise any form of automerge or does it ever retrain its depleted units?]

Thirdly, does the AI always aim to secure terrain and height advantages as It did in MTW and does it seek to consistently outflank my battle line?]

Finally, is the AI able to co-ordinate multiple stack armies to present me in a position of numerical inferiority?]

As stated above your response on the following points will be appreciated, so I wish to thank all those who contribute to my queries for their time and input.
~:thumb:

Sorry for joining the discussion late but to answer Amritochates´ questions: no, no, no and no!

Some of us, including me, has played TW since the first release of Shogun. And IMHO that game is a gem. Most of the time the AI really tries hard to maintain a favorable position, like a ford, hill etc. Many times I was located on top of the hill the AI would line up and send some cavs around my flanks. Not seldom both groups arrived at the same time.

In MTW the AI would still hold a ford/ bridge/ hill etc but was acting as individual units out in the open.

In RTW this behavior had been even more diluted (sp?) and you could actually lure the AI across a bridge or down from a hill.

In M2 the AI rarely do anything right, I´m sorry but I´m so fed up with this game going in the wrong direction. Everything has improved graphically but not gamewise. Anyone can be a desktop general with M2 but back with Shogun or M1 your battlefield maneuvering actually did matter.

One thing stands out as a huge improvement: siege battles, they are immensely fun to play and the AI holds favorable grounds, falls back on que drawing eager attackers into ambushes from sidestreets etc etc. I love playing siege-battles in M2 (and the graphs doesn´t make it less fun either).

[R'as al Ghul]

My guess is that in the 3rd test, the AI doesn't wish to commit its general for fear of losing him to a bad die roll. After all, it doesn't need to to win : either its spears win the 1-on-1 fight and it wins, or the player wins the fight with a severely depleted unit that the general unit is sure to beat. Whereas should both units attack, and the player kills the AI captain, both AI spears could end up routing...

Mmmh. If the enemy spear unit is engaged and the General walks around to launch an attack in the rear how can the General get killed? I think waiting for the melee to be decided offers a greater chance to loose the general because the AI has to launch a second frontal attack, which poses a greater risk to the Gen than attacking the rear.


If such close in tactical decisions are determined by random triggers then that might also explain why attacks by the AI are often so fragmented. Its has certainly puzzled me in the past why the AI would calculate that an attack was a good strategy and then only send one unit to be slaughtered at a time instead of its entire battleline. Likewise the fact that it doesn;t always do it, suggests some sort of random factor is involved.

Random factors would certainly explain a lot but I shudder to think that's purely random. I also think that any manouver should be available to all units if the situation is appropriate. Not having spears doing flanking manouvers would certainly be a programmed disadvantage or a simple omission.
I'm certain that the tactical AI in STW is better in that respect and that the same test would have played out differently. I may post some results if I find the time to test that theory.

[Kobal2fr]

Mmmh. If the enemy spear unit is engaged and the General walks around to launch an attack in the rear how can the General get killed? I think waiting for the melee to be decided offers a greater chance to loose the general because the AI has to launch a second frontal attack, which poses a greater risk to the Gen than attacking the rear.

Because flanked/rear-attacked elements in a unit will turn around to face the new threat. Sure, it's a lower chance of death than being attack frontally, but as they say, "next to none is worse than plain ol' none". And frankly, I'd rather have the AI use its generals timorously than recklessly. How many times have promising STW/MTW/RTW battles turned sour and boring because the AI just charged its general head-on and died stupidly ?

[TosaInu]

Because flanked/rear-attacked elements in a unit will turn around to face the new threat. Sure, it's a lower chance of death than being attack frontally, but as they say, "next to none is worse than plain ol' none". And frankly, I'd rather have the AI use its generals timorously than recklessly. How many times have promising STW/MTW/RTW battles turned sour and boring because the AI just charged its general head-on and died stupidly ?

Hello,

It's hard to judge whether it's too far away for moralesupport and a -hit, so we give it the benefit of the doubt there. By far the best action (keeping general safety in mind) is to move the generalunit closer to the action, ready to immediately backstab/flank in case unit one loses (against all odds). This will be a double cutting sword (which might already be the case) and provides the strongest position (in this case) if everything goes wrong.

More checks are needed, because the AI may 'know' it will win and just keep his general out of harm.

-What happens when there are more than 2 AI melee units? Two of them are not a general, so will it use quick hammer and anvil then?
-What does it do when it certainly needs more than one unit to defeat the player?
A. two AI units, so the AI general must fight.
B. more than two AI units, so the general may not have to fight.

[Kobal2fr]

Agreed on all points but one : if the general unit gets too close, then it gets too close to charge. The best charge distance for inf. units is 40 errr... meters ? squares ? game units ? Anyway, 40 of those. Or, 1/3rd shortbow range. And remember that spear units generally have a high charge bonus to go with their crummy combat stats and penalty against infantry, so unless 40... things is further than the morale boost area, it's probably a better idea for the general to stay back in case the 1-vs-1 fight goes pearshaped.

I wholeheartedly agree that said charge opportunity would probably be better used with the general on a flank, so that even if the first unit routs, it in turn gets a flanking chance should it rally later. But I also recall reading somewhere that units that are cutoff from their side of the map get a big morale penalty, so the AI might be taking that into account here, and not dare risk it.

[TosaInu]

Hello Kobal2fr,


Agreed on all points but one : if the general unit gets too close, then it gets too close to charge.

True, but it would get the complete benefit of a backstab, while the enemy unit could turn around and fight back when the general is keeping distance (1 cycle of combat is enough to get the AI general killed). It's a con and a pro vs another con and pro :juggle2: I think the unit would be seriously dented and not be able to do anything when receiving a proper blow in the back, instead rout instantly.


I wholeheartedly agree that said charge opportunity would probably be better used with the general on a flank, so that even if the first unit routs, it in turn gets a flanking chance should it rally later. But I also recall reading somewhere that units that are cutoff from their side of the map get a big morale penalty, so the AI might be taking that into account here, and not dare risk it.

Turn it around: the AI general could cut off the players side of the map (the player would be sandwiched). Indeed, of course still a chance that the AI general will get a fit too, but then it doesn't see propely what his ally is doing.

[R'as al Ghul]

Because flanked/rear-attacked elements in a unit will turn around to face the new threat. Sure, it's a lower chance of death than being attack frontally, but as they say, "next to none is worse than plain ol' none". And frankly, I'd rather have the AI use its generals timorously than recklessly. How many times have promising STW/MTW/RTW battles turned sour and boring because the AI just charged its general head-on and died stupidly ?

Yes, right, they can turn around now. I forgot. :laugh4:
Anyway, it hasn't happened to me before that the General gets killed in such a manoeuver and I think the chances are pretty low that it happens.
Of course I don't want any Suicide Generals either but that's something different.

[Didz]

I'm certain that the tactical AI in STW is better in that respect and that the same test would have played out differently. I may post some results if I find the time to test that theory.
Thats my recollection too.

I can recall that the AI in STW was very good at identifying beneficial terrain and pretty difficult to prize off it once it had established itself. Likewise the matching of units in close combat was also pretty effective, much more so than in MTW2.

However, I wonder how much of this is down to the sheer complexity of MTW2 combat as opposed to STW. STW after all had compartively few units types and all armies had the same whereas MTW2 has much more variety to cope with. Like wise I wonder how many of the current problems are due to the gradual introduction of more and more 'do something' triggers to the mix, which overrides the AI to force units to respond to specific situations. The one thing which is definately possible in MTW2 which was never the case in STW is the ability to 'kite' computer controlled units into suicidal attacks. In STW the AI simply stuck its nose in the air and refused to be suckered but in MTW2 the 'taking excessive missile casualties' trigger alone can be used to 'kite' specific individual units into traps by triggering a foolish attack move.

Random factors would certainly explain a lot but I shudder to think that's purely random.
By random I didn't mean totally random, but rather a random selection from a pre-determined set of actions the programmer/designer considered appropriate under those conditions.

So, for example if a unit is hit by missile fire and takes an excessive amount of casualties there is a pretty good chance that the 'do something' code will kick in and override its current action with an order to 'attack the missile unit'. That seems to be an almost standard response for all except missile units with ammo.

However, sometimes the attack of the first unit with trigger a more general advance, which suggests that there is a secodnary 'do something' trigger that says something like 'if the unit next to you attacks, then you attack'. However, this is not consistent, sometimes it happens sometimes it doesn't suggesting that it is only one of at least two random actions available when that 'do something' trigger is activated.

I suspect that over the course of the series more and more of these triggers have been incorporated into the code in an attempt to combat specific complaints about the performance of the AI. After all the easiest way to deal with complaints about a 'passive AI' is to code enforced actions into the battle enegine that make units do something.

However, the problem is that because they are triggered by specific events they do not take into account the wider tactical situation and so not only can they be explioted by human players but they frequently produce 'dumb' results.

[TosaInu]

Hello,

A quick test in MTW VI using equal infantry units, on a flat map.

AI attacking: general and another unit.
Player defending: general.
Both AI units attack immediately. The unit acts as anvil, the general moves into flank position, but doesn't coordinate well and hits its own unit first (thus cancelling the charge).


AI defending: general and another unit.
Player attacking: general.
Idem.

When I 'cheat' and run the last yards, the AI general gets into flanking (almost).



AI attacking: general.
Player defending: general.
AI attacks after a few seconds.


AI defending: general.
Player attacking: general.
AI waits, launches attacks when the player marches closer to a distance of 3 tiles. I guess fatigue, however small, is the perceived opportunity.

[FactionHeir]

Custom battles in M2TW are no way to test the AI.
Even if the AI is defending and you have a grand advantage, it will "do something" and attack you. It will also not get stuck if its army consists partially of missiles/artillery.

In campaign battles, this is very different, with the AI holding its ground as defender when outnumbered and even as attacker often getting stalled if it has artillery.

[Didz]

@TosaInu
That sounds a lot more sensible, the AI seems to be using its General unit sensibly in every example, or at least trying to.

@FactionHeir
I think that was where I came in...the key point being that campaign battles are pretty much never going to be balanced and so the AI will be able to calculate a solution from 'Start Battle' and seems to have more control over what happens.

I think much of the problem with sally battles is that the AI has no options at the start of the battle because basically all the enemy units are beyond its reach inside a city wall. For some reason this seems to get the AI off to a bad start from which it rarely recovers fully and certainly this is the type of battle where it is most easy to play the computer for a sucker.

[Orda Khan]

Well I must say I've had a good laugh anyway :laugh4: :laugh4: :laugh4: :laugh4: :laugh4: :laugh4: :laugh4: :laugh4: :laugh4: :laugh4: :laugh4:

......Orda

[Noir]

Just for the record:

I've run 6 custom games in MTW/VI : i've chosen 1 unit type of each for me and the AI, and played like this 3 games in the offensive and 3 in the defensive in a flat & featureless (no forests etc) map.

All battles were very good (that is hard fighting on the line to the end), and quite varied in their development - the AI reacted promptly to my deploying, skirmishing and melee, created and took opportunities, skirmished with his missiles, avoided being shot, protected his shooters as well as the General that he deployed at crucial stages (in one case to take out my own outumbered general). He also held reserves amd used them appropriately. His match ups were well calculated and he went for flanking attacks in melee nuclei in order to win them.

There was no sign of a do-nothing routine nor of any puzzlement of the AI in any of the 6 battles, nor of a "passive" AI if being shot.

Many Thanks

Noir

*edit*=all units for both sides were 0 valour and all were from the same roster.