why arent they talking at all about the AI?

[Silver Rusher]

IF a company decided to make a historically accurate game of the order of EB, with the dramatically reduced audience for such a game then they would have to charge a lot more, say theres 10% of the market that there is for RTR, they would have to charge £250 just to make the same kinda profits as CA.

Maybee CA could be incouraged to allow mods to use their engine like UT2004 has, the main reason i brought it was because of all the cool mods there are for it, its like buying 6 games in one!

How would charging £250 allow them to make the same profits as CA? No, seriously, I don't think there is a single person who would be willing to pay £250 for Europa Barbororum, sorry to the developers it's a good mod but that price is insane.

[BeeSting]

sorry about this edited below... didn't mean to spam

[BeeSting]

edited below

[BeeSting]

Instead of just moving x number of units with individual agendas towards the enemy but keeping them within a close enough distance of each other to appear as if they're moving in unison with one singular plan.

I think the problem occurs with having the attack/charge bonus for units via selecting or clicking the individual enemy units. This problem is most evident for a line of phalangite units, they have a hard time keeping a solid unbroken line as human players can expose this game feature by stretching them out, especially in defensive or guard mode the units become as in broken blocks tackling enemy lines in all sorts of angles instead of tackling parallel to the enemy line. in order to avoid this, a player can normally clicks behind enemy line and hit backspace as the line engages, sacrificing the attack bonus--but it's better than breaking up the solidity of your line--but for the AI.... let's just say my emersion is ruined when ever i see AI use phalanx units. This is just many problems with phalangites with AI--simply because of the game feature and AI is not spohisticated enough to think out side that feature to keep form. How to resolve this is beyond me. Just to take some pot shots at this I think they would have to recode the phalanx to approach the enemy differently more in reserved manner, not to order individual units to tackle their own targets but to appraoch the center of the line as a whole as if--as mentioned in my previous posts--say five or six phalanx units were as one unit. So they as one unit when selecting their target to attack it will choose the very center unit of opponent's line, as in phalanx unit in guard mode, so to not cave into that enemy unit in the center line, giving the attack bonus to all of it's line yet keeping the solid unbroken parallel line. I hope i made sense. FYI, English is my second language. Back to the point, this can be applied to all formations if the AI can organize its units in 2 flanks, 1 center, and perhaps reserves to fill in the holes.

[Duke John]

By asking multiple units to 'glue' themselves together the way you suggest you're actually asking for worse ai than the computer already uses. A group of five hoplite units being stopped by one unit of whatever is utterly pointless. It would cause all five of those units to sit and do nothing as you flanked or wrapped around and charged them from the rear, this is easy enough to do now the way it is, but you're asking for a free, resistanceless shot to the back.

Easy to fix that. Use the mass of the entire group formation. There is no way that 100 men in 2 ranks could stop 500 men in 10 ranks or the 100 men in 10 ranks but heavily outflanked. Well there is a way in R:TW because of the game mechanism but things like that can be coded. Soldiers already stick together in a single unit and use mass to push an opposite unit, the same can and should be applied to groups of soldiers.

[BeeSting]

If AI were to organize its units in 2 flanks, 1 center, and reserves to fill gaps in the line--it gives each unit specific roles that fall under these three categories--will give cohesiveness to the battle formation. This would be far superior to the seemingly random orders given out to "seek the weakest unit and charge it with your strongest" or "flank your opponent no matter if you are at the very center of the battle line" leaving a gaping hole in your own line to flank the enemy.

Center would be formed mostly of infantry, mainly to hold the enemy line, and the flanking units on its side are given the order to make sorties from the base (the center) to neutralize the enemy flanking units before charging the centerline on its sides. Current RTW AI just charges whatever that exposes its sides and rear, easily falling for traps. Cavalry units should not just charge at enemy infantry without engaging the enemy cavalry first. AI use of cavalry often if not always trashes its own first by throwing it away at the nearest enemy infantry units with the weakest stats or ones with exposed flanks or rear, which leads to its mounts being wiped out before the player's cavalry, by players cavalry, for exposing its flanks from engaging the player's infantry. Its decisive forces are gone within minutes after the engagement and the fate of AI's army is obvious--repetitive process begins for the player--clean up time.

Again flanks and rear reserves should work off of the center, the base. Specific roles for units in reserve to stay put till gaps occur in the battle line or to back up a wavering unit whether it is on the centerline or on the flanks. I don't know how hard it would be to code this to prevent battles from ending up to be a meaningless goose chase or all units doing their own thing oblivious to what others of its own are doing. I may be wrong in my observation but whatever method CA uses for its solutions, I hope the AI in MTW2 will be "improved" as promised to such extent that we will forget its faults in RTW.

[sunsmountain]

I don't know how hard it would be to code this to prevent battles from ending up to be a meaningless goose chase or all units doing their own thing oblivious to what others of its own are doing.

Exception leads to exploitation. In other words, if programmers program anything to prevent anything from happening, players can always find the condition on which this rests, and exploit this. This is particularly true in STW and MTW, where the AI simply relies on certain constraints in the map (defined pointers, initial formation before engagement, etc.) that cannot be exploited by the player. For randomly generated maps, and any possible army of 20 units, these constraints become impossible to declare.

You're describing the "standard" military strategy of deploying in the center and on the flanks, but there are many more of course. And what about 3-way battles? What you need, and what CA have tried to program (and the basis IS there), is a AI from the ground up, ie:

1. unit level AI responds to threats nearby. This is the most important part of the AI and needs to function before you can program the rest. That means: Archers must skirmish properly (they don't always), Hoplites must select their target properly (they don't), etc.

Unfortunately the development time for Rome TW was gone and they didn't get to finish this part, so that they could continue with:

2. army level AI, consisting of group level AI (if groups are defined or recognized by the AI, which is very hard to program) and if not, AI that all units must respond to.

This last part AI is in fact programmed, in battles where the AI has already lost half of its army, it gets the withdraw option which it uses. (the enemy army is routing... enemy reinforcements chose withdraw)

My criticism in all of this is that CA aimed too high when initially designing this AI, in particular with the second part of the AI, the army level AI. Recognizing situations on the battlefield and knowing what to do is typically something only humans can do, while I would give the AI a static set of strategies and formations to choose from, and simply make a somewhat educated pick.

This ensures coherence on an army level, gives the AI a strategy, but doesn't ensure all units will attack at once (if you outmaneuver it), which is also essential. This is why i understand they have left the RomeTW AI as is (without the army level AI), knowing full well that players will already be challenged by unit level AI, ie a good skirmishing AI with lots of horse archers is difficult to beat by any new player.

MTW veterans however, find it simply annoying, and expect to be challenged by army level AI, which was present in MTW and STW (though very static and constrained, most vets didn't and still don't notice).

[Fwapper]

I think people are really giving CA a hard time over the AI. A lot of what people say is just rumors which have overgrown.

It is very easy for CA to show off their cool graphics and new animations and so on, but much harder to demonstrate the AI. We'll have to wait for the demo or actual release before we can see it for ourselves and know for sure.

Of course the marketing will be largely driven by the graphics - it is the most obviously improved and visible part of the game. They're not going to charge some cavalry into a city and say:
"Oooh look! They went where we told them to go!! They didn't run around like headless chickens!!"

What I'm trying to say is don't bach CA too much over the AI as we haven't actually seen it yet.

[Lucjan]

What I'm trying to say is don't bach CA too much over the AI as we haven't actually seen it yet.

Oooh, another voice of reason attempt. hhehe, we've tried, it's just so much more fun being all "GAHH, CA WE WANT BETTER AI" *charges CA headquarters with angry viking horde like capital one commercial*

[The Wizard]

...I'm going to start referring to the people in this thread as the "gloom-n-doomers"

The end of the world is nigh!...

Regardless, we love this game dearly. We want to see its full potential realized. What makes us sound bitter is our disbelief that we are in any way listened to, which is quite frustrating as far as I'm concerned.

It's quite bad style, too, not listening to your hard core. This -- in a world where there are developing teams that develop patches with professional gamers (Blizzard) or actively helps and supports the modding community, sometimes even offering jobs to top mod makers (Bioware)?

Sad.

The hard core is your friend. Listen to the hard core. They are your access to real, active player experience.

[Nemeo]

Unlike many people here, I think that making a good AI for a game is far from impossible. On the battlefield we don't ask it to simulate feelings or anything arbitrary. The game has many restrictions and a great set of rules, like "units can't fly", "units can't fight underground", "units can't go through other units and walls", "cavalry beat infantry", "pikemen beat cavalry"...etc. From then on its easy for a coder who knows his game to have it check things like "I'm a cavalry, there's a phalanx unit in front of me, I must not charge" and then have it attack another unit or flank the phalax unit. The AI can check every units and have them moving at the same time in the same second, it should be able to beat a casual player from time to time. The AI IS the game, it should know better than the player how it works, what would not play in its favor. If coders don't tell it to analyse the situation it faces and just put a routine to have front unit making a straightforward charge and side units attacking from the side (they don't really flank in RTW as they often charge before the rest of the army, thus their target is rarely fixed before they charge), then this is no more a strategy game. It's like being a cat toying a mouse.

There can't really be a unit over another in TW. So it's almost like a chess game. There are good chess game around there, with very potent AI. I've never beaten one of them in the normal difficulty. So no one will ever persuade me that it's impossible to have a decent AI in any game, as we don't ask computers to simulate artificial life.

[Tellos Athenaios]

With designing AI there are always a few problems:

Who are you designing it for?
How much does it take for a slow system to make use of it? People with slow systems should be able to play your game by simply tuning the graphics down - and not the AI.
(Ever tried to create a snake game on a TI84plus with some AI to make the food move in order to escape from you? If yes, you know what I mean, even a relatively simple AI (with the above mentioned system and program this means: advoid obstacles, and if your route is obstructed try and flee the second best way away from the snake) slows down a lot.)
Do you really want it to be ALL seeing, and ALL knowing (this isn't realistic, since on a true battlefield most commanders wouldn't have been able to see ALL of it...)? If yes, how do you get it to respond in a sensible way?
How long do you have to develop it? I mean, an ALL seeing AI would require a self learning program - that's what they are into at universities and this has lead to 'speaking' systems. For companies however it would be nice if a new version of a series didn't take several years to complete...
How, when, and where should AI change it's battle formations? You can't just change your battle formations at any time, you can't do that at any place, and the key to do it succesfully is to do it in the right order, and in the best way adapted to the new given situation. To humans with a little sense of tactics this all comes natural but how do you copy the human way to the AI?
And to refer to my prior point of slowing down your system: to make effective use of battle formations and if needed to change those, the AI has to constantly check it's troop positions, movement directions, speeds, enemy troops and so on. And not only checking those, it has to adapt to new situations too, wich often requires entire new formations and always requires a new forecast. And so, a real killer AI in battle would also be one that takes processor minimal requirements to a whole new level.
And probably loads of other problems, to make it even worse. Such as the game is based on an earlier game wich had it's specific faults and bugs, and since this is some way just a new version of older software many of these faults and bugs can't be edited out without having to completely rewrite either very large portions or the complete game. Rewriting would then lead to a new game that simply doesn't fit as successor in the TW series nor as a start of a whole new series. (Often rewriting such large amounts of code requires rewriting other code not concerned with AI too - to make sure the comletely changed AI is going to work you might need a new 'enviroment' as well.)

Now why don't we see improved AI (yet)?

This is, in my opinion, simply a matter of REALLY good AI can't be seen (as it would seem as if as human was in charge and not the AI), improved AI is hard to notice (since this is mostly noticed in detail or by trying it - but there is no demo yet so this way in cancelled out for now) and to show AI you need to show a lot of your game. And all we have seen for now is just to build expectations not to give us answers on these sorts of questions: a company probably doesn't like when it's new products lose all of their 'surprise' values. So we don't get to see everything that's in the new game - which I feel confident about will include an improved AI. I'm very positive about this, since I doubt CA wouldn't make use from some of the excellent work Modders have done on AI battle formations like DarthVader's formations mod.

[sunsmountain]

Do you really want it to be ALL seeing, and ALL knowing (this isn't realistic, since on a true battlefield most commanders wouldn't have been able to see ALL of it...)? If yes, how do you get it to respond in a sensible way?

On Very Hard, I would LOVE the AI to cheat and know the exact make up of my army before we even deploy, and have its battle line already in an ideal counter position at the start. Responding in a sensible way needs 2 parts: A good unit level AI, which will not do stupid things while carrying out the second part which is army level AI, trying to get the best match-ups, best flanks, best path of attack etc.

They tried to finish the first part and didn't succeed enough. Basics were just not right. Things that could have been prevented by a better program, but were not. They certainly will have to be with MTW2, or a lot of RTW/MTW/STW owners will not be buying it. At least I won't. I will not play a game that screws up my generals, phalanxes, or archers marching back and forth playing their bagpipes and licking each others ...

... the second part will be hard enough as it is.

And to refer to my prior point of slowing down your system: to make effective use of battle formations and if needed to change those, the AI has to constantly check it's troop positions, movement directions, speeds, enemy troops and so on.

Ahum. Right now they probably check no more than enemy troop position, and troop type. Movement direction is a variable that adds too many dimensions into the problem, speed is redundant and inferior to troop position, as the AI has to evaluate its decision every so many seconds anyway.

I'm very positive about this, since I doubt CA wouldn't make use from some of the excellent work Modders have done on AI battle formations like DarthVader's formations mod.

Fair enough. CA have already implemented a measure of army level AI in patch version 1.5/1.6, where starting army formations were implemented, and if you look closely, the AI does try to maintain them while engaging. Of course, when it comes close to your troops....

[caravel]

"units can't go through other units and walls", "cavalry beat infantry", "pikemen beat cavalry"

This is how MTW and STW AI, for all it's flaws, worked. Cavalry did not hurl themselves at a formation of Chivalric Seargents, it tried to flank the more vulnerable units. It would also try to adjust it's formation to ensure that it could have a go at countering/beating the player's. For example if If your cavalry were on the left wing, it would move it's spear units to engage/obstruct them. Simple stuff. STW was in some ways better,as the AI seemed far more intelligent in battle. If it couldn't win it would withdraw immediately after inspecting your troops, it was more effective at flanking and far more cautious, often deploying much better defensively and not being easily drawn out. MTW AI was better at bridge defending.

Also, and very importantly, the AI, and the player, had the time to deploy correctly and change formation using some kind of tactics. This is unlike the RTW cavalry that all appear to buzzing about on motorcycles, not horses, changing direction like flocks of birds. This is just a few of the reasons I simply cannot get into RTW, and despair at the fate of M2TW.

The campaign map is a different issue of course. There is a sort of AI on the campaign map, well a 'diplomatic engine' for lack of a better description, and this has never been that hot, but the RTW map uses a different system that was totally new at the time, unlike the MTW risk map which had previously been in STW. Personally I didn't like it, and it's RTS/Civ style look. I prefferred the risk style general's map to manage my campaigns. MTW and especially STW had a sort of 'sim' feel about them. You didn't feel as if you were playing a game at all. RTW felt and looked like a game. In the ending video of STW, there is a highly polished fmv sequence of a risk style campaign map with pieces being pushed towards a castle to assault. This is the sort of map for M2TW (and RTW) that I would like to have seen. (I will post an image later)

I agree 100% with Puzz3D also and most of the other doom 'n gloomers for that matter.