So...the customers wont be cheated this time?
 

I am very happy with Dominions 3 really...but I feel myself cheated with Dominions 2. I remember when I bought the game, Shrapnel's site listed the features for the game, and there was like something "very good AI for solo play" or something like that. When I bought the game I realized that the AI was very bad, and I was very mad about it, I felt myself cheated. If I would know, that the AI is so bad, I wouldnt buy Dominions 2, hence I prefer singleplayer. I still dont understand why Shrapnel listed Dominions 2 with a very good AI. That was like lying.

Now Dominions 3.

"An already challenging AI made even more challenging. If you can't handle the unrelenting computer player, you can always try your hand at playing against humans with Dominions 3 robust multiplayer suite."

Already challenging AI? http://forum.shrapnelgames.com/image...s/confused.gif Pardon me but who found the Dominions 2. AI challening? The Dominions 2. AI was very weak.

Now I decided something. I wont buy Dominions 3. until I tried out the demo and the "new" AI. If its really challenging I will buy the game, if it is not, I wont.

Btw..

"Brand new graphics, with an overhauled and streamlined interface. Transparent mechanics decreases
the learning curve."

I hardly see any brand new graphics, the interface is somewhat similar to Doms 2, the battle map is a bit better looking, the unit's graphics is the same, the spell effects are slightly improved...so not a big deal graphics wise at all, but I don't really care about the graphics..all I want is a challenging singleplayer.

I can't stop playing Galciv2, the AI is so good and challenging.

Re: So...the customers wont be cheated this time?
 

Maybe it has something to do with posts like this or this.

Re: So...the customers wont be cheated this time?
 

Oh come on! You cant say that the Dom2 AI was really bad as such efforts go. It was AT LEAST good for tutorial level. For teaching a newbie the game and getting them convinced that they knew it well enough to challenge humans. You might feel that it was never a challenge for someone who had learned the game. But thats true of most AI's in games. Remember that "artificially intelligent" is not the same as "artificially human". If its able to play the game then its a good AI.

Now you know that I was the loud and strong advocate of solo play in Dom2. My fansite Dom2Minions was the unusual one in that it almost entirely involved things designed to make the game challenging later on for the solo player. I have had conversations with Illwinter on aspects of AI, and Random Maps, and Events, and other things that werent entirely MP things. I think that its safe to say that there is enough difference to make it fun for solo-play past the tutorial level. Probably not to expert level but we will have to see how good it got. Im definetly getting it so I will let you know early on my impressions on solo play.

Re: So...the customers wont be cheated this time?
 

There's not much else to Dominions, graphics-wise, that those things I emphasized from your post. They have been improved, except for the units' graphics. How that isn't brand new? Or were you excepting so radical a change that you wouldn't know this was Dominions game?

Three out of the four seems pretty good, especially as the last one would be lots of work. Dom:PPP alone had 600, or was it 800?, units. It would be nice to get them redrawn, but that would be immense amount of work.

On DomII forum, in the discussion about GalCiv2, some people thought the AI's weaknesses were too big, and it was too easy to exploit it. I'm not sure how good it actually is, as I haven't played the game, but at least that proves that people's opinions on AI can vary.

Re: So...the customers wont be cheated this time?
 

600, 800 or 1500 units, it wont make a huge difference. Basically most of the units are cannon fodders, or we aint using them at all strategically.
The galciv2 AI is much, MUCH better then the Dominions 2 AI generally. There are weaknesses in it of course, but those are mostly player exploits. I am pretty sure that the Galciv2 team knows about those player exploits and those will be fixed, maybe in patch 1.1, hence they are saying that the AI will be upgraded again, probably they gonna add AI counters for those "exploit tactics", like the hit with fast ships/planet bombing/run tactic.
Like I said the galciv 2 AI generally much better, especially if you fight against 3+ AIs on the hardest settings, it is almost impossible to win, I am usually ending up on a 2+ front war, and you cant win that in Galciv2. In Dominions 2, I could hold 3+ AIs alone without much probs on the hardest settings.

So all in all, lets wait for the doms 3 demo, and we can decide to buy the game or not. I am pretty sure that many players would buy this game for singleplayer, may I ask that how many players who bought Dominions 2 were playing multi games? I was checking out this side, and I only seen a "few" ongoing games most of the times/AARs about mp games.

I just hope that the devs can fix the Doms 2 AI errors, I am very very happy that they gonna do something with the AI castle building thing at least, that is a good start indeed.

Re: So...the customers wont be cheated this time?
 

Brad has been mentioned in AI professional documents for as long as AI has had any. I was reading about him years before he finally decided that Windows had improved enough to run a trimmed down version of his GalCiv game. Most of the complaints about GalCiv are more from the fact that its AI with a game attached later. Most games tend to be a game with AI attached later. Of course now people are also complaining that adding mltiplayer should be a simple no-brainer (hahahahha WRONG!).

My perfect game would involve bits of many games but I doubt anyone would be able to put it all together into one game. I suspect that Dominions will always be a turn-based send-file hosting multi-player strategy game. That pretty much defines what it will and wont do easily.

But I wouldnt mind a solo-player game with the rock-paper-scissors balance of Dominions and its effort to support so many diffrent tactical styles, the ai of GalCiv, the huge randomized setups of Magic Realm, the player-side moddability of SEIV, the storyline depth of King of Dragons Pass, and run on Linux & Windows.

Re: So...the customers wont be cheated this time?
 

Of course, I'd like for national units being made more feabile to use. Most of us probably want to. But even if only two or three of every nation's units were ever used, the quality of those troops differs between nations. Cannon fodder has many layers of power. Have you ever seen a Wyrm or a Great Mother mowed down by Knights, or by Glaive-wielding Tien Chi medium infantry backed by archers?

Provincial defense is the only true cannon fodder in the game, because it can be lost many times for the same cost. However, it is also an efficient defense for some nations, at least against minor stealth attacks. And this isn't only Jotunheim, human nations like Ulm or Marignon aren't bad either!

Re: So...the customers wont be cheated this time?
 

Like I said, I barely seen many MP games on the MP/AAR section for this game. Sure there are a couple of PBEM games going usually, but its like nothing. I mean probably much much more people bought Doms 2, who only played the singleplayer part. I dont think that a turb based game should force a player to play it PBEM. Personally I hate PBEM games, they are way to slow. I also know that there cant be other option for MP for this game, hence 1 turn takes enormous amount of time sometime. Thats all good. However Gandalf... if your opinion would be valid, the game's features are fake. It should be something like... "PBEM multiplayer game with a singleplayer learning segment", but surely not "challenging AI" for singleplayer. I meant Dominions 2. We shall see how it turns out for Dominions 3.

Btw may I ask the developers, that why the AI problems were not fixed ever in this game? Dominions 2 is an old game, but we havent seen a single major fix for the AI, like a castle building fix for the AI etc. etc.
I found it very odd, but after a few tries I gave up posting my ideas about the AI on these forums, hence the devs couldnt be bothered by the AI in Dominions 2. I had a feeling like the devs are like this. "here you got this AI, live with it."

Re: So...the customers wont be cheated this time?
 

> "here you got this AI, live with it."

Yep http://forum.shrapnelgames.com/images/smilies/happy.gif

Re: So...the customers wont be cheated this time?
 

!! You cant be serious Kristofer. I hope that you are just kidding, I mean what is the point in NOT fixing an AI? http://forum.shrapnelgames.com/image...s/confused.gif http://forum.shrapnelgames.com/images/smilies/shock.gif

Re: So...the customers wont be cheated this time?
 

The fact that they did continual fixings on the MP part of an MP game? And as far as I know, most people who have dominions are playing mp games even if its just friends nad family with no need for public scoreboards. You sem to feel you are the voice of the silent majority but I dont know if anyone is agreing with that.

I thought that the AI thing in dom2 was answered long ago. Its linear so there isnt much fix that can be done without a massive rewrite. And I asked continually for players to develop a better linear layout for the ai that didnt amount to teaching it how to play just one nation better.

MY request was for a map command that would allow map scenarios to add castles that were freebie giveaways. A #patrol or a #defence 21 type of command. Im still hoping for that

Re: So...the customers wont be cheated this time?
 

And well, a two man team must focus on certain things. I'm not sure if the expansion of Illwinter is what the two want, or if it is economicaly feasible.

Re: So...the customers wont be cheated this time?
 

Actually, it's listed as four now. Johan K. and Kristoffer O., with help from Johan O. and Daniel Serra. Surprised me, too.

Re: So...the customers wont be cheated this time?
 

Well the feature list was fake then. It clearly stated "very good AI". Not to mention that they forgot to mention "We wont care about the singleplayer part after the release."
Just tell me Gandalf that I am not correct.
Anyways this is the past, the future can be better, lets hope...Dominions 3 can be different in support.

Re: So...the customers wont be cheated this time?
 

You are of course perfectly correct. No one can tell you that you are not correct. Since you are stating an opinion, about an opinion, then it must be true for you.

Re: So...the customers wont be cheated this time?
 

You admitted [or someone else?] that the singleplayer was only good to "taste" the game and play multi. It is clear that the AI was weak, I think everyone can admit it.

I mean we mainly seen some weak AI troops, running all over the map without a point really, not to mention the other AI weaknesses. We cannot call this "very good AI", what was stated on the feature list originally. Now we got this "even more challenging AI" description for Doms 3, after the Doms 2. AI failed.
I think we can say more after a demo like I said, the question is....Illwinter will forget developing/fixing the singleplayer part if it is needed or not. This would be good to know also.

Re: So...the customers wont be cheated this time?
 

Maybe this 'even more challenging AI' can build castles now so you should see much more national troops running around (without goal???).
I think you are greatly overstating your case here. AI is not nearly as bad as you claim.

Re: So...the customers wont be cheated this time?
 

You still seem determined to put your "fact" words on other peoples opinions. I think that any game with an AI that can get you up to human-player level is an above average AI.

Did you ever find out who's opinion it was that you were quoting in the first place about the AI?

Re: So...the customers wont be cheated this time?
 

I remember reading somewhere that the AI was not updated because it was not fun to work on. Which is somewhat understandable.

The community has asked for tools to be able to script the AI, or to get specs for the .trn and .2h files (so you can write a program to read a .trn file and build a .2h file). Either one of these would be much appreciated.

Re: So...the customers wont be cheated this time?
 

We did get some additions to the maps in order to build more challenging scenarios for play vs the AI. Having the AI ally together agains the player helped alot. Also beefing up their armies with extra troops and equipment. The modders took things even farther.

I think that when players actually laid out specific AI formulas that some of those might have made it in also

Re: So...the customers wont be cheated this time?
 

The AI isnt challegening http://forum.shrapnelgames.com/images/smilies/frown.gif
I feel so pitiful.... I have yet to be 16 impossible AIs at once.

Re: So...the customers wont be cheated this time?
 

lol I must be kinda pathetic I guess. http://forum.shrapnelgames.com/images/smilies/happy.gif While I have won against AIs on small maps with few opponents I have not yet won a game against many (14+) opponents on a big map. The sad thing is that I find this AI to be much harder to beat than the GalCiv2 AI.

Re: So...the customers wont be cheated this time?
 

When you do get there, let me know and I will show you how to ally the AI's with each other so they will gang up on you.

Check out the WE vs THEM scenario.
http://www.dom2minions.com/WEvsTHEM.shtml

Or the Random Opponents program which also allies the AIs
http://www.dom2minions.com/Opponents.shtml

The "Poke in the Eye" scenario is another way to boost the difficulty of solo playing
http://www.dom2minions.com/PokeEye.shtml

Re: So...the customers wont be cheated this time?
 

I didn't think there was much wrong with Dominions 2. If Illwinter only fixed the remaining bugs and didn't release a Dom3, I'd be happy. As far as the AI is concerned, Dom was designed as a MP game first and foremost, so it's no wonder the AI did not really evolve. I'm mainly a SP gamer, so any improvement in the AI is appreciated, and I'm not going to look a gift horse in the mouth. It's unfair to compare Dom AI with GalCiv AI, since GalCiv was designed around the SP AI in the first place. The 2 games have different design goals. Also, Brad is one of the best AI programmers in the business. If I were rich, I'd personally pay Brad to develop an AI for Dom, but I'm only dreaming now.

Re: So...the customers wont be cheated this time?
 

The combat model in Doms is in a different league than GalCiv.

Brad has done a very good job with strategic AI in GC2 - last night, for example, all the AIs realized I had the smallest military and started taking me out. My tiny navy clustered around a fully-upgraded Military Starbase, and were mostly unassailable. So, eventually, the AI started ignoring my planets and fleets to go straight for my military starbase, and they almost killed it... and whenever my attention slipped for a second - like, when I intercepted their fleets so they couldn't attack my starbase - they'd sneak a fast troop transport in and steal a planet. All in all, I was very impressed.

However, you can't even evaluate strength and weakness in Doms2. An Ice Devil with a Hell Sword can kill 2000 militia, and die to 500 longdead. The militia can kill the longdead, but die to a mage casting Foul Vapors. The mage would die to the longdead, who would be killed by 20 priests, who would be killed by the militia, who would be killed by 500 Salamanders, who would be killed by 25 Devils, who would (maybe?) be killed by 25 Angels, and etc.

Writing the AI for Dominions is not easy, or even difficult; it's mind-blowing. Rule-based AI just can't work perfectly because there are too many rules.

I would compare Dominions to English (or some other highly hybridized human language) and Chess / Go / GalCiv to computer languages. Sure, GC2 is nondeterministic and more complex than the other two, but they all have fairly simple rulesets without many exceptions. As long as something is defined by rules, it can be taught to a computer - that's why compilers can be auto-generated. But anything that is largely defined usage, exceptions, or context becomes exponentially (with size) more difficult to teach to a computer. If Illwinter had stopped with just Pythium and Arco (without any mages), then it would be possible to write a very strong AI. But with spells, damage types, gems, items, immunities, unit attributes, and so forth, a complete and accurate ruleset probably would be larger than the universe.


I think that Illwinter has done a very good job on an AI that copes with this complexity, and can still beat intermediate players... and advanced players too, unless they follow very specific sets of steps, that would also allow them to beat virtually any human player who doesn't know the recipe for countering them. Perhaps the AI should be given some "recipes" that it seeks - like massing Druids with ivy crowns and summoning vine-creatures, taking a 9/9 Earth/Water pretender and rushing to Clay Men, starting a Tartarian Factory + Gift of Health, or rushing Staff of Storms + Wrathful Skies. It might be interesting. But no matter what recipes are given to an AI, nothing you see in this decade will ever detect the use of an advanced recipe, much less correctly choose and employ a counter-recipe. Consider how much work Brad put into making his AI detect that an opponent is primarily using Beam Weapons, and eventually compensate by developing Shields? That's in a game with 3 different damage types (and no immunities), or something like 9 basic units... and the AI is still slow to react, when it does at all, because it has to also consider every other player in the game, and other possible uses of research points. Now consider the scenario with 1100 units and 500 spells, mutated by various combinations of Dominions and blessings and items. Still seem easy to write an AI that counters some unseen strategy that a player cooks up?


In summary, I think Dominions II had a decent AI with a few important flaws, like castle-building, TAC-AI spell selection, ignorance of supply constraints, poor site-searching, sending pretenders to the deathmatch, choosing random scales and paths on pretenders, and the inability to use any powerful recipes (like putting good items on an Ice Devil, rather than random items on a scout). If those sorts of things are addressed by Dominions III, it will be much stronger in singleplayer... and it won't be obvious how to make it better. But even with flaws, it still beats most people over and over and over for months, or until they come to the forum for advice... or even then. http://forum.shrapnelgames.com/images/smilies/happy.gif

Re: So...the customers wont be cheated this time?
 

This is a good analysis, and from what I've learnt from my *humble* experience of programming AIs I completely agree.

I hope that at least some the obvious AI flaws that you mentioned will be fixed in DOM-III.

Re: So...the customers wont be cheated this time?
 

I must agree with my brother, however I wasnt really playing singleplayer after a period, because of the unintelligent AI, I used to play multi most of the time with friends 2-3 weeks after Doms2 release. If I remember correctly, the game was sold as a sp/mp game with a nice sp ai. Even if the AI failed as 'challenging', its ok for a release, just consider that AI bugged or whatever. There kicks in the developer's support. If I understand correctly, the devs werent supporting the singleplayer part of the game (I stopped following the singleplayer - AI discussion after a period, since the first patches contained no AI fixes), and I think that it is sad, since it was (the singleplayer) a part of the game. This is like a half support and its never good.

Now as Saber Cherry said:

-> if these AI errors will be fixed for Dominions 3, the AI will be really challenging.

As for the Galciv2 AI, I dont agree with Marcus. The Galciv 2. AI is weak also, but just because there are some ways to beat the AI easily, and the AI cant do anything about it. Those are major AI flaws, like the Doms 2. major AI flaws.
Well if I am not using any "cheap" tactics in Galciv2, the AI is much harder to beat of course. The "general" AI is better in Galciv2 for sure, since it doesnt have major flaws like the Doms 2 AI. In Doms 2 the AI rarely built castles/recruited/summoned strong troops for example. In Galciv2, the AI builds kickass ships/starbases,colonizing hella fast etc. The only problem with the Galciv2 AI is that its "allowing" players to use tactics what the AI cant beat. So overall yes, the Galciv2 AI is surely better, but the Doms 2 AI problems can be fixed for Doms 3, and then, the Doms 3 AI will be >> Galciv2 AI imo.

Re: So...the customers wont be cheated this time?
 

Saber, You are quite right about the headaches of programming AI in Dominions. That was shown in the forum when we asked for flowcharts and ai scripts which would fix things. Some tried and quickly shifted from whiners about the ai to just low grumbles.

Eventually (maybe Dom4) Id like to see dominions broken up. The maintaining would be a headache so it ight have to wait for multiple programmers. But Id like to see the host seperated from the tools and the client. DomHost.exe, DomTools.exe, and DomClient.exe, then probably a DomMain.exe to tie them together so that its transparent to the average user.

Then Id like to see changes made which would heavily support 3rd party scripting. Im not much interested in Dom3 for multiplayer but Id be VERY interested if it got to the point that really involved scripts that made the host take hours to do a turn could be generated. That would turn the AI into something exciting. Someone who plays Arcos really well could script an Arcos-bot, and someone who plays Ctis well could do the same for Ctis. Eventually we would even end up with variants such as an Ulm which plays as a massive army and one which plays guerrila warfare. Setting up grames which randomly selected various styles and the server spent most of the time between turns actually thinking.. *sigh*.

Anyway, If I won the lottery that would be on my list. http://forum.shrapnelgames.com/images/smilies/happy.gif

Re: So...the customers wont be cheated this time?
 

I think it could help the AI a bit, but I have no idea how it would be in practice, and how hard/unappealing it would be to make. Perhaps there could be a set of "guideline" pretender setups for AI so it won't pick completely strange scales or magic.

Re: So...the customers wont be cheated this time?
 

Can't I http://forum.shrapnelgames.com/images/smilies/happy.gif
There are exactly seven points in NOT fixing the AI.
Three of them are PR related.
Three are secret.
Last tone has to do with strange cults and the orbital mind control lasers of the swedish 'Svärmöga', who sees it all.

Seriously though: there is no *point* in not fixing stuff. Fixing is good. That is why people do it from time to time.

Re: So...the customers wont be cheated this time?
 

Daniel is mostly in as the company form 'ekonomisk förening' requires at least 3 members. He does some economic stuff and is the main contact with Erik and Anna of Falsobordone and Dråm.

My brother comes up with ideas and speaks some in the fora, but he doesn't work on the game per se. He is also the devils advocate http://forum.shrapnelgames.com/images/smilies/evil.gif (when JK is silent).

Re: So...the customers wont be cheated this time?
 

Both of these would be doable if the format for .trn and .2h files were released.

I think the main concern is cheating, but you already can cheat even without the information.

I know others have asked for this, although I never saw an offical response. Perhaps there is some problem that I'm not aware of. The community could make some interesting AIs with this info.

Re: So...the customers wont be cheated this time?
 

How many games did it take you before you could guarantee a win against 16 AI opponents on a properly sized map? If it's more than four or five games, then you did get a good AI, but you personally are one of those people who have no concept of the difficulty of programming an AI.

So you've never played any mods singleplayer?

Re: So...the customers wont be cheated this time?
 

Graeme, I dont know how hard is to program the AI, you are right...but since Illwinter is talented, we can clearly see this from the Dominions games, I am pretty sure that they could fix it. Someone here posted that it wasnt fixed because scripting the AI was boring http://forum.shrapnelgames.com/image...s/confused.gif!

I think after a company is releasing a game, they must support it, interact with the players that what is wrong, what to fix etc. This is how its working for all games.

..and yeah I never played any mods for Dominions 2..there were any mods to fix the AI errors? I really missed those if there are any.

Re: So...the customers wont be cheated this time?
 

I asked you how many games it took you before you can guarantee that you will win against 16 other AI players on a ~160 province map. If you can't guarantee that win, then the AI is clearly still capable of beating you. As for your ignorance about the difficulties of AI programming, it doesn't just take talent to build an AI capable of defeating humans. It took decades of research effort to build an AI that could defeat humans at Chess, a completely deterministic game with an 8 X 8 movement grid, and 6 individual unit types. Dominions is only has a few dozen orders of magnitude more choices to make.

Are you claiming that Illwinter hasn't supported their game? Do I need to go through the patch list to point out where you are mistaken?

So you are complaining that Illwinter didn't support single player games, while at the same time you never even _used_ the things they did add for single player games?

Re: So...the customers wont be cheated this time?
 

Some people like scripting AI, some dont.
You could at LEAST take a stab at it. Here, I will get you started...

(turn 1)
send out scout (if you have one)
put pretender on research, set research to Evocation
put leader with troops on patrol
turn taxes to 200 percent
make bid on mercs cost+10
recruit priest (in case random event burns temple)
if cant afford priest then recruit leader
if cant afford leader then recruit scout
spend rest of cheapest troops (more patrollers)

(turn 2)
assign troops
set infantry to the front
set archers to the sides and slightly back
set cavalry to the far sides

give gems to mage equal to magic type and level*3

select weakest nearby province and attack
set taxes back to 100 since everyone is leaving but pretender
bid on mercs cost+20
use leftover cash to purchase cavalry
if cavalry at 10% of troops then purchase hvy infantry
if hvy infantry 50% of troops then recruit archers
if archers 20% of troops then lt infantry

(turn 3)
in new province:
purchase defence of population divied by 1000
recruit leader
if cant afford leader then recruit scout
recruit 10 archers or slingers


Doesnt that sound good? You want to keep it going?

Re: So...the customers wont be cheated this time?
 

ioticu's post is dead on concerning the AI.
Illwinter made it clear that they were developing a multi-player game. During the testing of DOM2, a lot of it was done in multi-player mode. If something 'feasible' was found that could affect the AI, it was implemented.

Re: So...the customers wont be cheated this time?
 

It would be great if DOMINIONS_3 had the option for gamers to adjust/improve the AI. With time we'd see some interesting AI's appear.

Also it would be great if the AI has multiple personalities instead of one fixed AI used by all computer opponents. HOPEFULLY AI opponents are randomly given an AI personality at the start of each game which would make it MUCH more difficult for gamers to predict what the AI will be doing in the game. Also the option to disable one of the personalities in case it's weak or bugged.

Examples:

The_Sage: Usually peaceful and more concerned with mastering spells than conquering the land. Beware he/she hungers for territories with strong magic sites. He/She's also known to attack anyone with greater research then him.

The_Cutthroat: Untrustworthy and usually organizes his assassins to attack the same time his troops march into provinces. He will make temporary alliances to knock down the greatest threat(most troops & highest gold/gems). He's known to attack territories which bring high gold income.

The_WarMonger: Purely focused on taking territories to expand his kingdom. Weak enemies nearby will be his primary target unless someone owns more territories which he truly hates. Sometimes known to use assassins.

The_Paladin: His expansion is based off those which do the most evil which includes using death magic, blood magic, death scale, and most importantly who is killing the most warm blooded creatures. His actions are usually honorable... in fact always honorable in his eyes. Known to ally with a weak nation to focus on a common stronger enemy.

The_Psycho: The actions and behavior are completely random and unpredictable. Sometimes he will attack the strongest and sometimes the weakest. It's known those with the highest dominion are more frequently attacked.

The_ArchMage: No known flaws or weaknesses and is an aggressive expansionist. A hunger for magic sites has the archmage searching more intense than the others. Also very effective for using spells before marching into territories. The player with the most gem income are more likely to become an enemy.

The_General: Able to focus on specific bottlenecks between provinces as he expands. Sometimes known to proceed in setting up supercombantants(SCs) to march and take territories as a team and occasionally send stronger ones solo. He will form allies with anyone willing to attack the player with the total largest army size.


==========

Just my ideas. http://forum.shrapnelgames.com/images/smilies/wink.gif
Maybe a Dominions_3 expansion can provide this suggestion??

Re: So...the customers wont be cheated this time?
 

Cheating, as in "alter the world state", can be addressed by using a signature scheme on the .trn file. If the .trn file doesn't match its signature, then it means someone tampered with it.

Cheating, as in "peek at what other players can see", can be addressed by encrypting each nation's section of the .trn file (assuming there is one) with the nation's password. Leave an unpassworded section unencrypted.

But then, this would proabably require a complete overhaul of the game's turn files. It would be nice, though, to have the turn and order file formats open, so players could use their own external programs to help in their micromanagement (generate partial .2h files, like, or generate "report" files that tell them, say, where everything is and so on). Most likely, though, this will not happen until Dom4 or Dom5... http://forum.shrapnelgames.com/images/smilies/happy.gif

Re: So...the customers wont be cheated this time?
 

Hey - I put quite some work into my suggested algorithm(s) to fix the castle and troop building.

Noone ever tested them, you included.
And I can't do a dozend of test games all on my own for sure!

And IMHO MStavros is mostly correct. It's a shame that the most obvious, AI-crippling bugs are not fixed in Dom2: That it can't build castles, but only get them from sites/events, and does not recruit indie mages.

If you put a bunch of admin-30 castles on a map, it will go for them quickly, though, and start building 'serious' armies. It will not start recruiting indie mages, unfortunately.

And I can't really understand why this is not fixed - other than because the devs/shrapnel think we better should buy Dom3 ... .

Alas, this is not the kind of support I want for games I buy... .

Re: So...the customers wont be cheated this time?
 

The AI sure as hell will recruit indie mages if there are laboratories present. It just has the same problem with labs as it does with castles: it doesn't build them.

Edi

Re: So...the customers wont be cheated this time?
 

16 AIs on 160+ province maps. I usually played smaller games, and some bigger games too. I NEVER ever had a hard time to beat the AIs. Hm, as for time, I think after 2-3 weeks of playing I was beating the AI in all games, except in a very few where I had some horrible luck or 4+ AIs were constantly attacking me.
The thing is that if a player can beat 3 AI players 1v3!!!!, what can you say about that AI? Just go and try to beat 3 AIs in Galciv2 on harder settings, you will loose very fast. Yes I know that Galciv2 is a singleplayer game. My _ONLY_ complaint was, that the Illwinter team _NEVER_ fixed the major AI problems at all, like the castle building and the correct use of troops/summonings.
This is why I said that this is a bad support. They supported the MP part, thats pretty good, but they released a SP/MP games, I bought it because I've seen "VERY GOOD AI" on the feature list. Do you understand what I am trying to say now?

Please, drop this idea "how hard is to script AI" topic, if a company cannot do it properly, they shouldnt add a singleplayer part then.
Once again, take Galciv2 as an example. Try the AI on the easiest and on the hardest setting. IF you aint exploiting the AI, and ur using normal tactics, you will barely win against the hardest AIs. It is possible to make a good AI, especially if the company knows about the exploits/ai bugs, and they are fixing it. The Galciv2 team is fixing most of those exploits for patch 1.1 AFAIK.
If you want to say that its a SP game, and that the developer team is much bigger then Illwinter, all I gotta say is that the product prices are the same, thus we deserve support for our money. If I buy the game to play singleplayer, I expect fixes in patches for singleplayer if it is needed.

I tried out the latest Dominions 2 patch btw, and the AI is the same, so I dont know that what singleplayer support are you talking about Graeme. Crapload of months passed, and I havent seen a single fix for the AI in Dominions 2. So yeah, this is why I am gonna wait the Dominions 3 demo, and see how it turns out, I am just very cautious with Illwinter now.

Re: So...the customers wont be cheated this time?
 

You got several weeks out of the single player game. That's all any person can realistically expect from an AI even with current state of the art AIs. I still assume that you've never played a game against 16 AIs all allied against you, and I also assume that you deliberately use cheesy tactics to defeat the AI.

The Gal Civ 2 AI cheats far, far, more than the Dominions 2 AI on the higher difficulty settings. That's the only way that it is possible for it to have a chance of defeating a human.

Those "problems", as you call them, are only noticeable to somebody who has already risen far above the capabilities of the AI.

You _got_ a very good AI. It's not my fault that you're whinging about it not being able to defeat a human opponent. There are virtually no AIs that can accomplish that for any games that incorporate even slightly random elements. You'd notice, if you weren't ignorant of the field of AI research, that computers are just barely beginning to be able to play Poker competently against humans. That's another game that's orders of magnitude simpler than Dominions, especially when a computer is capable of determining the exact probability that any given hand will win in a atter of microseconds.

I'm not going to drop an argument when my opponent is clearly ignorant of the difficulty of the tasks that they are demanding. I assume that you must just be barely out of high school from the immaturity and general lack of information displayed by your arguments.

Of course not, because you don't have the AI's perfect understanding of the rules, and the AI starts with massive information and production advantages over you.

They weren't needed, nor would any fixes provided be capable of providing you an opponent that is the equivalent of a human. If you want to give the computers a chance to beat you, then there are numerous changes that you can make through mods, map starting locations that will give the computer the massive production advantages it needs to be able to defeat a human.

Thanks for admitting that you aren't even interested in taking any of the easily available actions to "fix" your game, if you think that playing with the most current patch is some kind of rarity.

Perhaps you should develop realistic expectations of an AI, rather than complaining that it doesn't reach your impossible requirements for performance.

Re: So...the customers wont be cheated this time?
 

Crap, having the AI build castles and labs can't be that difficult.

After all, they havn't recoded Dom3 from scratch. Therefore they have fixed both issues. Or haven't they?

And about AI strength and weakness: give the AI some pre-build castles and labs on the map to conquer. You'll find that it would get at least 50% stronger.

But it didn't take days to notice that the AI does not build castles at all and therefore recruits only indie crap because it lacks the ressources to go for heavy troops.

Obviously, a human player can always find a way to exploit some tactics the AI can't understand. But I might get some fun out of SP (and I bought Dom2 mostly for SP, too) by staying away from such abusive tactics. Therefore, having a unnecessary crippled AI destracts heavily from the fun I can have with the game.

Re: So...the customers wont be cheated this time?
 

Hmm. I think this discussion needs to stop before anyone gets annoyed. So I'll round up this with the following.

It is clear that MStavros is unsatisfied with how the single player AI works, and I can't blame him. I do not have any big problems defeating the AI at most levels.

MStavros says that we shouldn't add SP AI if we can't do it properly. Here I disagree. I have had many good and fun games watching my realm grow and heros come and go. There are people who enjoy playing SP games as roleplaying experiences where the empire is the character.

If I'm correct it is not the AI but the lack of improvement that bothers you the most. This is understandable. It is probably a bad desicion (or non-desicion) PR-wise not to improve it.

Unfortunately were not paid to do stuff. We don't have a publisher or lead developer that tells us what we must do and in what time frame. In school I get paid to teach my students this and that before summer etc. If I don't I get fired (or something). As a developer I have few restraints on me and I'm a quitter. If I get frustrated I quit the project and start another one. If I get feedback, from my own work or from players, I work harder. Thus many ideas are started, discarded and eventually restarted. If an improvement doesn't work I give up and do something else. That probably makes me a bad developer.

So, to sum this thread up:
Our skills in AI programming will never make the dominions AI a match for a human player one vs one.
On the other hand there are some issues that if fixed would improve the AI performance to some extent. How much I do not know, but it would perhaps remove some cheesier anti-AI tactics available to the experienced player. Some suggestions on cheezy-fixes have been posted here and earlier on the forum. Some of them might appear, probably not all. One of these fixes are fort-building.

I hope this clears matters a bit. Thanks!

Re: So...the customers wont be cheated this time?
 

Forts are reported as fixed. I expect AI to be significantly stronger as a result.

There is this mod for Orania map (forgot it's exact name) that is basically map file for stock Orania map and gives 5 provinces 2 castles and some special sites to all nations in game. AI performs very, very well there.

If AI can build castles (and labs of course) then there is no reason to believe that AI will be weak.

Which reminds me, that Orania mod/map (called something like Orania hard, but not sure) is one of those mods Graeme is talking about. It's designed primarily for SP and to help AI while keeping things balance. Another mod/map was provided by no other then Gandalf Parker who has random maps generated with random alliances and random bonuses which make AI much more difficult and the game more unpredictable as a result.

Edit: Whoopsie, K.O. beat me to it.

Re: So...the customers wont be cheated this time?
 

Actually if the AI builds castles some heavy extra programming would be needed such as coding the AI to stock and guard its castles... and then code to especially guard those castles on the front lines. And then code the AI to build labs in those castles so the mages can research and cast spells. And then code the AI to focus on building castles with strong magic sites and bottlenecks in the map. Now you'd have to QA test the new AI to verify humans won't be able to exploit a weakness from this building of castles.

Currently there's only two main developers working Dominions_3 and one so so helping. Besides very basic chess games I have yet to see smart AIs for any game.

Ideally the best feature we can hope for is some method where us gamers can modify/adjust the AI via scripting or another method. This will greatly open the door for more challenging AIs.

Re: So...the customers wont be cheated this time?
 

Graeme, your words are getting a bit insulting. Yes I dont know anything about scripting, but what the hell is your point? It is my fault that the Dominions 2. AI never been fixed? What are you trying to say? I understand that AI scripting is hard, but is it impossible? Do you know what "game support" means? It means that the game developers are trying to fix the bugs,errors etc. and they keep upgrading the game.
I respect Illwinter, the game is really good, but I cant respect their lack of support for the singleplayer part in Dominions 2.

Secondly, I dont expect any AIs what can beat a human player 1on1...but I expect an AI what is challenging. Beating 3 AIs alone is just annoying, 3 AIs should easily beat a human player. Do you agree? Go and try Galciv2 on the hardest vs. 2-3 AIs, and tell me that it is impossible to script a good AI.

Graeme, you dont get my whole point, and you just keep insulting me with your words. I think I was objective enough, and you call it whining.

Anyways its nice to see that Kristoffer agreed on these, and explained the situation. I hope that Dominions 3. will get singleplayer support too, thats all I can say.

Re: So...the customers wont be cheated this time?
 

First off, giving castles to the AI has been done (by me) and it didnt work. If there is a map command to add defence or patrol then Id consider it fixed.

And MStacros, THATS what your noise is about? That you can beat 3 AI's? Im not one to usually say "then switch to multiplayer" but if all this namecalling is because you want an AI that can challenge you one-on-one without cheating, then thats a human player. Sorry, but its not FAILING on illwinters part if they cant provide a level of AI that hasnt even been invented yet. You need to give up on the noise and switch to asking for improvements

Re: So...the customers wont be cheated this time?
 

Yes forts are said to be fixed... lets hope human players don't find a way to exploit this. I have lots of questions myself about the game, but waiting for the demo which should answer most of them.

Hopefully the death match has been made more worthwhile for attending.


Ideally the best feature we can hope for is some method where us gamers can modify/adjust the AI via scripting or another method. This will greatly open the door for more challenging AIs.