Why..?

General feedback and discussion of the game.

Moderators: Forum Moderators, Developers

Tux2B
Posts: 1217
Joined: March 29th, 2005, 8:18 pm
Location: Toulouse (South of France)
Contact:

Post by Tux2B »

Elvish Pillager wrote:The key to not losing units is in not putting them where they can be killed. A straight line of undamaged units, for instance, spaced with one hex between them, is a powerful defensive formation that rarely sustains casualties.

Of course, the units do get damaged. When the line is attacked and units are damaged, those units can either retreat or attack the attackers, then be defended as the line reforms one hex ahead around the dead attackers. If you start running out of high-HP units, then it may be a good idea to start destroying the enemies at all costs; agressive strategies that leave all the enemies dead don't leave any enemies to kill your units.
It is good to put that line where you have a village, or when you can put either healers or leaders (I mean units with leadership) behind.
Such a line can be very strong, most of all since the AI is completely unable to make such one.
"There are two kind of campaign strategies : the good and the bad ones. The good ones almost always fail because of unforeseen consequences that make the bad ones succeed." -- Napoleon
Chris Byler
Posts: 99
Joined: April 14th, 2005, 2:32 pm
Location: Blacksburg, VA, USA

Post by Chris Byler »

Darth Fool wrote:It is important to realize that the AI is pretty brutal about attacking and killing weakened units. It generally considers that killing one unit is a lot more valuable then wounding several. And usually, that is correct.
However, the AI will also expose its own units in vulnerable positions to get that kill, and usually, *that* is *not* correct. The AI can sort of get away with it because of its massive material advantages in almost every scenario.

Using a low-value, wounded unit as bait to draw AI units into exposed positions and/or draw fire away from more valuable units is a very useful strategy against the AI. Just don't expect it to work as well on human opponents.

Actually, it doesn't even have to be a wounded unit: a unit in a surroundable position will work just as well. The AI does take into account how it can improve its odds of killing something by ganging up on it.
Dave
Founding Developer
Posts: 7071
Joined: August 17th, 2003, 5:07 am
Location: Seattle
Contact:

Post by Dave »

Chris Byler wrote:
Darth Fool wrote:It is important to realize that the AI is pretty brutal about attacking and killing weakened units. It generally considers that killing one unit is a lot more valuable then wounding several. And usually, that is correct.
However, the AI will also expose its own units in vulnerable positions to get that kill, and usually, *that* is *not* correct. The AI can sort of get away with it because of its massive material advantages in almost every scenario.
The AI won't do it nearly so much if you set its aggression lower. In most scenarios though, AI aggression is set pretty high. It is also largely because it's difficult for the AI to play defensively.

However, I don't think the AI has a 'massive material advantage' in most scenarios. On the contrary, if the player is doing well in a typical campaign, the material on either side is usually reasonably close. A scenario where the AI has 150% of the player's strength and the player has to destroy the AI is generally considered a difficult scenario.

For instance, the Siege of Elensefar, a scenario that is generally recognized as being fairly difficult, has the AIs with combined starting gold of 340. Their starting village advantage might possibly be worth an additional 40 gold or so, giving them around 380 gold total. If you look at http://stats.wesnoth.org/?W_SCENARIO=Th ... IFF=NORMAL you will see that players have between 350 and 700 gold at the start of the scenario (a small sample size so far, but I think others can agree that this is typical).

Considering that the player will typically have six or more level two units, and perhaps one or two level three units that can be recalled for 20 gold each, while the AI will have to pay full price for its level two units, I think the player has a reasonable material advantage in this scenario.

The advantage the AI does have in a campaign is that the player has future battles to fight, and the AI can 'win' by doing sufficient damage to the player, while still losing the battle. I will concede that this is somewhat of a weakness in the game: the AI should appear to be trying to win the current battle, not take an approach of "oh well, I'll let you beat me, but will try to do enough damage to you that you lose in a later battle."

Of course, the main reason for this is that programming a very good AI is very hard. :)

David
“At Gambling, the deadly sin is to mistake bad play for bad luck.” -- Ian Fleming
User avatar
Ken_Oh
Moderator Emeritus
Posts: 2176
Joined: February 6th, 2006, 4:03 am
Location: Baltimore, Maryland, USA

Post by Ken_Oh »

First time poster, just saying I think what the developers have done with this game is incredible. Mad props.

The points that the first poster, Orgemage, brings up are an excellent list of reasons why I like the game and think it's something really unique.
arno
Posts: 42
Joined: January 25th, 2006, 2:12 pm
Location: France

Re: Why..?

Post by arno »

Dave wrote:
Elvish Pillager wrote: The key to liking this game is to see its strategies, not to apply preconcieved notions of strategy to it. Before playing, forget all you know about TBS games and relive the discovery all over again... ;)
Very well said. This is the key to liking Wesnoth.
I also agree.
Aloo
Posts: 18
Joined: July 21st, 2005, 10:49 am
Location: Poland

Post by Aloo »

Players who always load a saved game whenever they lose a levelled unit in other strategy games or RPGs will probably not like Wesnoth.

Players who always 'must' have their army at maximum power, winning every battle 'perfectly' will probably not like Wesnoth.
Thats me on my first go on HTtT. I played on easy and save reloaded each time I lost any leveled unit (getting mad in the process). I play normal now and reload only when losing very important units (once per 3-4 scenarios).

I guess I just had to get used to the way the game works - you have to loose units.
Chris Byler
Posts: 99
Joined: April 14th, 2005, 2:32 pm
Location: Blacksburg, VA, USA

Post by Chris Byler »

Hmm, I guess that's true Dave. It just seemed like I was outnumbered because I was trying to win with few casualties, since unlike the AI, I need to save my army for later.

The only ones where the enemy really has massive material advantages are scenarios where you don't have to (and shouldn't try to) destroy all enemies, like Home of the North Elves, and ones where you have only minimum gold because the previous scenario drained it all (sometimes avoidable, sometimes not really). IIRC I came into Return to Wesnoth pretty much broke from the ultra-long, village-poor, must-have-lots-of-cockatrice-bait Valley of Statues, for example. But on Battle for Wesnoth I had 1300 gold... huge early finish bonus on TotC, which is a large map with tons of villages.

Sometimes, though, even if you have lots of gold and advanced units, you still end up outnumbered because of keep space imbalance, at least for a while. You having 6 keep spaces while 2-3 enemies have 6 keep spaces *each* is fairly typical. Since, by design, one level 3 unit is not as strong as 3 level 1 units (unless fighting on a narrow front, which you can't always arrange), this can be a problem.
arno
Posts: 42
Joined: January 25th, 2006, 2:12 pm
Location: France

Post by arno »

Darth Fool wrote:It is important to realize that the AI is pretty brutal about attacking and killing weakened units. It generally considers that killing one unit is a lot more valuable then wounding several. And usually, that is correct.
War is brutal.
toms
Posts: 1717
Joined: November 6th, 2005, 2:15 pm

Post by toms »

One thing would make the AI far better: If he could use abilities like backstab or berserk. :o
First read, then think. Read again, think again. And then post!
torangan
Retired Developer
Posts: 1365
Joined: March 27th, 2004, 12:25 am
Location: Germany

Post by torangan »

toms wrote:One thing would make the AI far better: If he could use abilities like backstab or berserk. :o
Describing what the AI should do is easy, putting it into any programming language a very complex task under active reasearch. There's a reason why AI is a huge subdivision of computer science - it's hard and complex work!
WesCamp-i18n - Translations for User Campaigns:
http://www.wesnoth.org/wiki/WesCamp

Translators for all languages required: contact me. No geek skills required!
toms
Posts: 1717
Joined: November 6th, 2005, 2:15 pm

Post by toms »

torangan wrote:
toms wrote:One thing would make the AI far better: If he could use abilities like backstab or berserk. :o
Describing what the AI should do is easy, putting it into any programming language a very complex task under active reasearch. There's a reason why AI is a huge subdivision of computer science - it's hard and complex work!
I know and it is no bad thing that should be fixed immediately.

But have you seen what a chess engine can do? :shock:
First read, then think. Read again, think again. And then post!
User avatar
Elvish_Pillager
Posts: 8129
Joined: May 28th, 2004, 10:21 am
Location: Everywhere you think, nowhere you can possibly imagine.
Contact:

Post by Elvish_Pillager »

Chess engines do not have to deal with chance, and there are also far, far fewer choices for a chess engine to choose from than for a Wesnoth engine.
It's all fun and games until someone loses a lawsuit. Oh, and by the way, sending me private messages won't work. :/ If you must contact me, there's an e-mail address listed on the website in my profile.
Dave
Founding Developer
Posts: 7071
Joined: August 17th, 2003, 5:07 am
Location: Seattle
Contact:

Post by Dave »

toms wrote: But have you seen what a chess engine can do? :shock:
There are some chess engines that can play chess extremely well. However, these are usually centered around optimizing for the specific, narrow rules of chess.

Imagine, for instance, you had to write a chess engine that played standard chess, but allowed rule customization. It allowed any starting position (this immediately eliminating its use of an "openings library"). It allowed custom pieces which moved in non-standard ways, such as 'camels', 'maharajahs', and other such fantasy chess pieces, with their movement rules specified by a configuration file.

It also allowed custom winning conditions, such as perhaps moving a certain piece to the other end of the board, or capturing a certain piece. It allowed any board size -- even a much larger board like 50x50. It even allowed more than two players to play -- you could have a 2 vs 2 game, or a free-for-all.

How well could such an engine play? Not nearly so well as current chess engines. In fact, with things like the board getting larger, the possibilities would increase dramatically, and its performance would plummet.

If you made a 'standard', rather small Wesnoth map, set up a scenario where each player gets a certain amount of gold, and have fixed factions, it would probably be possible to write a very powerful AI that can just play that scenario. Especially if you invested the time and energy that has been invested in making powerful chess engines.

Of course, I haven't even touched on the actual gameplay differences that make chess much simpler than Wesnoth.

David
“At Gambling, the deadly sin is to mistake bad play for bad luck.” -- Ian Fleming
toms
Posts: 1717
Joined: November 6th, 2005, 2:15 pm

Post by toms »

Heh, I didn´t say that I want a AI wich is unbeatable or plays very awesome.
It´s good that it´s possible to set the AI at a better position at gold or recruitment.

I just think that the AI for an example can´t think over more turns than one by figuring out the probablities what the human player will do.

Also the never tries to trick me with a pseudo attack or so. And he can´t handle flanking strategies very good. If you break the line somewhere or you come from aside and get close to his leader, he will pull his units back and you can go forward.

These are the main points where the AI is weak, IMO. But as said, if he has to be stronger, I make a map wich promotes him or give him more gold. And the ai WML is also useful here.
:)
First read, then think. Read again, think again. And then post!
jonadab
Posts: 148
Joined: October 7th, 2005, 2:33 am
Location: Ohio
Contact:

Re: Why..?

Post by jonadab »

OgreMage wrote:especially when I have to restart 1 scenario in a campaign 13 times to save one of my units from some ridiculous combat situation that was as unavoidable as it was infuriating.
There are various strategy guides in these fora and on the wiki that will help you here.
OgreMage wrote:1. Units cannot support each other in battle, so 1 guy in the middle of 3 other allies is almost as vulnerable as one guy alone.
This is only true if your enemies have the skirmisher ability, which is fairly rare (e.g., Saurians have it). If you don't understand how to use Zones of Control to protect your units, see the basic strategy guides.
OgreMage wrote:This also reduces the importance of archers and stresses the importance of infantry units, since infantry are more likely to survive the punishment of a concentrated attack.
Depends. I find Elvish Archers survive better than Elvish Fighters, on average.
OgreMage wrote:2. Why can't healers stack, so that if one unit that is poisoned and damaged be both healed and cured if he has 2 healers next to him? Instead, a unit that is poisoned and injured is out of commission for 2 turns, not 1. Also, since poisoning always does 8 damage the turn after it is delivered, the unit is almost guaranteed to be badly injured.
Dealing with poison is a learned skill. There are tips about it in the strategy guides. Among other things, poisoning does not necessarily have to do 8 dammage the turn after it is delivered, if you are backing your units with healers. Also, units with poison attacks tend to be weak in other ways.
OgreMage wrote:3. Why was poison given to a class of units that can dodge most attacks, making it almost a given that they will survive long enough to poison their attacker? If anything, it should be made into a melee attack, so that the one unit that has a decent chance to hit them, mages, can do so without being poisoned for their trouble.
Oh, Assassins. Yeah. They *are* a bit hard to hit, and it *can* be frustrating. There are several major ways to deal with this. First, you can use units with attacks that ignore the defender's elusiveness. Sharpshooters, for instance. Second, you can use units with high repeat values on their attacks, so they have more chances to hit. Third, you can use units that are immune to poison, such as woses or ghosts. Fourth, you can fight poison with poison. At least one of these options should be available to you almost no matter which faction you are. If you are playing Heir to the Throne, the most obvious way to deal with Assassins is to kill them with mage fire, and place your Druid behind the mages on the same turn, to cure the poison before it does any dammage.
OgreMage wrote:4. Why does a charge attack confer an advantage to both the attacker and defender, and why does this always seem to benefit the defender?
There are some situations in which charge attacks are not what you want to use. If you try to use them against e.g. Skeleton Archers at night, they will die. In that case, use a different kind of attack.

And yes, some units are harder to keep alive than others, and the horseman line and the mage line are two of the hardest ones to keep alive. Different units have different strengths and weaknesses.
OgreMage wrote:I understand the rationale behind this, that a charging horseman can be taken out with a single stroke more easily than a horseman who is using more caution,
Oh. And here's me thinking the rationale was that they'd be terribly overpowered otherwise.
OgreMage wrote:but since the horse units are only given 2 attacks and also cannot receive good defense bonuses, they are not only less likely to hit their target than the target is likely to hit them,
Depends on the target. Some units are inappropriate targets for horsemen.
OgreMage wrote:5. Why aren't units allowed to attack then move,
If all units could do that, it would change the tactics of the game considerably, and probably the strategy as well. If only certain highly specialized units could do it, it might fit into Wesnoth better.
OgreMage wrote:why does day/night confer such large bonuses? This both deeply disadvantages neutral races
And here I was thinking that neutral races had an advantage.
OgreMage wrote: as well as provides a large incentive to the attacker, since the attacker is often able to choose what time of day to battle in.
Not always -- and in any case the defender usually gets to choose the terrain.
OgreMage wrote:The bonus from day/night is so large that I often find my level 2 and 3 units being routed by level 1 units in a single turn
Definitely see the basic strategy guides.
OgreMage wrote:7. Why does being injured not impare combat effectiveness?
Because on the whole it would be complicated, hard for the player to keep track of, and confer no substantial gameplay benefit.
OgreMage wrote:This also reduces the importance of healing, since it encourages all-out attacks over careful coordination
I don't think it does that. Careful coordination is very important in Wesnoth's tactics. From your comments, it doesn't seem that you have mastered it yet.
OgreMage wrote: And maybe the problems that I'm coming up with are mostly capaign related
There are threads with tips for the specific campaigns, and of course you can always start a new thread if you run across an issue that's not well covered.
Post Reply