Dancing

[Dakara]

14 hours ago, elexis said:

On hostrules:

There's always some overpowered unit or mechanism that is ruled out by players. Last alpha skirmisher cavalry accuracy was off by 20% or something and it lead to players banning the unit from matches altogether, or resulted in every player being forced to only train cavalry as soon as possible and as much as possible. Another common user-specified matchrule is to prohibit wonders (because they allow each player +50 pop and that can lead to significant additional lag when playing with 8.). Also every player has a different understanding of what they consider toxic chatting, or when they chose to hand out bans. If that's put into some field that needs to be accepted like the other gamesettings, the player at least cannot start the game unless having formally agreed to the hosts rules.

On dancing:

Since it was in doubt what dancing is, I would define it as every command where the player moves units not to relocate the unit but to dodge arrows. (It means the sum of all units move vectors is about zero and covers a very small area.)

Hannibal_Barca has demonstrated how far one can take the bug. Practically one can send out a single hero or champion unit and do the dancing, and even if 100 archers shoot at it, there will just be 100 arrows piling up left and right of the unit every second. So to me it appears to be something that can be addressed in the accuracy code. By chance some of the arrows are expected to hit. On the other side, just reducing the accuracy to accomplish that might make the unit ineffective. So I suppose it needs a good idea how to change the projectile trajectory model without changing how much damage the unit practically delivers.

Given that one can with slightly more effort than using the patrol feature trigger the 'trajectory bug' using manual moves, changing the patrol feature just covers up the symptom. If a player patrols in place, they should be an easy target for the attacker in theory.

Bonjour,

Ce n'est pas le sujet de la discussion mais où pouvons nous trouver les % de précision de chaque unité ?

On ne peut pas augmenter légèrement la précision des archers ?

Cordialement,  

[Sundiata]

I'd just like to add that this discussion is good illustration of why it's difficult to move forward on gameplay decisions.

Dancing is an exploit, plain and simple. But there are some experienced players who have a vested interest in maintaining this exploit because it allows them to win vs noobs and/or people who refuse to stoop to dancing an individual unit to win a battle. Such cases should be clearly identified and tackled. Dancing on the frontline of a battle is just dumb, and if it allows the dancer to win the battle, it's hella frustrating to everyone that can't be bothered to go into first person Billy Eliot mode to win. 

 

This is basically the inverted of something I've suggested several times myself (I suggested ranged units randomly selecting 1 out of the 10 nearest units, spreading fire across the group being attacked, not just the nearest individual):

7 hours ago, nani said:

I have a better idea. Much fair and more easy to implement too. Set a default maximum number units that will automatically attack at the same enemy unit. It would go as follows:

Default_max: 5

1. Search closest enemy unit to attack

2. Attack that enemy unit if currently is being attacked by less than Default_max of your own units

3. Othewise and if there are more enemy units repeat steps 1 and 2 for the next closest enemy unit.

4 if all enemy units have Default_max of your own units attacking them then turn to the current standard proximity attack.

This should apply to only ranged units.

The only caveat I can see with nani's suggestion is how would you focus fire on a specific unit, like an elephant storming towards your CC? I don't think you'd have that problem if you follow my original suggestion, which is the most natural, organic, logical and intuitive solution (don't know if it's feasible from a programming side though)

I repeat my suggestion:

When a group of ranged units attacks another group, the ranged units each individually and randomly select 1 out the 10 nearest enemy units in range to fire on. This would cause groups of ranged units to fire across a front, not an individual. Dancing would be rendered ineffective and ranged combat would start looking a lot more natural. Perhaps "focus fire" could be a tech to research, to help sniping heroes? 

Secondly, directional defence should be a thing. So that units carrying a shield would become less vulnerable when attacked from the front, and far more vulnerable when flanked or attacked from the back. Currently a unit's attack has a direction, no? How come the defence stats can't be applied differently to a direction (front, back, left, right)? 

 

Also, the poll is misleading, as option 2 and option 3 are basically the same option and option 1, 2, 3 are more similar to each other, while people are allowed to vote on more than one option at a time... Skewed results.

Edited March 19, 2019 by Sundiata

[elexis]

7 minutes ago, Sundiata said:

When a group of ranged units attacks another group, the ranged units each individually and randomly select 1 out the 10 nearest enemy units in range to fire on. This would cause groups of ranged units to fire across a front, not an individual. Dancing would be rendered ineffective and ranged combat would start looking a lot more natural.

Unless the opponent dances with 10 unit. Though I agree that might reduce the problem already to the point where people don't collectively push each other into using this thing. Notice that even if we'd nuke or restrict patroling, players can script such things (at least Lefo scripted his base building commands already). Another issue is that the likelihood to strike a unit is reduced with the distance to the unit.

Mostly I wonder whether the trajectory code itself doesn't offer a nearby solution.

9 minutes ago, Sundiata said:

why it's difficult to move forward on gameplay decisions

We only need a consensus amongst coders, not everyone. If the problem is gone either way, the topic will be gone. If the problem only morphed into another problem, we have something new to talk about the alpha following that

14 minutes ago, Sundiata said:

Secondly, directional defence should be a thing. So that units carrying a shield would become less vulnerable when attacked from the front, and far more vulnerable when flanked or attacked from the back. Currently a unit's attack has a direction, no? How come the defence stats can't be applied differently to a direction (front, back, left, right)?  

[Silier]

Directional attack damage does not need battalions strictly. This mod has it https://0ad.mod.io/balancing-mod. 

Edited March 19, 2019 by Angen

[nani]

4 hours ago, Sundiata said:

The only caveat I can see with nani's suggestion is how would you focus fire on a specific unit, like an elephant storming towards your CC? I don't think you'd have that problem if you follow my original suggestion, which is the most natural, organic, logical and intuitive solution (don't know if it's feasible from a programming side though)

I repeat my suggestion:

When a group of ranged units attacks another group, the ranged units each individually and randomly select 1 out the 10 nearest enemy units in range to fire on. This would cause groups of ranged units to fire across a front, not an individual. Dancing would be rendered ineffective and ranged combat would start looking a lot more natural. Perhaps "focus fire" could be a tech to research, to help sniping heroes? 

The procedure I described would only apply the attacks make automatically (like what happens when you press halt hotkey in the middle of a battle with your units selected),

you can still focus on a specific unit by selecting your units and  sending an order to attack that specific enemy unit.

Your suggesting would work too and maybe would even be more realistic of real battle situations :D

[camel]

Do u have ketchup, cuz i dont like mayonesa.

Edited March 19, 2019 by camel

[Feldfeld]

On 3/18/2019 at 6:14 PM, elexis said:

On hostrules:

The problem is that I think it is difficult to define exactly what is dancing and what is not :
- say I have a cav targeted by multiple ranged units, I want it to get out without getting damage, then I move it out while changing its directions sometimes to avoid arrow, the cav receives less damage than it should, then would it be considered dancing ? If so consider the next situation :
- 2 big ranged armies are going to fight. The best way to engage is to have my units lined up as much as possible (with a limit to the width of the line) when ennemy enters range so that as much unit as possible fire at same time. But, crap, i didn't enable formation so i have some units more advanced than others and i want to place them back in the line. Doing so might attract the ennemy units if he doesnt care, and might have some dancing properties like in the first situation (in order to not receive too much damage), which could turn battle well in my favour, would that still be okay ?

And in case first situation is not okay, then what is the bound ? some change in directions that aren't dancing could be interpretted as such, and maybe some players who like to spam click could be wrongly called for actions that change nothing actually.

In practice, i'm pretty sure i had many games with Valihrant where there was tacital agreement not to dance, but since limit is not clear and some actions can be interpretted as dancing then, well in the end we both dance because no choice.
I mean, the host does what he wants anyway of course, but i can expect heads on fire and facepalms

19 hours ago, ffffffff said:

MAKE MAX QUEUED ORDERS PER UNIT TO 5 DANCING GONE TX

nubest suggestion of the thread

13 hours ago, Dakara said:

Bonjour,

Ce n'est pas le sujet de la discussion mais où pouvons nous trouver les % de précision de chaque unité ?

On ne peut pas augmenter légèrement la précision des archers ?

Cordialement,  

Si tu regardes dans les fichiers du jeu, tu pourras trouver dans les templates des unités la valeur "spread", qui n'est pas un pourcentage mais une valeur telle que plus elle est basse, meilleure sera la précision. Augmenter la précision des unités ne changerait rien, en réalité ça aurait juste tendance à avoir l'effet inverse : une bonne précision implique de prédire la trajectoire de la cible, alors que la dance consiste justement à changer constamment de direction pour brouiller cette prédiction. On pourrait alors penser que la solution serait de baisser la précision, mais en réalité, dans la pratique ça ne changerait rien IIRC.

12 hours ago, Sundiata said:

When a group of ranged units attacks another group, the ranged units each individually and randomly select 1 out the 10 nearest enemy units in range to fire on. This would cause groups of ranged units to fire across a front, not an individual

I think that this suggestion would bring quite a bit of side issues of units not acting like we want them to. Say i have a group of skirmisher cavalry raiding, i move them close to a group of 5 women to kill them quickly. But there's another group of 5 women (or unit) barely in range, maybe that are leaving, then only half of my army will fire effectively, where i want them to, and half of my raiding party would be firing inefficiently or drawn away and that would be quite annoying. As for dancing, its efficiency would not be reduced that much, a group of 5 units can draw 50% of enemy arrow which remains quite practical.

 

I think that this is a problem for the engine, that something can be changed like maybe how a hit is determined (make unit "hitbox" larger and perhaps compensate with lower accuracy for units ? perhaps projectiles faster as well ?). I don't think at all that it will be easy to fix, some other RTS like AoE2 and Empires Apart have some sort of dance as well.

When i play personally, what i do generally is that i don't dance (or "too much") until ennemy does it, which is why i picked my choice in the poll (closest choice)

Edited March 19, 2019 by Feldfeld

[elexis]

16 minutes ago, Feldfeld said:

then I move it out while changing its directions sometimes to avoid arrow, the cav receives less damage than it should, then would it be considered dancing ?

Dancing means moving without intending to move a unit away, but only to dodge bullets. In your example you actually move the unit elsewhere.

17 minutes ago, Feldfeld said:

i didn't enable formation so i have some units more advanced than others and i want to place them back in the line.

Meh, switching formations back and forth is another exploitable thing. In that case the units even receive run speed. That's not the dancing we're talking about, dancing is only moving backwards and forth to dodge bullets with either many move commands or patrol command. Like actual dancing where you're not getting into formation with others or transporting units (ministry of silly walks doesn't count as dancing because it's moving units).

20 minutes ago, Feldfeld said:

In practice, i'm pretty sure i had many games with Valihrant where there was tacital agreement not to dance, but since limit is not clear and some actions can be interpretted as dancing then, well in the end we both dance because no choice.

The question is whether one of you wants to report that as a cheat, or as as unintentional exploit that still cost you the score. (If you're playing top 5, you want the game result to be acurate). I guess the burden of proof is on the reporter to show that a specifc series of moves has resulted in the loss of the game. It's easy to show for most dancers and formation wigglers.

24 minutes ago, Feldfeld said:

I mean, the host does what he wants anyway of course, but i can expect heads on fire and facepalms 

That's always the case with enforcing custom terms of use. Where is the limit to jokes on the chat if the host says he will ban if players are making jokes he doesn't like? Mostly players need to organize and decide for themselves if it doesn't affect the WFG score database or WFG lobby chat. They should join the host they trust or host themselves and become a source of player trust.

30 minutes ago, Feldfeld said:

i don't dance (or "too much") until ennemy does it

That's even the more messed up part than just the invulnerable units. It forces everyone to play with the silly exploit thing (or to train nothing but the most recent OP unit).

31 minutes ago, Feldfeld said:

maybe how a hit is determined (make unit "hitbox" larger and perhaps compensate with lower accuracy for units ? perhaps projectiles faster as well ?).

I mostly think about the pile of 100 arrows in one place left and 100 arrows right of the unit. That inherently seems wrong. Perhaps one can reformulate the trajectory somehow. If we would actually look at the code, we wouldn't have to guess. It should be in Attack.js. But it's hard if not impossible to change it without affecting the balancing.

[Feldfeld]

10 minutes ago, elexis said:

Dancing means moving without intending to move a unit away, but only to dodge bullets. In your example you actually move the unit elsewhere.

Yeah but still it could be quite frustrating if a player sees that, especially in a game where dancing would be forbidden, and it could have quite an impact in the 2nd situation I described. Formations does indeed that but fixes the "issue" quickly whereas with the other way, the units manually moved would take longer to relocate, and dodge more arrow.

17 minutes ago, elexis said:

The question is whether one of you wants to report that as a cheat, or as as unintentional exploit that still cost you the score. (If you're playing top 5, you want the game result to be acurate). I guess the burden of proof is on the reporter to show that a specifc series of moves has resulted in the loss of the game. It's easy to show for most dancers and formation wigglers.

By tacital (meant tacit*) I mean that it was unsaid. It's not like it's important anyway especially since those games are unrated, and i believe still that there is no clear bound for dance. And even still, if moderators had to handle those kind of report the amount of work could be potentially much bigger than that of rated leavers

 

21 minutes ago, elexis said:

I mostly think about the pile of 100 arrows in one place left and 100 arrows right of the unit. That inherently seems wrong. Perhaps one can reformulate the trajectory somehow. If we would actually look at the code, we wouldn't have to guess. It should be in Attack.js. But it's hard if not impossible to change it without affecting the balancing.

I agree with that. Also i think it would be very fine even if balance was changed after fixing dancing (that i belive should be hard). Balance can be re-done after, and only ranged units should be affected

[av93]

Possible final solution: ranged attacks should always hit although the objective is moving, but the odds of failing are determined by the attacker template.

Edited March 20, 2019 by av93

[Feldfeld]

15 minutes ago, av93 said:

Possible final solution: ranged attacks should always hit although the objective is moving, but the odds of failing are determined by the attacker template.

I think there would be a problem to visually convey that, are we going to make projectiles magically attracted to a unit, or will we have a projectile hitting despite obviously not landing in the game ? (or projectiles at supersonic speed ?)

[av93]

1 hour ago, Feldfeld said:

I think there would be a problem to visually convey that, are we going to make projectiles magically attracted to a unit, or will we have a projectile hitting despite obviously not landing in the game ? (or projectiles at supersonic speed ?)

AoE 3 doesn't have a miss probability, and it looks good. I think that they used fast projectiles solution

[Gurken Khan]

3 hours ago, av93 said:

Possible final solution: ranged attacks should always hit although the objective is moving, but the odds of failing are determined by the attacker template.

Might be good in any situation with moving targets.

Yesterday I had an enemy swordsman walk away, and five of my skirmisher cavs (about twice the speed) chased him across half the map without landing a single hit. Of course it doesn't help that those idiots always seem to be tripping over their own and their pals' hooves, but I really think that pedestrian should've died.

[nani]

1 hour ago, ffffffff said:

On 3/19/2019 at 10:08 PM, Feldfeld said:

nubest suggestion of the thread 

u wann fight

13 hours ago, av93 said:

Possible final solution: ranged attacks should always hit although the objective is moving, but the odds of failing are determined by the attacker template.

nub

2xNub combo

[Guest]

Need to launch projectiles where it would hit instead of the targets position. For a unit moving straight, this is pretty simple.

Edited March 21, 2019 by Guest

[wowgetoffyourcellphone]

23 hours ago, Gurken Khan said:

Might be good in any situation with moving targets.

Yesterday I had an enemy swordsman walk away, and five of my skirmisher cavs (about twice the speed) chased him across half the map without landing a single hit. Of course it doesn't help that those idiots always seem to be tripping over their own and their pals' hooves, but I really think that pedestrian should've died.

You should have ran your horsemen ahead of the fleeing coward and showered him with javelins.

[Gurken Khan]

@wowgetoffyourcellphoneYeah, I guess. I thought it'd be an easy kill, didn't expect my guys to perform that poorly.

Spoiler

That swordsman was running around there because his AI spawned as Iberian and its walls blocked that northeastern chokepoint on 'Marmara'; naturally the sensible choice for the AI was neither to stay on one side of the wall nor to upgrade that piece of wall to a gate, but to run its units all around that gigantic inland sea...

 

[Luiko]

the hitbox area of units should be a bit wider than his collision area and the hitbox area should be less or equal of the outer collisions

[wowgetoffyourcellphone]

On 3/19/2019 at 5:52 AM, elexis said:

 

You. I like you.