Non-random BuildingAI

[real_tabasco_sauce]

not too hard to make a26 compatible:

@chrstgtr @borg- @ValihrAnt maybe try some 1v1s with the mod? it is compatible also with community mod.

btw you can use ctrl by default to set just a rally point and not target arrows.

non-random.zip

Edited January 11 by real_tabasco_sauce

[chrstgtr]

@real_tabasco_sauce and I did some testing tonight. We tested it using 120 skirm armies on both sides. In the first test, we had 20 units garrisoned in CC. In the second test, had 0 units garrisoned in the CC. Although the defender seemed to do better when ungarrisioned, the effect wasn't super dramatic (something like 28ish units left over when garrisioned vs 40ish units when ungarrisioned). This addressed a lot of my late game concerns. 

If people like the idea, I think we should try it. If people prefer the current/old way, we can revert in a28 (or the community mod). I expect gameplay will be impacted in more ways than we can immediately tell but the community can weigh in on whether those changes are desirable later. 

TLDR: My concerns were present but I don't think they were dramatic enough to warrant blocking this patch. Up to the community to decide if it's desirable. 

Edited January 12 by chrstgtr

[real_tabasco_sauce]

Here is a replay of a 1v1 using the mod between @ValihrAnt and @hamdich. 

non-random_1v1.zip

[Atrik]

Here some critics I came by also talking to other players:

Currently, defense buildings feel balanced the way they inflict attrition over time to a army. If turrets can kill soldiers 1by1 more efficiently, the cost of going under a turret is dramatically increased.
Healing (temple, hero..) is used mostly to heal units after taking damages from defense buildings. Healing doesn't feel overused currently. The exception that can come up is healer hero but maybe there are others ways to keep them good but not as OP.

If you try to ram a fort with garrisoned swords (something extremely difficult already), you have have very low chances because your army shrinks every time you go into the fort's range. In others words every time you get into a defensive building attack range, it cost you permanently (can't heal dead mens) more then with random arrows.

Dancing: Good micro with non-random would be to try to engage with a fast moving unit and dance to wast turrets shots until the opponent react and order the turret to attack another unit. I consider this not desirable.

Haven't tested the mod with other players yet, if someone want to test or prove me wrong about something above I'm up to it.

[real_tabasco_sauce]

1 hour ago, Atrik said:

If you try to ram a fort with garrisoned swords (something extremely difficult already), you have have very low chances because your army shrinks every time you go into the fort's range. In others words every time you get into a defensive building attack range, it cost you permanently (can't heal dead mens) more then with random arrows.

Well, I would argue this is more about the rams and less about the arrows. btw I have a patch for that too.

[alre]

15 hours ago, Atrik said:

If you try to ram a fort with garrisoned swords (something extremely difficult already), you have have very low chances because your army shrinks every time you go into the fort's range. In others words every time you get into a defensive building attack range, it cost you permanently (can't heal dead mens) more then with random arrows.

I don't believe this is true. it's a proven fact that random targeting is more effective than closest-unit targeting - see @chrstgtr comment as well. this patch is going to buff buildings against early rushing, but nerf them against late game armies.

[real_tabasco_sauce]

On 16/01/2024 at 2:14 PM, Atrik said:

Dancing: Good micro with non-random would be to try to engage with a fast moving unit and dance to wast turrets shots until the opponent react and order the turret to attack another unit. I consider this not desirable.

also, I doubt this will be a problem. Dancing was a problem when whole armies could avoid damage. In this case, someone might be able to avoid a little damage, but that will be very situational. Raiding units would not want to spend their time dancing, especially when the entities attacking them don't cost pop, and when they need to be still to attack.

Edited January 21 by real_tabasco_sauce

[BreakfastBurrito_007]

6 hours ago, real_tabasco_sauce said:

Dancing: Good micro with non-random would be to try to engage with a fast moving unit and dance to wast turrets shots until the opponent react and order the turret to attack another unit. I consider this not desirable.

This won't happen but is absolutely desirable. We always love opportunities for a player to out-skill the other.

For some context, a23 dancing (most infamous) had no counter play other than to also dance with an opposing army. In this case you can just pay attention and click a different unit. if you switch targets while your enemy dances with his cav you are killing him and his units are doing nothing productive.

[wowgetoffyourcellphone]

The "random arrows with no control" just feels like a gimped or bugged feature rather than something desirable. 

[BreakfastBurrito_007]

12 minutes ago, wowgetoffyourcellphone said:

The "random arrows with no control" just feels like a gimped or bugged feature rather than something desirable. 

Thats quite a good point. The random arrow system seems like a placeholder for some system that could yield better gameplay results. I know there will be some growing pains because players and balance have only known random arrows, but its definitely for the best to have something more intentional and situational in place. 

Edited January 22 by BreakfastBurrito_007

[chrstgtr]

11 minutes ago, wowgetoffyourcellphone said:

The "random arrows with no control" just feels like a gimped or bugged feature rather than something desirable. 

More like a straw man that I don't think anyone wants. An alternative choice of something like "random arrows with control" would be better.

I conceptually like random arrows more than non-random arrows. But no one will ever actually know until it is widely tested. Even then, this is ultimately just a preference issue.

Regardless, I think we should give this a try to see if it is in fact more enjoyable and if it is balanced. 

[wowgetoffyourcellphone]

7 minutes ago, chrstgtr said:

An alternative choice of something like "random arrows with control" would be better.

I would like this, actually. Random arrows by default, overridden by the player to target a desired enemy unit until it is dead or out of range, then it's back to random until it's retargeted by the player. 

[chrstgtr]

8 minutes ago, wowgetoffyourcellphone said:

I would like this, actually. Random arrows by default, overridden by the player to target a desired enemy unit until it is dead or out of range, then it's back to random until it's retargeted by the player. 

I think that is my preference too. But I understand how others could prefer the proposed patch. Again, this is all preferences--people can like different things and that is ok.

I think we should try the patch and see if people prefer "non-random arrows" or "random arrows." Then let popular opinion decide. I think the next iteration of the community mod will be really helpful for that and the melee rebalance, which are both huge changes

The control piece is really different feature that everyone seems to want. 

Edited January 22 by chrstgtr

[Atrik]

1 hour ago, BreakfastBurrito_007 said:

This won't happen but is absolutely desirable. We always love opportunities for a player to out-skill the other.

As always we disagree on what is skill. The current meta for attacking and defending is all about mitigating defense building attrition. When you see an attacker retreat units in time and heal after inflicting damages to defender's eco, I see it as skill.
When a defender use the attrition effect of defense building to weaken the attacker and slaughter them in a ambush in his city, seems like good plays.

With @real_tabasco_sauce patch, defense buildings mechanic are about: For attacker, bringing a/a few units closer to cc and dance, for defender, shift-click all non-dancing or weaker units (new spam sniping mechanic). These micro seems to me extremely predictable (I wouldn't have posted if it wasn't that obvious).
The random by default and focus-fire by user control patch the dancing obviously, but also increase the importance of sniping with building.

Besides, it is expected that a lot of re-balance will have to happen if you change such a thing then how defense building works. Starting by the fact that towers and cc will be far less forgiving to rushes and raids, and turtling in general would be more efficient. A full garrisoned fort already shot over a couple of seconds to kill a solo unit so I doubt overkills are compensating the efficiency of focused fire.

It's pointless to favorise spam-click meta mechanics, the more importance they have, the less others things are making any difference, like counter units, game plan, teamwork... The skills that also probably take the most time to master.

The problem isn't user controlled abilities. You could have ideas to make some buildings have some user controlled abilities, if they are fun they'll add more to the game then adding more sniping similar spamy stuff.
In that case since healing will not be rly useful anymore, I do think you remove some possible strategies and tactics to the profit of some mechanics that aren't so worth it.

[wowgetoffyourcellphone]

Y'all act like this hasn't been solved in like 100 other games, the most applicable being AOE2, which has non-random fire and does not "break" the game. 

[chrstgtr]

1 hour ago, wowgetoffyourcellphone said:

Y'all act like this hasn't been solved in like 100 other games, the most applicable being AOE2, which has non-random fire and does not "break" the game. 

Can't remember the last time I saw someone "dance" effectively in a game. I can't recall a single instance when someone tried to "dance" around a defensive building. There are a ton of reasons not to like the proposal. Dancing isn't one of those. This whole dancing discussion is so specious

Edited January 22 by chrstgtr

[BreakfastBurrito_007]

7 hours ago, Atrik said:

ith @real_tabasco_sauce patch, defense buildings mechanic are about: For attacker, bringing a/a few units closer to cc and dance, for defender, shift-click all non-dancing or weaker units (new spam sniping mechanic). These micro seems to me extremely predictable

Well this micro is really only theoretical until someone can make it work. I agree that the logic is simple, but this would be very difficult to execute. Don’t throw around the word sniping here as it’s not applicable here. The value of shift clicking in the situation is not there either, since the defender would want to respond in real time to the movements of the attacker. “Spam clicking” is a highly unrealistic projected outcome, like your personal boogeyman that you see around every alley. 

[real_tabasco_sauce]

anyone know how to refresh a poll? I would like to start a new one with more detailed questions.

[Stan`]

Removed the old poll, you can you try again?

[chrstgtr]

@real_tabasco_sauce suggestion: you should change the text of third option in the last Q. Everyone wants balanced arrows—some want random balanced arrows while others want non-random balanced arrows. I know that’s not what you mean but it’s unclear and is the type of option that suggests a “right” poll response. 

Options like below seem better:

(1) non-random arrows with community mod 26.6 values

(2) non-random arrows with different values than community mod 26.6 values

(3) random arrows unless manually targeted

(4) random arrows (pre community mod 26.6)

 

OR, better, just ask the question straight up: 

“Should buildingAI have random  arrows or attack the nearest unit”

(1) random

(2) attack nearest unit 

the second question is more straight forward and doesn’t get into the dirty details.  

[real_tabasco_sauce]

3 minutes ago, chrstgtr said:

@real_tabasco_sauce suggestion: you should change the text of third option in the last Q. Everyone wants balanced arrows—some want random balanced arrows while others want non-random balanced arrows. I know that’s not what you mean but it’s unclear and is the type of option that suggests a “right” poll response. 

That question is about what to do going forward. Do we balance arrows to solve the above issues? or revert/semi-revert the random arrows.

It doesn't make sense to have a "random balanced arrows" option because the arrows counts were already balanced for random arrows. That would just be the same as a full revert.

6 minutes ago, chrstgtr said:

OR, better, just ask the question straight up: 

“Should buildingAI have random  arrows or attack the nearest unit”

(1) random

(2) attack nearest unit 

the second question is more straight forward and doesn’t get into the dirty details.  

That's basically the first question.

[real_tabasco_sauce]

What do u all think should be done with sentry towers? Slight bump in cost? Increase build time? change arrows counts/fire rate?

[chrstgtr]

Man you gotta accept some feedback when you’re clearly wrong…at least acknowledge it could’ve been better if you don’t think it’s necessary 

7 minutes ago, real_tabasco_sauce said:

That question is about what to do going forward. Do we balance arrows to solve the above issues? or revert/semi-revert the random arrows

 Then ask that. That is what I tried to in my suggested alternative question. 
 

8 minutes ago, real_tabasco_sauce said:

It doesn't make sense to have a "random balanced arrows" option because the arrows counts were already balanced for random arrows. That would just be the same as a full revert.

You missed the point. Your third option choice could be random or non-random. Both could be hypothetically be “balanced.”

Using the term “balance” also suggests an answer. It’s like asking a person “do you support (A) the good candidate or (B) the other guy. There is a suggested answer.  

11 minutes ago, real_tabasco_sauce said:

That's basically the first questio

No. First question is do you like 26.6, including its current values, more than pre-26.6. 
 

What I suggested is “do you want random or near-unit arrows.” For some, they could say “I want non-random but not at 26.6 value.” That person could say no to your first question. what I suggested is actually more favorable to your position 

[chrstgtr]

10 minutes ago, real_tabasco_sauce said:

What do u all think should be done with sentry towers? Slight bump in cost? Increase build time? change arrows counts/fire rate?

1 and 3 if non-random

[real_tabasco_sauce]

18 minutes ago, chrstgtr said:

 Then ask that. That is what I tried to in my suggested alternative question. 

18 minutes ago, chrstgtr said:

No. First question is do you like 26.6, including its current values, more than pre-26.6. 
 

What I suggested is “do you want random or near-unit arrows.” For some, they could say “I want non-random but not at 26.6 value.” That person could say no to your first question. what I suggested is actually more favorable to your position 

Ah i see.

18 minutes ago, chrstgtr said:

You missed the point. Your third option choice could be random or non-random. Both could be hypothetically be “balanced.”

Sure, so would a full revert. The reason the option involves the term "balance" is because that is what would be done by changing the arrow counts. The system is different, so it should be made balanced. Basically, it describes well the task of adjusting arrow counts now that they work differently.

We have a new system that has disrupted balance as evidenced by questions 2-4, and that solution would involve restoring balance by changing arrow counts

I suppose I could avoid the term "balance," but then I would need a lot more words than necessary.

Edited March 12 by real_tabasco_sauce