Latest Posts

Topic: give some bonus to the defender?

WorldSavior
Avatar
Joined: 2016-10-15, 04:10
Posts: 2091
OS: Linux
Version: Recent tournament version
Ranking
One Elder of Players
Location: Germany
Posted at: 2019-07-19, 18:55

king_of_nowhere wrote:

WorldSavior wrote:

king_of_nowhere wrote:

defending troops do not go back in their buildings to heal as long as there is a single attacking soldier still fighting. when the attacker strikes again, all defending troops move out to intercept.

including troops that are barely alive, that should have been left to heal.

True, but I suggested that one could tell them to stay. And one can send wounded defenders simply away by clicking on them, so they don't stay in the building, so they will not fight.

well, it's a bit more complicated. you click on a wounded soldier to send it away, if it's a rookie all if fine. if the soldier has some promotion, then generally a rookie will be sent in its place.

No: The first soldier in queue instead.

and then once the rookie gets in, another soldier is generally sent to substitute it.

Not necessarily

and then maybe the very soldier you kicked out of the building is sent back because it has more promotions.

Can be prevented

messing around by manually removing wounded soldiers can backfire in spectacular ways.

But it doesn't have to

What about compensating it by more possible orders for defending troops?

that could be an even better solution. since the problem is "the attacker has more control and can use it to advantage", making things even by giving control to the defender would work. the only problem is that the attacker strike first, the defender will react, so we need options that will work assuming the defending player will look at the battlefield some ten seconds later.

Why ten seconds? The defender should check immediately the attack.

I think making it so that you can still see a shadow version of soldiers from a building when they are out, and you can manually recall them back - or send them out to intercept an attacker, or order them to not get out until they are healed - would be enough. that way, both attacker and defender have a similar lever of control over picking soldiers that go to battle.

Nice options

JanO wrote:

What is the current way how widelands picks the defenders? Could it be set to something like "form left to right" in the list of soldiers inside a building?

"From left to right" is already the case when they are leaving the building to intercept. If they don't have time for interception I'm not sure, generally full hp heroes will go out first then, but full hp rookies are favored over wounded heroes. So there is already a helpful algorithm, but controls could still be useful.

Then all problems may be solved by the possibility to drag and drop them into desired positions and set a limit, how many of them are allowed to leave.

That would be another possibility. I think that they don't even contradict each other...

Still I would prefer a menu similar to the one from settlers2 face-wink.png

How does that look like?

hessenfarmer wrote:

WorldSavior wrote:

hessenfarmer wrote:

WorldSavior wrote:

hessenfarmer wrote:

did some calcualtions. Defence value can't be used as results are to tribe specific. But evade could be used

Yes, but why? One could also change "attacker hits first" into "defender hits first" - but is this really necessary? (If two attackers intercept each other the first hit could be randomly chosen.) Defenders already have the advantage that they are healed in the buildings, and attackers often have to walk long distances until they can be healed again. Furthermore the attackers usually don't exactly know who they will be facing, so players should often think twice or more if they really should attack.

Well we could always discuss necessity. perhaps we should do this first. But the thread strted by the feeling that we gave the attacker a bonus by being able to select attacking soldiers which we should compensate a bit on the defence site.

What about compensating it by more possible orders for defending troops?

May I conclude from this answer that you agree on the necessity to help the defender a bit due to the new options we made for the attacker?

No, I don't. At the other hand it will maybe not decrease quality of the game if defenders get a bonus. So I'm not strictly against that.

From your last posts I was unsure about that. If yes we can discuss implementation details.

Giving the defender a bonus by giving him more orders is difficult imho, as one might might miss to react to a specific attack in case a well orchestrated multiple attack at different borders is launched. So only general settings might help, which might fail as there is no such thing as a general setting helpful in any case.

Maybe one could implement "defender hits first" and more orders both.

Making the defender strike first is a rather big change in some situation I believe from the numbers.

But smaller than complicated evade bonuses or group bonuses, right?

Yes my first idea was to overambitioned for this task. I now beleive that just adding 2 to the current defender evade value will decrease the attackers chance over all levels (checked for L10 vs L10 and L0 vs L0) by 5% (e.g. 68% instead of 73% for fri L0 vs bar L0). This will be just a very minor coding issue and easy to implement. If this proves to be to much then we could try just using 1.

Sounds hard to believe to me, because increasing evade from 64% to 66% is a bigger bonus than increasing from 35% to 37%. And it's still complicated...

hessenfarmer wrote:

Hi these are the actual values of the chances to win as in r20. https://www.widelands.org/forum/post/26265/

These values are calculated with the attacker listed in the first column striking first. So the difference is dependent on the level of soldiers for example for Level 0 this would mean quite a big change.

One of the highest difference would be that some 57% success rate turns into 43% (bar vs bar), which looks like 14% difference. But most other duels look more like 10% difference. And it's very uncomplicated, and don't you want that the defender gets a bonus? face-wink.png

Yes but 10 % or even more would be too high as you correctly stated before.

Did I? Why would it be too high?


Wanted to save the world, then I got widetracked

Top Quote
king_of_nowhere
Avatar
Topic Opener
Joined: 2014-09-15, 18:35
Posts: 1668
Ranking
One Elder of Players
Posted at: 2019-07-19, 23:10

WorldSavior wrote:

king_of_nowhere wrote:

WorldSavior wrote:

on e can send wounded defenders simply away by clicking on them, so they don't stay in the building, so they will not fight.

well, it's a bit more complicated. you click on a wounded soldier to send it away, if it's a rookie all if fine. if the soldier has some promotion, then generally a rookie will be sent in its place.

No: The first soldier in queue instead.

yes, which most of the times is a roookie. you can cycle soldiers by closing and reopening the spot repeatedly, but if other soldiers are out of the building sometimes it keeps the new soldier and kick those out. you can cut road and remake road, provided you have a road that you can afford to cut to isolate the military building.

but regardless of details, all those options are time-consuming and attention-consuming, and they require considerably more skill to be used by the defender than the skill required from the attacker to only pick healty soldiers to attack.

that could be an even better solution. since the problem is "the attacker has more control and can use it to advantage", making things even by giving control to the defender would work. the only problem is that the attacker strike first, the defender will react, so we need options that will work assuming the defending player will look at the battlefield some ten seconds later.

Why ten seconds? The defender should check immediately the attack.

because, related to the above answer, I assume that those options should be useable by a regular person.

I have no doubt that a korean starcraft champion the routinely performs 500 actions per minute will check the attack in 0.5 seconds and will perform the road-cutting trick in real time with 4 different buildings with enough attention span left to watch a youtube video and play a poker game.

however, most people can only focus their attention on one thing at a time, and they aren't all that fast either. and the game should still be usable by those players.

I think making it so that you can still see a shadow version of soldiers from a building when they are out, and you can manually recall them back - or send them out to intercept an attacker, or order them to not get out until they are healed - would be enough. that way, both attacker and defender have a similar lever of control over picking soldiers that go to battle.

Nice options

glad you also like the idea. can we get someone to try and code something like that?


Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2645
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-07-20, 21:03

WorldSavior wrote:

Making the defender strike first is a rather big change in some situation I believe from the numbers.

But smaller than complicated evade bonuses or group bonuses, right?

Yes my first idea was to overambitioned for this task. I now beleive that just adding 2 to the current defender evade value will decrease the attackers chance over all levels (checked for L10 vs L10 and L0 vs L0) by 5% (e.g. 68% instead of 73% for fri L0 vs bar L0). This will be just a very minor coding issue and easy to implement. If this proves to be to much then we could try just using 1.

Sounds hard to believe to me, because increasing evade from 64% to 66% is a bigger bonus than increasing from 35% to 37%. And it's still complicated...

It is definitly not complicated as not much code more than a single +2 in one line is needed. I' ll try to make the numbers and post them to show you.

hessenfarmer wrote:

Hi these are the actual values of the chances to win as in r20. https://www.widelands.org/forum/post/26265/

These values are calculated with the attacker listed in the first column striking first. So the difference is dependent on the level of soldiers for example for Level 0 this would mean quite a big change.

One of the highest difference would be that some 57% success rate turns into 43% (bar vs bar), which looks like 14% difference. But most other duels look more like 10% difference. And it's very uncomplicated, and don't you want that the defender gets a bonus? face-wink.png

Yes but 10 % or even more would be too high as you correctly stated before.

Did I? Why would it be too high?

Yes you did it. see https://www.widelands.org/forum/post/28909/ face-wink.png and I believe it to be too high as it will change the game a lot and not only a bit if you have 10 % chance less to win a fight if you attack.


Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2645
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-07-20, 21:44

ok this is the actual value for L0 vs L0

vs. bar_00 emp_00 atl_00 fri_00
bar_00 56.9% 41.8% 31.6% 37.9%
emp_00 70.0% 56.1% 42.9% 47.2%
atl_00 78.6% 67.6% 55.7% 54.0%
fri_00 72.9% 63.6% 56.5% 55.2%

this is L10 vs L10

vs. bar_00 emp_00 atl_00 fri_00
bar_00 53,72% 51,96% 49,94% 52,64%
emp_00 55,89% 52,45% 49,51% 55,27%
atl_00 57,27% 57,83% 53,5% 53,38%
fri_00 58,98% 55,12% 55,94% 59,17%

this is L0 vs L0 while the defender gets 2% higher evade

vs. bar_00 emp_00 atl_00 fri_00
bar_00 53,2% 37,19% 27,47% 33,96%
emp_00 66,06% 50,89% 38,37% 42,88%
atl_00 74,27% 63,34% 50,77% 49,19%
fri_00 69,03% 58,33% 51,85% 50,88%

and this L10 vs L10 with the bonus

vs. bar_00 emp_00 atl_00 fri_00
bar_00 51,02% 48,71% 46,39% 51,19%
emp_00 51,48% 49,48% 44,01% 51,24%
atl_00 53,51% 53,17% 48,52% 50,07%
fri_00 54,09% 51,12% 51,38% 56,37%

Top Quote
WorldSavior
Avatar
Joined: 2016-10-15, 04:10
Posts: 2091
OS: Linux
Version: Recent tournament version
Ranking
One Elder of Players
Location: Germany
Posted at: 2019-07-25, 19:58

hessenfarmer wrote:

ok this is the actual value for L0 vs L0

vs. bar_00 emp_00 atl_00 fri_00
bar_00 56.9% 41.8% 31.6% 37.9%
emp_00 70.0% 56.1% 42.9% 47.2%
atl_00 78.6% 67.6% 55.7% 54.0%
fri_00 72.9% 63.6% 56.5% 55.2%

this is L10 vs L10

vs. bar_00 emp_00 atl_00 fri_00
bar_00 53,72% 51,96% 49,94% 52,64%
emp_00 55,89% 52,45% 49,51% 55,27%
atl_00 57,27% 57,83% 53,5% 53,38%
fri_00 58,98% 55,12% 55,94% 59,17%

this is L0 vs L0 while the defender gets 2% higher evade

vs. bar_00 emp_00 atl_00 fri_00
bar_00 53,2% 37,19% 27,47% 33,96%
emp_00 66,06% 50,89% 38,37% 42,88%
atl_00 74,27% 63,34% 50,77% 49,19%
fri_00 69,03% 58,33% 51,85% 50,88%

and this L10 vs L10 with the bonus

vs. bar_00 emp_00 atl_00 fri_00
bar_00 51,02% 48,71% 46,39% 51,19%
emp_00 51,48% 49,48% 44,01% 51,24%
atl_00 53,51% 53,17% 48,52% 50,07%
fri_00 54,09% 51,12% 51,38% 56,37%

Looks good. In that case it should be mentioned in the warfare tutorial that the defender gets an evade bonus...

hessenfarmer wrote:

WorldSavior wrote:

Making the defender strike first is a rather big change in some situation I believe from the numbers.

But smaller than complicated evade bonuses or group bonuses, right?

Yes my first idea was to overambitioned for this task. I now beleive that just adding 2 to the current defender evade value will decrease the attackers chance over all levels (checked for L10 vs L10 and L0 vs L0) by 5% (e.g. 68% instead of 73% for fri L0 vs bar L0). This will be just a very minor coding issue and easy to implement. If this proves to be to much then we could try just using 1.

Sounds hard to believe to me, because increasing evade from 64% to 66% is a bigger bonus than increasing from 35% to 37%. And it's still complicated...

It is definitly not complicated as not much code more than a single +2 in one line is needed.

Okay...

hessenfarmer wrote:

Hi these are the actual values of the chances to win as in r20. https://www.widelands.org/forum/post/26265/

These values are calculated with the attacker listed in the first column striking first. So the difference is dependent on the level of soldiers for example for Level 0 this would mean quite a big change.

One of the highest difference would be that some 57% success rate turns into 43% (bar vs bar), which looks like 14% difference. But most other duels look more like 10% difference. And it's very uncomplicated, and don't you want that the defender gets a bonus? face-wink.png

Yes but 10 % or even more would be too high as you correctly stated before.

Did I? Why would it be too high?

Yes you did it. see https://www.widelands.org/forum/post/28909/ face-wink.png and I believe it to be too high as it will change the game a lot and not only a bit if you have 10 % chance less to win a fight if you attack.

Oh right, I'd forgotten that already ; )

king_of_nowhere wrote:

WorldSavior wrote:

king_of_nowhere wrote:

WorldSavior wrote:

on e can send wounded defenders simply away by clicking on them, so they don't stay in the building, so they will not fight.

well, it's a bit more complicated. you click on a wounded soldier to send it away, if it's a rookie all if fine. if the soldier has some promotion, then generally a rookie will be sent in its place.

No: The first soldier in queue instead.

yes, which most of the times is a roookie. you can cycle soldiers by closing and reopening the spot repeatedly, but if other soldiers are out of the building sometimes it keeps the new soldier and kick those out. you can cut road and remake road, provided you have a road that you can afford to cut to isolate the military building.

but regardless of details, all those options are time-consuming and attention-consuming, and they require considerably more skill to be used by the defender than the skill required from the attacker to only pick healty soldiers to attack.

I see

that could be an even better solution. since the problem is "the attacker has more control and can use it to advantage", making things even by giving control to the defender would work. the only problem is that the attacker strike first, the defender will react, so we need options that will work assuming the defending player will look at the battlefield some ten seconds later.

Why ten seconds? The defender should check immediately the attack.

because, related to the above answer, I assume that those options should be useable by a regular person.

I have no doubt that a korean starcraft champion the routinely performs 500 actions per minute will check the attack in 0.5 seconds and will perform the road-cutting trick in real time with 4 different buildings with enough attention span left to watch a youtube video and play a poker game.

however, most people can only focus their attention on one thing at a time, and they aren't all that fast either. and the game should still be usable by those players.

Okay


Wanted to save the world, then I got widetracked

Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2645
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-07-25, 22:55

maybe I'll find some time next week to prepare a branch with the evade bonus so it can be tested and properly evaluated.


Top Quote
king_of_nowhere
Avatar
Topic Opener
Joined: 2014-09-15, 18:35
Posts: 1668
Ranking
One Elder of Players
Posted at: 2019-08-22, 17:49

bump: is this going forward?


Top Quote
teppo

Joined: 2012-01-30, 09:42
Posts: 423
Ranking
Tribe Member
Posted at: 2019-08-31, 12:05

king_of_nowhere wrote:

bump: is this going forward?

I kind-of would like to do the 2019-07-07 suggestion, despite not having time.. If that turned out to be easy to implement, would you like it?


Top Quote
GunChleoc
Avatar
Joined: 2013-10-07, 15:56
Posts: 3324
Ranking
One Elder of Players
Location: RenderedRect
Posted at: 2019-08-31, 13:01

There are 6 posts from thay day with differing opinions in them - which one do you mean? You can use the anchor symbol to link to individual posts face-wink.png


Busy indexing nil values

Top Quote
GunChleoc
Avatar
Joined: 2013-10-07, 15:56
Posts: 3324
Ranking
One Elder of Players
Location: RenderedRect
Posted at: 2020-03-03, 11:08

I have created an experimental branch for testing: https://github.com/widelands/widelands/pull/3736

  • All soldiers present in a military site are healed simultaneously
  • Soldiers stationed but not present are 1/2 healed if they are on home territory and not actively fighting an opponent

If course, this will also speed up how frequent follow-up attacks can be triggered.


Busy indexing nil values

Top Quote