Discussion of changes to slow?
Moderator: Forum Moderators
Discussion of changes to slow?
Did I miss it? I think I like the new changes but I can't see the benefit of:
"* hitting with slow changes attack order to allow attacker to strike again"
Aren't most slow attacks ranged anyway? Who would want to slow a ranged unit to capitalize on such an advantage? I suppose it helps defensively but two free shots at the start is balanced by two free shots at the end...I can see how this is supposed to create a "get them before they get you" finishing opportunity but with slow attacks doing so little damage...
"* slowed units lose initiative on fight"
Again I can see a defensive bonus here but what about "first strike" units?
Obviously slow can't now be used as it was to weaken high damage low swing units - why was that change considered necessary? Was it over or underpowered?
Not complaints, just thoughts!
"* hitting with slow changes attack order to allow attacker to strike again"
Aren't most slow attacks ranged anyway? Who would want to slow a ranged unit to capitalize on such an advantage? I suppose it helps defensively but two free shots at the start is balanced by two free shots at the end...I can see how this is supposed to create a "get them before they get you" finishing opportunity but with slow attacks doing so little damage...
"* slowed units lose initiative on fight"
Again I can see a defensive bonus here but what about "first strike" units?
Obviously slow can't now be used as it was to weaken high damage low swing units - why was that change considered necessary? Was it over or underpowered?
Not complaints, just thoughts!
Signature dropped due to use of img tag
Re: Discussion of changes to slow?
it was mainly on IRCdibblethewrecker wrote:Did I miss it?
"* hitting with slow changes attack order to allow attacker to strike again"
Aren't most slow attacks ranged anyway? Who would want to slow a ranged unit to capitalize on such an advantage? I suppose it helps defensively but two free shots at the start is balanced by two free shots at the end...I can see how this is supposed to create a "get them before they get you" finishing opportunity but with slow attacks doing so little damage...
new slow is cumulative with first strike"* slowed units lose initiative on fight"
Again I can see a defensive bonus here but what about "first strike" units?
no, it can't. The reordering of attacks is actually quite interesting since it gives you an edge when you're trying to finish off a wounded ennemy (which is often the case for slowing units like the shaman)Obviously slow can't now be used as it was to weaken high damage low swing units - why was that change considered necessary? Was it over or underpowered?
the other usage of slow is more tacticall... slowed units have no ZoC, use it to reach escaping wounded units or help your wounded units get out of a ZoC loc.
it was not considered because the old ZoC was over or under powered, but because it was not interesting enough... slow was a cool side effect, but of little strategic use...
Fight key loggers: write some perl using vim
- Elvish_Pillager
- Posts: 8137
- Joined: May 28th, 2004, 10:21 am
- Location: Everywhere you think, nowhere you can possibly imagine.
- Contact:
-
- Retired Developer
- Posts: 2633
- Joined: March 22nd, 2004, 11:22 pm
- Location: An Earl's Roadstead
Ok, having missed the discussion on IRC, what does slow do currently? I take it that it is different then what got sent to the mailing list a few days ago.
"you can already do that with WML"
Fight Creeeping Biggerism!
http://www.wesnoth.org/forum/viewtopic. ... 760#131760
http://www.wesnoth.org/forum/viewtopic. ... 1358#11358
yes, we had a couple of coding/playtesting cycles with dragonking, this is the effect of slow in current trunk.
note that it's 1.1.x only and it's here for testing, it might have taticall effects we havn't forseen and might be changed again.
basically it's at the point where we think it is as balanced as we can see without the huge playtesting we get by including it
What hasn't changed
* slow divides ths movement by 2
* slow heals at the end of the slowed unit's next tur
what has changed
* slowed units don't emit a ZoC
* slowed units don't lose an attack anymore
* slow reorders the attacks in a fight
let me give a few examples of the reordering effect
A is the attacker
D is the defender
s means the last attack touched (only meaningfull with slowing attack)
both have 4 attacks
first, A is slowed, the order would be
DADADADA
D is slowed
AADADADD
A is slowed, D has first strike
DDADADAA
D is slowed, D has first strike
ADADADAD
and now when one of them has a slowing attack
A has slowing attack and lands his second and third blow
ADAsAsADDD
and so on
hopes this helpd
note that it's 1.1.x only and it's here for testing, it might have taticall effects we havn't forseen and might be changed again.
basically it's at the point where we think it is as balanced as we can see without the huge playtesting we get by including it
What hasn't changed
* slow divides ths movement by 2
* slow heals at the end of the slowed unit's next tur
what has changed
* slowed units don't emit a ZoC
* slowed units don't lose an attack anymore
* slow reorders the attacks in a fight
let me give a few examples of the reordering effect
A is the attacker
D is the defender
s means the last attack touched (only meaningfull with slowing attack)
both have 4 attacks
first, A is slowed, the order would be
DADADADA
D is slowed
AADADADD
A is slowed, D has first strike
DDADADAA
D is slowed, D has first strike
ADADADAD
and now when one of them has a slowing attack
A has slowing attack and lands his second and third blow
ADAsAsADDD
and so on
hopes this helpd
Fight key loggers: write some perl using vim
- Elvish_Pillager
- Posts: 8137
- Joined: May 28th, 2004, 10:21 am
- Location: Everywhere you think, nowhere you can possibly imagine.
- Contact:
How the heck is anyone going to remember all this? It would be ok if slow had one or two, effects, but three?!?
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.
-
- Retired Developer
- Posts: 2633
- Joined: March 22nd, 2004, 11:22 pm
- Location: An Earl's Roadstead
So is this what you mean by cumulative? A unit that is hit by 3 slowing attacks has its attacks delayed by three? So the next unit that attacks it (with 4 attacks) will getBoucman wrote: and now when one of them has a slowing attack
A has slowing attack and lands his second and third blow
ADAsAsADDD
and so on
hopes this helpd
AAADADDD
"you can already do that with WML"
Fight Creeeping Biggerism!
http://www.wesnoth.org/forum/viewtopic. ... 760#131760
http://www.wesnoth.org/forum/viewtopic. ... 1358#11358
Darth Fool wrote: So is this what you mean by cumulative? A unit that is hit by 3 slowing attacks has its attacks delayed by three? So the next unit that attacks it (with 4 attacks) will get
AAADADDD
hmm no
being hit by a slowing strike will delay your strike immediatly, whether the unit was slowed before or not
being slowed at the beggining of combat will delay your first attack
however the number of time you were slowed does not change anything.
when I said slow was cumulative, it was in a very precise context, I said it when discussing it's effect on units with initiative.
attacking a unit with unit while slowed will delay your attack twice
it's actually pretty simple once you get the underlying rules
Fight key loggers: write some perl using vim
-
- Retired Developer
- Posts: 2633
- Joined: March 22nd, 2004, 11:22 pm
- Location: An Earl's Roadstead
Ok, I got it now. I do think that the description of the effect needs to be worked on. Overall, EP might be right that it is too complicated, even if it is balanced. A short and accurate description is needed. Currently it seems there are four important effects:Boucman wrote:Darth Fool wrote: So is this what you mean by cumulative? A unit that is hit by 3 slowing attacks has its attacks delayed by three? So the next unit that attacks it (with 4 attacks) will get
AAADADDD
hmm no
being hit by a slowing strike will delay your strike immediatly, whether the unit was slowed before or not
being slowed at the beggining of combat will delay your first attack
however the number of time you were slowed does not change anything.
when I said slow was cumulative, it was in a very precise context, I said it when discussing it's effect on units with initiative.
attacking a unit with unit while slowed will delay your attack twice
it's actually pretty simple once you get the underlying rules
1) unit that is slowed has no ZOC
2) unit that is slowed has 1/2 movement
3) unit that is slowed at the beginning of combat has first attack delayed
4) unit hit by a slowed attack has the next attack delayed.
5) Slowed units heal at end of their next turn. Presumably this means if they are slowed on their attack, they must wait through a full cycle of turns before they are healed?
"you can already do that with WML"
Fight Creeeping Biggerism!
http://www.wesnoth.org/forum/viewtopic. ... 760#131760
http://www.wesnoth.org/forum/viewtopic. ... 1358#11358
- Dragonking
- Inactive Developer
- Posts: 591
- Joined: November 6th, 2004, 10:45 am
- Location: Poland
It is really simple.
Slowed unit always is slower - so while attacking, defender is faster than him(defender attacks first), and while slowed unit is defending, attacker does 2 first strikes in a row.
It is cumulative with firststrike.
Plus, if shaman/other unit wits slow hits, he can attack again immediately after every slow attack which hits (of cours if number of attacks allows him to do so).
Slowed unit always is slower - so while attacking, defender is faster than him(defender attacks first), and while slowed unit is defending, attacker does 2 first strikes in a row.
It is cumulative with firststrike.
Plus, if shaman/other unit wits slow hits, he can attack again immediately after every slow attack which hits (of cours if number of attacks allows him to do so).
This is a block of text that can be added to posts you make. There is a 255 character limit
- Elvish_Pillager
- Posts: 8137
- Joined: May 28th, 2004, 10:21 am
- Location: Everywhere you think, nowhere you can possibly imagine.
- Contact:
That takes a while to comprehend though, AND it removes ZoC AND reduces speed. It's hard enough to fully percieve the effects of Slow currently, which only has two, simple, effects...
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.
- Elvish_Pillager
- Posts: 8137
- Joined: May 28th, 2004, 10:21 am
- Location: Everywhere you think, nowhere you can possibly imagine.
- Contact:
The point is not that it's hard to understand - it's that among all the little potential happenings that the player has to analyze, it is near impossible to do subconciously. It's not that thinking can't make you understand, it's that you have to go out and think in the first place, while you should be spending your thoughs handling higher-level concepts.
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.
My big problem with slow, both as it stands and this proposal, is that its fiddling with opponent attacks is potentially super-powerful but has been balanced by making slowing attacks very wimpy. So it has the net effect of sometimes affecting the skirmish outcome hugely, but often being irrelevant.
If we keep in mind the basic principle of keeping things simple, I can categorically say that slow is extremely difficult to analyse. The basic damage model, with berserk, drain and firststrike, already constitutes a random walk in the positive quadrant which is more complicated than anything I have found in the research literature (for the current frontier, see for instance the recent papers by Mireille Bousquet-Mélou, Marko Petkovsek, and Stan Zachary). The only feasible way to analyse this basic model, excluding slow, is through expanding the state space as prkill does, or to use simulation as done by the AI code (100 simulated skirmishes).
What worries me here is that firststrike is a powerful effect -- it swops the attacker and defender percentages-to-kill in a skirmish. However, it applies always or not at all, so it is easy to work with. Slow's effect on opponent attacks is a similarly powerful effect, but because it only sometimes applies, 100 simulated skirmishes are going to have a large variance from the mean outcome. So it seems to me that the AI's actions will vary wildly when slowing is involved, since the outcome of the simulation is likely to be inaccurate. Note that the current slow has a similar problem, so I am not arguing for keeping the status quo.
I think we get enough flak as it is for the outcome of skirmishes being heavy-tailed, that perhaps we should reconsider this effect. Reducing the movement rate and removing ZOC are both interesting effects, and have strategic appeal, but regarding the effect on attacks, I would like to see some other way forward.
My proposal is that we should aim for an effect that can be explained fully in one short sentence. As it is, we have diagrams of slow vs firststrike to explain how it works, and I still don't understand how it works when the number of strikes of attacker and defender are unequal. I certainly also can't explain the old slow's effect on opponent attacks properly without several sentences.
Any ideas for a slow that is fast to explain?
If we keep in mind the basic principle of keeping things simple, I can categorically say that slow is extremely difficult to analyse. The basic damage model, with berserk, drain and firststrike, already constitutes a random walk in the positive quadrant which is more complicated than anything I have found in the research literature (for the current frontier, see for instance the recent papers by Mireille Bousquet-Mélou, Marko Petkovsek, and Stan Zachary). The only feasible way to analyse this basic model, excluding slow, is through expanding the state space as prkill does, or to use simulation as done by the AI code (100 simulated skirmishes).
What worries me here is that firststrike is a powerful effect -- it swops the attacker and defender percentages-to-kill in a skirmish. However, it applies always or not at all, so it is easy to work with. Slow's effect on opponent attacks is a similarly powerful effect, but because it only sometimes applies, 100 simulated skirmishes are going to have a large variance from the mean outcome. So it seems to me that the AI's actions will vary wildly when slowing is involved, since the outcome of the simulation is likely to be inaccurate. Note that the current slow has a similar problem, so I am not arguing for keeping the status quo.
I think we get enough flak as it is for the outcome of skirmishes being heavy-tailed, that perhaps we should reconsider this effect. Reducing the movement rate and removing ZOC are both interesting effects, and have strategic appeal, but regarding the effect on attacks, I would like to see some other way forward.
My proposal is that we should aim for an effect that can be explained fully in one short sentence. As it is, we have diagrams of slow vs firststrike to explain how it works, and I still don't understand how it works when the number of strikes of attacker and defender are unequal. I certainly also can't explain the old slow's effect on opponent attacks properly without several sentences.
Any ideas for a slow that is fast to explain?
This quote is not attributable to Antoine de Saint-Exupéry.