Less random game
Moderator: Forum Moderators
We've rehashed this argument several times. I have yet to see anyone post any counterarguments to the detailed analysis at http://www.wesnoth.org/forum/viewtopic. ... 1620#91620 which Dave summarizes in a neat and straightforward way above.
Dave designed this game so that terrain has a critical influence on skirmishes. Increasing the number of attacks would make the distribution of damage closer to normally distributed. This would reduce the essential strategic imbalance of units fighting on different terrain, an imbalance that is expressed by the binomially distributed number of attacks, where terrain defence determines the shape of this distribution. Making everything normally distributed may make some of you grade-curve obsessed folks happy, but personally I like the binomial distribution.
Dave designed this game so that terrain has a critical influence on skirmishes. Increasing the number of attacks would make the distribution of damage closer to normally distributed. This would reduce the essential strategic imbalance of units fighting on different terrain, an imbalance that is expressed by the binomially distributed number of attacks, where terrain defence determines the shape of this distribution. Making everything normally distributed may make some of you grade-curve obsessed folks happy, but personally I like the binomial distribution.
This quote is not attributable to Antoine de Saint-Exupéry.
-
Casual User
- Posts: 475
- Joined: March 11th, 2005, 5:05 pm
Good afternoon!
Ott wrote : "I have yet to see anyone post any counterarguments to the detailed analysis"
Okay, I'll have a go at it :
Since the average damage done would still be the same (at x% chance to hit, doing 5-4 or 5-8 with the ennemy having doubled hp is the same amount of damage), the impact of terrain would still be roughly the same, i.e. units would still get the same amount of relative damage. Hiding a low-hp unit in defensive terrain would be no safer or less safe. The only thing that really would see its importance on the game changed would be luck. The distribution would still be a binomial, and its shape would still be dictated by terrain. Luck would still play a (reduced) part in it, and terrain would play an equal one. The only thing that would change would be 1-3 hp left situations, and while they happen, they aren't that frequent. Besides, with doubled hp, 1-3 hp situations might become very infrequent while situations where you need two hits might become more frequent.
Ultimately, all this would do would be to reduce the incidence of extraordinary events, not the importance of terrain. Right now, you do feel the difference between forest and grassland, but also feel it is small on a larger scale. Skirmishes small enough for you to feel the difference is huge are very rare, and they are strictly a question of chance. Four hits in a row against an opponent in forest can happen just as well as four misses in grassland, and results balance each other out.
Just as you can argue it can reduce the importance of terrain because having a higher to-hit chance becomes less important in some cases, I can argue it increases the importance of terrain by preventing luck from nullyfying differences in defence. Terrain remains as useful overall.
So the question isn't how important terrain is in skirmishing, but how important luck is in skirmishing, and there, the disscussion is open.
One can argue it adds risk assessment tactics, while others can argue it nullyfies strategy in favor of mass attacks. I don't know what the best way is, but I can say that the current deviation seems excessive to me. Maybe not to you, but to me (and obviously a sizeable portion of Wesnoth players), it does. This justifies a mod. I'm not saying it would be better, but I am saying it is worth trying.
Ultimately, don't judge until you try both alternatives. Wesnoth being an open-source game, one can make and try 'DeterminNoth' and see which is best. It all comes down to the 'chess to dice' ratio.
A limited version of this could be made easily. By steering off slowing, healing or poisoning units, we could make some 'DeterminNoth' by screwing around only with the unit files and the traits files. I say once 1.0 (final) is out, let's try and give everyone the possibility to choose...
P.S. Coders would obviously be those who like the idea.
P.P.S. Poisoning and slowing attacks would reuire more refined balancing.
Ott wrote : "I have yet to see anyone post any counterarguments to the detailed analysis"
Okay, I'll have a go at it :
Since the average damage done would still be the same (at x% chance to hit, doing 5-4 or 5-8 with the ennemy having doubled hp is the same amount of damage), the impact of terrain would still be roughly the same, i.e. units would still get the same amount of relative damage. Hiding a low-hp unit in defensive terrain would be no safer or less safe. The only thing that really would see its importance on the game changed would be luck. The distribution would still be a binomial, and its shape would still be dictated by terrain. Luck would still play a (reduced) part in it, and terrain would play an equal one. The only thing that would change would be 1-3 hp left situations, and while they happen, they aren't that frequent. Besides, with doubled hp, 1-3 hp situations might become very infrequent while situations where you need two hits might become more frequent.
Ultimately, all this would do would be to reduce the incidence of extraordinary events, not the importance of terrain. Right now, you do feel the difference between forest and grassland, but also feel it is small on a larger scale. Skirmishes small enough for you to feel the difference is huge are very rare, and they are strictly a question of chance. Four hits in a row against an opponent in forest can happen just as well as four misses in grassland, and results balance each other out.
Just as you can argue it can reduce the importance of terrain because having a higher to-hit chance becomes less important in some cases, I can argue it increases the importance of terrain by preventing luck from nullyfying differences in defence. Terrain remains as useful overall.
So the question isn't how important terrain is in skirmishing, but how important luck is in skirmishing, and there, the disscussion is open.
One can argue it adds risk assessment tactics, while others can argue it nullyfies strategy in favor of mass attacks. I don't know what the best way is, but I can say that the current deviation seems excessive to me. Maybe not to you, but to me (and obviously a sizeable portion of Wesnoth players), it does. This justifies a mod. I'm not saying it would be better, but I am saying it is worth trying.
Ultimately, don't judge until you try both alternatives. Wesnoth being an open-source game, one can make and try 'DeterminNoth' and see which is best. It all comes down to the 'chess to dice' ratio.
A limited version of this could be made easily. By steering off slowing, healing or poisoning units, we could make some 'DeterminNoth' by screwing around only with the unit files and the traits files. I say once 1.0 (final) is out, let's try and give everyone the possibility to choose...
P.S. Coders would obviously be those who like the idea.
P.P.S. Poisoning and slowing attacks would reuire more refined balancing.
-
Elvish_Pillager
- Posts: 8137
- Joined: May 28th, 2004, 10:21 am
- Location: Everywhere you think, nowhere you can possibly imagine.
- Contact:
It all boils down to the assumption that these are equivalent, which is incorrect. For a fairly normal example to illustrate:Casual User wrote:the impact of terrain would still be roughly the same, i.e. units would still get the same amount of relative damage.
A Dark Adept, with 28 HP, sits on grassland or a hill or a mountain. At Dawn, he is menaced by two Spearmen.
Current scheme: 6 swings, 4 hits to be killed.
Grass: 54% chance to be killed.[1]
Hill: 34% chance to be killed.
Mountains: 18% chance to be killed.
Other scheme: 12 swings, 8 hits to be killed.
Grass: 44% chance to be killed.
Hill: 19% chance to be killed.
Mountains: 6% chance to be killed.
As you can see, if chance to kill is your object, the different system is hugely different. Instead of becoming 1/3 as likely to be killed by moving from grass to mountains, you become 1/7 as likely to be killed!
You can't just remove luck without changing a lot of game dynamics.
[1]percentages are from Wesnoth damage calculations, using a single spearman with twice (or four times) the number of attacks.
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.
I personally think that the big problem is that some people don't "trust" a computer to calculate random outcomes, rather than things being too random. If the game was played on a board using dice with the same amount of randomness, I don't think nearly as many people would complain.
David
David
“At Gambling, the deadly sin is to mistake bad play for bad luck.” -- Ian Fleming
I agree with the statement that some don't trust the computer to calculate the random outcomes. I happen to be one of them. My subjective feeling is that the computer seems to cheat though I know it probably does not. However, I do think that the game has been carefully balanced and do not believe that the proposed changes would be beneficial.Dave wrote:I personally think that the big problem is that some people don't "trust" a computer to calculate random outcomes, rather than things being too random. If the game was played on a board using dice with the same amount of randomness, I don't think nearly as many people would complain.
David
Uhh...you do realize that this game is Open Source, right? Anyone competent in C++ (of which there are many people) can look at the source code and see for themselves that there is no cheating going on....BillG3 wrote:My subjective feeling is that the computer seems to cheat though I know it probably does not.
David
“At Gambling, the deadly sin is to mistake bad play for bad luck.” -- Ian Fleming
-
Casual User
- Posts: 475
- Joined: March 11th, 2005, 5:05 pm
Good afternoon!
Actually, the discussion has nothing to do with trusting the computer or not. When playing a big battle, you eventually find the odds to be fairly even.
The discussion has to do with deviation from the norm. I personnally think it to be excessive. As I said, in big battles, it's no problem. In small skirmishes with ~5 units on both sides, which happen quite often in campaigns, the deviation makes strategy almost moot (beyond some very basic stuff).
I honestly think this idea is one that should be tried. It requires little scripting and might truly improve the game. I'm not saying it will, but I'm saying it deserves being tried.
P.S. Elvish Pillager, I never said it wouldn't change the game, I only said the changes would be balanced. By the way, did you just make my point for me that the new system would actually increase the importance of terrain, and hence good planning and army deployment, at the expense of luck?
Actually, the discussion has nothing to do with trusting the computer or not. When playing a big battle, you eventually find the odds to be fairly even.
The discussion has to do with deviation from the norm. I personnally think it to be excessive. As I said, in big battles, it's no problem. In small skirmishes with ~5 units on both sides, which happen quite often in campaigns, the deviation makes strategy almost moot (beyond some very basic stuff).
I honestly think this idea is one that should be tried. It requires little scripting and might truly improve the game. I'm not saying it will, but I'm saying it deserves being tried.
P.S. Elvish Pillager, I never said it wouldn't change the game, I only said the changes would be balanced. By the way, did you just make my point for me that the new system would actually increase the importance of terrain, and hence good planning and army deployment, at the expense of luck?
For you, possibly, or possibly not.Casual User wrote: Actually, the discussion has nothing to do with trusting the computer or not.
But how can you speak for all users? I have seen many users more or less admit that they don't really trust the computer to generate fair random numbers.
David
“At Gambling, the deadly sin is to mistake bad play for bad luck.” -- Ian Fleming
@Elvish Pillager: that was the whole purpose of this change, it reduces chance of 'unfair' (lucky) kills, the relative average damage will be the same but variation will decrease, that's why chance for killing unit decrease.
And about dynamics - i think i won't change it that much, don't forget that now you got higher chance for killing unit but also higher chance for not doing any ( or just little ) damage.
Computer cheat? Yes i got that feeling too, which i know is impossible but still. Imho the main problem are campaign scenarios, there are some reasons why:
- players don't like lose to a computer, they look for a excuse so they think that AI cheats, they also tend to remember bad luck situations rather that good ones
- It's partly a RPG game, you level your units, play with them in every scenario and don't want to lose them
- It's very important to win every scenario in best possible way, as it will affect later scenarios
About problems with slow - this change was sugested as a way for getting less random game without code modification ( not every player is able to compile Wesnoth ) ( i just keep two data folders and change them when i want to play normal or moded game ). But if changes in code are needed it's better to keep current number of attacks and change chance to hit to 100% and apply terrain defence to resistance. ( it may require some aditional work though )
PS. the other solution to 'randomness problem' without editing the code is to mess with terrain defence and make chance to hit between 80%-100% in most cases, but HP needs to be increased as well.
And about dynamics - i think i won't change it that much, don't forget that now you got higher chance for killing unit but also higher chance for not doing any ( or just little ) damage.
Computer cheat? Yes i got that feeling too, which i know is impossible but still. Imho the main problem are campaign scenarios, there are some reasons why:
- players don't like lose to a computer, they look for a excuse so they think that AI cheats, they also tend to remember bad luck situations rather that good ones
- It's partly a RPG game, you level your units, play with them in every scenario and don't want to lose them
- It's very important to win every scenario in best possible way, as it will affect later scenarios
About problems with slow - this change was sugested as a way for getting less random game without code modification ( not every player is able to compile Wesnoth ) ( i just keep two data folders and change them when i want to play normal or moded game ). But if changes in code are needed it's better to keep current number of attacks and change chance to hit to 100% and apply terrain defence to resistance. ( it may require some aditional work though )
PS. the other solution to 'randomness problem' without editing the code is to mess with terrain defence and make chance to hit between 80%-100% in most cases, but HP needs to be increased as well.
-
Elvish_Pillager
- Posts: 8137
- Joined: May 28th, 2004, 10:21 am
- Location: Everywhere you think, nowhere you can possibly imagine.
- Contact:
Well, what do you define as a lucky kill? Even when the Spearmen had a 54% chance to kill, at which point you can hardly say they're unfair kills if they happen, your system still reduced the chances. It reduces the chances of all lucky kills, not just unfairly lucky ones, and almost all kills that matter in Wesnoth are lucky.Ambegen wrote:@Elvish Pillager: that was the whole purpose of this change, it reduces chance of 'unfair' (lucky) kills, the relative average damage will be the same but variation will decrease, that's why chance for killing unit decrease.
You are so wrong! Killing is *THE* combat effect. Everything in the game boils down to one of two goals: Capturing villages, or killing units. And villages aren't very effective. Almost every complex tactical choice will have a different best choice if the chances to kill are reduced! Attrition will rule and terrain will be way too strong. Mages will be overpowered and Horsemen will be underpowered. Healers will be useless and Skirmishers weak, and Drain will be even stronger than it is now.Ambegen wrote:And about dynamics - i think i won't change it that much, don't forget that now you got higher chance for killing unit but also higher chance for not doing any ( or just little ) damage.
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.
you still don't get it
this sugestion reduces luck factor, which means units will more often do average damage
now when damage distibutions is binomial there is much more chances for more extreme situations
yes it will reduces chances to kill , but you forgeting one thing, what will happen when you not kill the unit ? binomial distribution causes that chances for none or minimal damage are also high, which means you might still be unable to kill unit when you add another attacker to calculation, while in normal distribution at some point kill is almost guaranted.
So for example, chnaces for killing enemy with 2 units will be lower, with 3 similar and with 4 greater.
this sugestion reduces luck factor, which means units will more often do average damage
now when damage distibutions is binomial there is much more chances for more extreme situations
yes it will reduces chances to kill , but you forgeting one thing, what will happen when you not kill the unit ? binomial distribution causes that chances for none or minimal damage are also high, which means you might still be unable to kill unit when you add another attacker to calculation, while in normal distribution at some point kill is almost guaranted.
So for example, chnaces for killing enemy with 2 units will be lower, with 3 similar and with 4 greater.
-
Elvish_Pillager
- Posts: 8137
- Joined: May 28th, 2004, 10:21 am
- Location: Everywhere you think, nowhere you can possibly imagine.
- Contact:
That doesn't help the problem significantly!Ambegen wrote:So for example, chnaces for killing enemy with 2 units will be lower, with 3 similar and with 4 greater.
If three units face three enemy units, by the current scheme, the first side to get a kill is generally the winner, because two units are not enough to get a good chance at a kill.
With your scheme, it becomes less likely for one side's unit to be killed right away, but even more devastating when it DOES get unlucky and die right away, because the remaining two are even less likely to get a kill than they are now.
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.
As has been said many times before, if you reduce bad luck, you also reduce good luck. What occurs to me is that perhaps the reason this debate keeps reappearing (and really it should me marked OFW) is that some people are optimists and some are pessimists. Optimists like the luck factor, because, hey, you might get lucky. Pessimists dislike it because you might get unlucky. Currently Wesnoth is more geared towards the optimists, and I like that because I find the company of cheerful optimists far more pleasant than the company of gloomy pessimists.
"When a man is tired of Ankh-Morpork, he is tired of ankle-deep slurry" -- Catroaster
Legal, free live music: Surf Coasters at Double Down Saloon, Las Vegas on 2005-03-06. Tight, high-energy Japanese Surf-Rock.
Legal, free live music: Surf Coasters at Double Down Saloon, Las Vegas on 2005-03-06. Tight, high-energy Japanese Surf-Rock.