BfW is a game that relies on save-scumming too much

General feedback and discussion of the game.

Moderators: Forum Moderators, Developers

Post Reply
gnombat
Posts: 207
Joined: June 10th, 2010, 8:49 pm
Contact:

Re: BfW is a game that relies on save-scumming too much

Post by gnombat » April 2nd, 2019, 11:21 pm

Konrad2 wrote:
April 2nd, 2019, 11:01 pm
That's true, but I wonder why you quote me while saying that? :hmm:
Typically in difficult scenarios there is only going to be one way to win (or at least, a narrow range of strategies/tactics needed to win). The challenge is finding it.

It would be unusual (though perhaps not impossible) to find a scenario where there are multiple completely different ways to play which are equally likely to succeed.

User avatar
Pentarctagon
Forum Administrator
Posts: 3911
Joined: March 22nd, 2009, 10:50 pm
Location: Earth (occasionally)

Re: BfW is a game that relies on save-scumming too much

Post by Pentarctagon » April 3rd, 2019, 12:52 am

As far as increasing the number of strikes goes, how would that work with the Poison and Slows weapon specials?
99 little bugs in the code, 99 little bugs
take one down, patch it around
-2,147,483,648 little bugs in the code

User avatar
josteph
Developer
Posts: 613
Joined: August 19th, 2017, 6:58 pm

Re: BfW is a game that relies on save-scumming too much

Post by josteph » April 3rd, 2019, 12:15 pm

beetlenaut wrote:
April 1st, 2019, 10:11 pm
josteph wrote:
April 1st, 2019, 9:16 pm
Maybe add a scene to the tutorial where we have the player do an attack, force that attack to miss 4/4, and then the narrator explains that that happens sometimes and campaigns generally don't expect you to save/load when it happens
This message is a really good idea. However, I bet it's rare that someone saveloads when a fighter or archer misses 4 out of 4 times. More often it seems to be when an important unit dies. Maybe we should do it the first time a unit dies in the tutorial. We could even force that if it hasn't happened yet and the enemy is down to just a couple units. (The player would be winning handily in that case.)
Opened #4010, let's continue the discussion there.
Yomar wrote:
April 2nd, 2019, 9:37 pm
I think that someone already said, that no one is forced to save and load, if some campaign developers have this concern, they could implement a save limit or autodelete feature if possible.
Campaign developers can't delete savefiles (or any other files).

User avatar
Ravana
Moderator
Posts: 2115
Joined: January 29th, 2012, 12:49 am
Location: Estonia
Contact:

Re: BfW is a game that relies on save-scumming too much

Post by Ravana » April 3rd, 2019, 3:40 pm

It is possible to count how many times current game has been saved, and it is possible to invalidate save files that would be created.

User avatar
Aldarisvet
Translator
Posts: 736
Joined: February 23rd, 2015, 2:39 pm
Location: Moscow, Russia

Re: BfW is a game that relies on save-scumming too much

Post by Aldarisvet » April 3rd, 2019, 4:40 pm

Pentarctagon wrote:
April 3rd, 2019, 12:52 am
As far as increasing the number of strikes goes, how would that work with the Poison and Slows weapon specials?
Yea, that is a problem but can be solved in that way: if a damage is higher than average, then Poison and Slows works, else not.
Konrad2 wrote:
April 2nd, 2019, 9:49 pm
It seems to boil down to you wanting every single player that plays your campaign to have no other choice but to play it in one specific way.
Or putting it in different words, noone should be allowed to play it unless they follow your specific design. Is that about right?
Actually I care not about a specific way for a player to win but the fact that this tactics would be original, unusual, interesting.
If someone would win my scenario with an interesting way, I would be only happy.

Can you find a difference between a plain way to make a checkmate and a beautiful way? That is about it.
Yes, I want to force people to think. Many was not able to guess that in the defensive scenario with lvl1 dwarves against lvl2 trolls they should recruit Guardsmen because they have a steady ability. Nope, they would always recruit dwarvish fighters instead and use save-scumming to get Dwarvish Lords.
facebook.com/wesnothian/ - everyday something new about Wesnoth
My campaign:A Whim of Fate, also see Zombies:Introduction single map campaign
Art thread:Mostly frankenstains

Spirit_of_Currents
Posts: 75
Joined: April 26th, 2014, 4:44 pm

Re: BfW is a game that relies on save-scumming too much

Post by Spirit_of_Currents » April 6th, 2019, 1:04 pm

Aldarisvet wrote:
April 1st, 2019, 9:03 am
The Idea is to address an issue mentioned by Xalzar. Just to increase the number of dice rolls.
Each unit's strike must be splitted into 10 dices. It will stay visually as a single strike but it outcomes must be split from 2 outcomes (missed or hit) to 11 outcomes from no damage to full damage.
I suggest rounding in following way: if the ten dice roll 1.2, it has 80% chance to become 1 and 20% chance to become 2.

I really want this modification add-on if it doesn't change chance to poison or slow.
There are very much electrical currents in my brain.

Tad_Carlucci
Developer
Posts: 393
Joined: April 24th, 2016, 4:18 pm

Re: BfW is a game that relies on save-scumming too much

Post by Tad_Carlucci » April 7th, 2019, 1:51 pm

I don't understand. At present, if we have a 70% chance to hit, we "roll" a single "die" by generating a random number in the range zero (inclusive) to 1 (exclusive). If the result is less than 0.7, we score it a hit, otherwise we score a miss.

It sounds like you're suggesting we generate several random numbers and combine them.

This won't work.

We would need to rebalance everything. To keep the game the same "dificulty", in fact, we'd have to adjust the chance to hit downward, probably quite a bit. I've not done the math, but I'd guess to maintain a 70% actual chance to hit, we'd have to use something more like 52% or 53%. And, then, the number we're using is not "chance to hit" any more. It's the number dividing the area under a Bell curve at a 70%/30% split. That means every UMC author would need to solve an Integral equation rather than simply state "70% chance to hit". (We could make their job easier by pre-calculating the values, if we use a fixed number of samples.)
I forked real life and now I'm getting merge conflicts.

Byron
Posts: 34
Joined: May 23rd, 2013, 7:41 am
Location: West Virginia

Re: BfW is a game that relies on save-scumming too much

Post by Byron » April 8th, 2019, 2:57 am

I know, when the AI misses completely with a Mage, have it say, "That's not fair!" to make it realistic. I mean, what do we do, when the AI's "luck" is really bad?

Sorry, I just think this conversation is less than useful. When they changed the game of basketball, I stopped watching. When they changed chess, I stopped playing. If you have a problem with this game, then go away & play another, but leave this one alone.

Spirit_of_Currents
Posts: 75
Joined: April 26th, 2014, 4:44 pm

Re: BfW is a game that relies on save-scumming too much

Post by Spirit_of_Currents » April 8th, 2019, 8:54 am

The ten-dice-system would belong to an optional add-on.

There are already modifications that could make some scenarios seriously unbalanced, such as AMLA Mod.

Using the ten-dice-system would be a good way to hone one's mental calculation skills.
There are very much electrical currents in my brain.

User avatar
Aldarisvet
Translator
Posts: 736
Joined: February 23rd, 2015, 2:39 pm
Location: Moscow, Russia

Re: BfW is a game that relies on save-scumming too much

Post by Aldarisvet » April 9th, 2019, 11:21 am

Tad_Carlucci wrote:
April 7th, 2019, 1:51 pm
I don't understand. At present, if we have a 70% chance to hit, we "roll" a single "die" by generating a random number in the range zero (inclusive) to 1 (exclusive). If the result is less than 0.7, we score it a hit, otherwise we score a miss.

It sounds like you're suggesting we generate several random numbers and combine them.

This won't work.

We would need to rebalance everything. To keep the game the same "dificulty", in fact, we'd have to adjust the chance to hit downward, probably quite a bit. I've not done the math, but I'd guess to maintain a 70% actual chance to hit, we'd have to use something more like 52% or 53%. And, then, the number we're using is not "chance to hit" any more. It's the number dividing the area under a Bell curve at a 70%/30% split. That means every UMC author would need to solve an Integral equation rather than simply state "70% chance to hit". (We could make their job easier by pre-calculating the values, if we use a fixed number of samples.)
I also don't understand too what you mean.
Nothing need to be rebalanced only because of reducing deviations.
If you make an elvish fighter 0.5-40 (or less radical 2-10) instead of 5-4 (in calculations, not in a visual side), this would affect a difficulty how? The average will stay the same.
The difficulty depends on the initial number of enemy units/gold and enemy income.
Actually normalizing volatility of results will make all campaigns harder. Why so? Because battles results can deviate not only to harder outcomes but also to easiest outcomes. And I assure you that 99% players effectively use these deviations to get advantages with save and loads. If my 11-year son, for example, will reach a wounded 3HP enemy unit with a mage and attack him with that mage, and a mage misses all three times, he would get angry and reload it for sure. The probability of such outcome will be 0.3*0.3*0.3=2,7% (still many of such cases through the scenario shift the result to player's benefit from the average). In the system I suggest the probability of attacked unit's survival would be quite lower. Moreover, it would be difference if the attacked unit have 1hp left or 4hp left, in the second case he will have significally more chances to survive however still very low.
__________________________________

More thoughts. What is the feature of current RNG system? How it affects on the style of playing? I can say from my experience. It force a honest/experienced player to play very carefully because he understands that with that system he can always get a critical unluck. So a honest player always must see if there is any possibility of heavy losses (losing a hero/loyal unit/valuable high-level unit) and avoid that risk. And a common player just reloads in case of that critical unluck. Is it something of value? Is it a fun to be always over cautious? Obviously no (however in the real war you must be over cautious exactly because you always need to avoid a critical unluck or you will be killed relatively soon, but the real life is a boring thing that demands huge perseverance and patience, I doubt that players seek these while they go for games).

From the other side, a great instability creates a very poor ability to calculate future events in the short term. You cannot attack effectively except keeping the ranks together. And AI is too stupid to keep the ranks together. That creates quite a boring gameplay. Reducing RNG deviations would let a honest player to use a more aggressive style and will create a more interesting gameplay (and a common player in fact always uses an aggressive style, he just reloads in the case of it's failure).

...Surprisingly for myself I discovered that a boring gameplay in Wesnoth and a RNG system are connected.
facebook.com/wesnothian/ - everyday something new about Wesnoth
My campaign:A Whim of Fate, also see Zombies:Introduction single map campaign
Art thread:Mostly frankenstains

Caladbolg
Posts: 195
Joined: January 1st, 2016, 4:40 pm
Location: Hopelessly trapped within the Submachine

Re: BfW is a game that relies on save-scumming too much

Post by Caladbolg » April 9th, 2019, 4:01 pm

To get back to your Dragonguard example with 1-40 and 70% defence, with the standard mechanics, your chance to deal 0 damage is 70%, and cth is 30%.
With your proposal, you are bracketing the potential damage into 11 brackets, 10 of which deal at least some damage, while only the lowest one deals no damage. So your chance to deal 0 damage is 0.7^10 = 2.82%, and cth is 97.18%. So you lose the very intuitive idea of "70% defence means 30% chance to hit on an individual swing". With your proposal, low hp units would have almost no chance of surviving any encounter. This alone would require extensive rebalancing of the whole game.

Also, there would be very little chance of doing full damage, which might remove some of the frustration of dealing with e.g. Dragonguards, but it would also make such units less flavorful, and make units in general more bland imo.
Aldarisvet wrote:
April 9th, 2019, 11:21 am
And I assure you that 99% players effectively use these deviations to get advantages with save and loads. If my 11-year son, for example, will reach a wounded 3HP enemy unit with a mage and attack him with that mage, and a mage misses all three times, he would get angry and reload it for sure.
...
And a common player just reloads in case of that critical unluck.
...
and a common player in fact always uses an aggressive style, he just reloads in the case of it's failure
Where are you pulling that 99% statistic from? You continue hammering on the point that the average player extensively uses save-loading to avoid bad rng, but I see no reason to believe that is the case. And I don't think your 11-year-old son is a good representative of the average player, nor would the two of you make up a sufficiently large sample to draw conclusions from even if he were.

User avatar
Xalzar
Posts: 284
Joined: April 4th, 2009, 10:03 pm
Location: New Saurgrath

Re: BfW is a game that relies on save-scumming too much

Post by Xalzar » April 9th, 2019, 7:30 pm

Aldarisvet wrote:
April 1st, 2019, 9:03 am
The Idea is to address an issue mentioned by Xalzar. Just to increase the number of dice rolls.
Each unit's strike must be splitted into 10 dices. It will stay visually as a single strike but it outcomes must be split from 2 outcomes (missed or hit) to 11 outcomes from no damage to full damage.
That doesn't address my issue. My concept was totally different: the duality damage/no damage remains, but every single strike rolls more than one dice. This way probability is conserved but more dice* throws garantee better the expected results (e.g. an attack with 80% cth will hit more likely 80% of the time instead of missing every single time).

*I mean not standard dice, but 0-1 (hit/miss) or 0-9

Example:
Elvish Archer
Bow 5-4
Cth 60%
Every attack some dice are thrown, results:

a) 0-9 dice x5 dice throws/strike
0 required to hit with 100% cth (dice throws not necessary)
1-5 90%
6-10 80%
13-15 70%
16-20 60%
21-25 50%
26-30 40%
31-35 30%
36-40 20%
41-45 10%
To hit the sum of the results (for every singular attack) needs to be above certain values, in this case a result between 16 and 45.

Example of results: 14 / 11 / 25 / 42 -> miss / miss / hit / hit

or:

b) 0-1 dice x9 dice throws/strike
0 100%
1 90%
2 80%
3 70%
4 60%
5 50%
6 40%
7 30%
8 20%
9 10%
To hit the sum of the results (for every singular attack) needs to be above a certain value, in this case above 4.

Example of results: 4 / 1 / 3 / 7 -> hit/ miss / miss/ hit

The solution proposed by Aldarisvet is cool but indeed changes the balance of the game. A nearly-dead target unit at 1 hp for example, is pratically garanteed to die even to a unit with one strike which has 50% cth, instead of having 50% chance.

User avatar
Aldarisvet
Translator
Posts: 736
Joined: February 23rd, 2015, 2:39 pm
Location: Moscow, Russia

Re: BfW is a game that relies on save-scumming too much

Post by Aldarisvet » April 9th, 2019, 7:43 pm

Well, I wanted to argue and insist that nothing should be rebalanced. A case in which a dragonguard attacks a bat, for example, is quite rare, lvl3 attacks lvl0. Yes, in that case would be a change in the average outcome. Yes, a dragonguard would not be dragonguard anymore but that only for good.
But instead I got an idea that we not necessarily have to use a normal distribution. In the normal distribution all probabilities are concentrated in the middle. But it must be possible to create a plain distribution where a probability to make a zero damage by a dragonguard would be equal to probability of doing a half damage. Dices create a normal distribution but we have no need to stick with it.
Xalzar wrote:
April 9th, 2019, 7:30 pm
The solution proposed by Aldarisvet is cool but indeed changes the balance of the game. A nearly-dead target unit at 1 hp for example, is pratically garanteed to die even to a unit with one strike which has 50% cth, instead of having 50% chance.
Come on! Reducing deviations would not shift the balance!

Do you have any 1hp units to recruit in the game? If yes, then it would shift it. Because a dragonguard attacking 1 hp footpad on the hill in the current system of course is a waste. And a cattlefish will kill that 1 hp footpad easely.

But the fact is that 1hp wounded unit appears not from nowhere, it is a result of the previous battle. Think about it. The fact is that in the current system you have much more chances to kill a lvl1 shaman (26 hp) in the forest from the first strike (30% probability) with lvl2 thunderguard (28 damage) than in case with less deviations. So it is a wrong statement that weaker units would be easeier to kill in a new system.

Given in multiplayer we have units of the same level fighting each other and the damage a single unit can inflict in one fight is less than an average HP of these units, there would be no shift in balance at all.
Last edited by Aldarisvet on April 9th, 2019, 8:55 pm, edited 5 times in total.
facebook.com/wesnothian/ - everyday something new about Wesnoth
My campaign:A Whim of Fate, also see Zombies:Introduction single map campaign
Art thread:Mostly frankenstains

User avatar
Xalzar
Posts: 284
Joined: April 4th, 2009, 10:03 pm
Location: New Saurgrath

Re: BfW is a game that relies on save-scumming too much

Post by Xalzar » April 9th, 2019, 8:10 pm

Aldarisvet wrote:
April 9th, 2019, 7:43 pm
Well, I wanted to argue and insist that nothing should be rebalanced. A case in which a dragonguard attacks a bat, for example, is quite rare, lvl3 attacks lvl0. Yes, in that case would be a change in the average outcome. Yes, a dragonguard would not be dragonguard anymore but that only for good.
But instead I got an idea that we not necessarily have to use a normal distribution. In the normal distribution all probabilities are concentrated in the middle. But it must be possible to create a plain distribution where a probability to make a zero damage by a dragonguard would be equal to probability of doing half damage. Dices creates a normal distribution but we have no need to stick with it.
You mean like three values like miss/half damage/full damage? Could be interesting, I'd play that.
Or do you mean that a 6 damage attack has for example:
0 damage (miss) 20 %
1 damage 10%
2 damage 10%
3 damage (half) 20%
4 damage 10%
5 damage 10%
6 damage (full) 20%
Also interesting, not what everyone want though. Still I'd be curious to try it too.

User avatar
Aldarisvet
Translator
Posts: 736
Joined: February 23rd, 2015, 2:39 pm
Location: Moscow, Russia

Re: BfW is a game that relies on save-scumming too much

Post by Aldarisvet » April 9th, 2019, 8:18 pm

Xalzar wrote:
April 9th, 2019, 8:10 pm
0 damage (miss) 20 %
1 damage 10%
2 damage 10%
3 damage (half) 20%
4 damage 10%
5 damage 10%
6 damage (full) 20%
Also interesting, not what everyone want though. Still I'd be curious to try it too.
This is not a plain distribution. Why 20% at the middle and the ends of outcomes and 10% in others? Must be 100%/6 everywhere. The problem is correcting this accounting a defence level.
Look at the picture. We need a line that colored in black. Do not pay attention on these pi symbols in the x axis, it just the first picture i found in the internet. Instead must be a values from a zero damage to a full damage.

A cuttlefish attacking 3*10 attacks with a normal distibution. 10 stikes with either 0 or 3. I am speaking about spreading each strike's damage to some random little portions (outcomes from 0 to 3), and not with a normal distribution. Of course cuttlefish has already low deviations because of 10 dices.
Xalzar wrote:
April 9th, 2019, 7:30 pm
the duality damage/no damage remains, but every single strike rolls more than one dice.
Than it would change nothing. There is no "single dice" actually. For now every strike is generated as a random number between 0 and 1 and the result is compared with a defence level. If a defence is 70%, then the number must be more than 0.7 to hit. Then the number of a damage per strike multiplies with the resistance to get a final result. I do not know exactly but I am sure that it is, because it is obvious.
Attachments
1200px-WrappedNormalPDF.png
facebook.com/wesnothian/ - everyday something new about Wesnoth
My campaign:A Whim of Fate, also see Zombies:Introduction single map campaign
Art thread:Mostly frankenstains

Post Reply