Nope.Aloo wrote:Is there a place in PzC where I can check the statistics of damage caused vs expected and damage taken vs expected (similar to stats function in Wesnoth) for a scenario or campaing?
RNG (That's Random Number Generator)
Moderators: Slitherine Core, Panzer Corps Moderators, Panzer Corps Design
-
Aloo
- Sergeant - 7.5 cm FK 16 nA
- Posts: 217
- Joined: Mon Oct 31, 2011 12:38 pm
- Location: Poland
- Contact:
I see a lot of new topics coming up about the randomness favoring the AI. This is typical for games using RNG: people always remember bad results and quickly forget the good ones, this is especially true if they loose a unit.
The solution is something I asked about a few posts earlier - giving the player an easy to check comparison of the predicted vs actual battle results for the whole campaign. Then usually everything balances out after 2-3 scenarios and most "the AI has better rolls" arguments die out.
Im not sure how hard this is to do but if this isn't much work it would help solve this problem. The other benefit is in MP you can check how much of your victory/defeat was luck and how much skill.
The solution is something I asked about a few posts earlier - giving the player an easy to check comparison of the predicted vs actual battle results for the whole campaign. Then usually everything balances out after 2-3 scenarios and most "the AI has better rolls" arguments die out.
Im not sure how hard this is to do but if this isn't much work it would help solve this problem. The other benefit is in MP you can check how much of your victory/defeat was luck and how much skill.
The randomness is a bit TOO random, and the negatives towards the player as compared to the AI seem consistently less "random." I would say NO to removing randomness, but reducing it to a more reasonable level-- and eliminating the ridiculous advantage the AI seems to enjoy on the middle difficulty setting and up might not be a bad idea. It's truly frustrating to plan an attack with a series of well stocked, varied and well supported German units and watch a half strength polish infantry with 1 entrenchment do 4 or 5 damage to your elite units. Happens more often than not, too... which makes it seem less "random" to me.
You're right, except this thread isn't about that argument!Aloo wrote:I see a lot of new topics coming up about the randomness favoring the AI. This is typical for games using RNG: people always remember bad results and quickly forget the good ones, this is especially true if they loose a unit.
The solution is something I asked about a few posts earlier - giving the player an easy to check comparison of the predicted vs actual battle results for the whole campaign. Then usually everything balances out after 2-3 scenarios and most "the AI has better rolls" arguments die out.
This thread was (IIRC) posted because discussions about the discrepancy between predictor and outcome.
Unfortunately, the poll is phrased in a way that asks us between keeping things as they are and removing randomness COMPLETELY.
Nobody wants that. Nobody has asked for that.
And while there certainly has been complains about the computer "cheating" (i.e. getting more unfavorable discrepancies than favorable ones) that is not the issue either.
The issue is that the predictor far too often is inaccurate, forcing you into details dialogs to see the real probabilities.
The solution is to
1) contain any wild results into specified cases, such as "rugged defense", "unexpected encounter" etc
2) make it immediately clear to the player when he or she's at risk of getting such a "rugged defense" (etc) result
2) make sure actual outcomes follow predicted outcomes (sufficiently to make the predictor truly useful) when wild results aren't happening
The current combat routine needs no modification in itself. Only the presentation needs to be enhanced:
Instead of being told you will get a nice 1-6 result only to get a 5-2 slaughter, you might see "1-6 44%" telling you the prediction is itself only 44% accurate - the prediction is of LOW QUALITY and conversely the risk of a wild result is very high.
This would have told you to improve the odds before engaging and most importantly to reduce unknowns: you might need to make more suppressing attacks, move up more supporting units, take out a defending artillery, bomb the defender out of entrenchments, or any other of the usual host of approaches.
After one or three such preparatory actions, you might see a prediction of "0-7 94%" meaning that now the variables involved are much less prone to wild deviations (perhaps your initiative is so much higher than the enemy unit's that it shooting first is highly unlikely).
Instead of percentage numbers, I've already proposed settling for a traffic-light coloring system. Percentages easily lull you into thinking they are scientifically exact. Instead having GREEN, YELLOW or RED predictions would do the job just as fine: 44% would most definitely be RED (since over half the time, the prediction is worthless). 94% on the other hand, is probably good enough to merit a GREEN text.
After all, what you need to know, and what the game lacks today, is an easy way to see whether to trust the predictor or not. (If not, this indicates you should prepare the attack better, and the predictor accuracy should quickly increase)
So this isn't about having bad luck. It's about giving the player the info where he needs it - at his fingertips.
(This is more or less a repost from a few pages back for the benefits of new readers)
The main question is misleading.
Of course it is not about having 100% accuracy - but at least 50% would already improve game quality.
Coming off some online battles (mainly with the panzerliga.de players), I recognized the randomness of the battle results is a heavy factor to decide a game. This is especially true when both players are on an equal level of experience. You can still win if you have the better overall strategy - which is ok, when you play somebody who might not know the scenario very well or may be less experienced. However, if both players are veterans (like old PG players) and know the scenario well these random results wash up the game completely. Of course you can try to minimize them (even more emphasis on blocking and field possesion, even more preperations on full attacks) but as I said: if both players are experienced, there are moments, were you know "now it's on" and you did everything to get the best result and in the end there kicks in a "randomness" factor of 3 or 4 on a battle setup, that should be a sure thing.
What makes it even worse that ONLINE some players get better random results, than others. Maybe there is a logic behind it I havent found out yet. It is ok when you play the CPU on a big map and you know it equals up in the end. But on a short scenario 1:1 against an equal opponent there usually are fewer battles and them count more. I played a whole lot of PG games and the feedback I get from my peers is, that the basic battle results are not reliable enough to compete on a fair level and to prepare tactics.
Some examples: It makes no sense, when a 10-tank-unit attacks a non-entrenched infantry unit on open field/street and the result is 2:0 and you have the feeling it should be 2:2 at worst. Of course, it doesnt hurt when a prediction on an already-attacked-5-unit-strength city attack is 1:4 and the result will be 1:3 or 2:4 or 0:2. But it hurts when it ends up being 2:0 w/o any logic. Or a Stuka attacks infantry on open field and it is 0:0 altough prediction is - maybe - 0:4. Or a surrounded 3 unit infantry in a city survives 4 attacks from 4 different units, although already being hit by artillery and Stukas. And on the next turn a 10-tank-unit attacks an Opel Blitz with 0:1.
To conclude this: There has to be an undelying reliabilty in the results. I am not talking about fuming over a 1:3 result althuogh it should be 0:5.. This game never should be a simple looking-at-the-results-game.
BUT: It should be about battle logics. Randomness yes, but general battlefield and war basics simply have to be fulfilled. Right now, the game prefers an "attack with what you want" style -because you know you can and will get lucky. But simple war basics like entrenchment, soft on hard attack (and vice versa), type of ground, air superiority and structure of defensive lines should still be the main factors to decide battles, and remain untouched.
Of course it is not about having 100% accuracy - but at least 50% would already improve game quality.
Coming off some online battles (mainly with the panzerliga.de players), I recognized the randomness of the battle results is a heavy factor to decide a game. This is especially true when both players are on an equal level of experience. You can still win if you have the better overall strategy - which is ok, when you play somebody who might not know the scenario very well or may be less experienced. However, if both players are veterans (like old PG players) and know the scenario well these random results wash up the game completely. Of course you can try to minimize them (even more emphasis on blocking and field possesion, even more preperations on full attacks) but as I said: if both players are experienced, there are moments, were you know "now it's on" and you did everything to get the best result and in the end there kicks in a "randomness" factor of 3 or 4 on a battle setup, that should be a sure thing.
What makes it even worse that ONLINE some players get better random results, than others. Maybe there is a logic behind it I havent found out yet. It is ok when you play the CPU on a big map and you know it equals up in the end. But on a short scenario 1:1 against an equal opponent there usually are fewer battles and them count more. I played a whole lot of PG games and the feedback I get from my peers is, that the basic battle results are not reliable enough to compete on a fair level and to prepare tactics.
Some examples: It makes no sense, when a 10-tank-unit attacks a non-entrenched infantry unit on open field/street and the result is 2:0 and you have the feeling it should be 2:2 at worst. Of course, it doesnt hurt when a prediction on an already-attacked-5-unit-strength city attack is 1:4 and the result will be 1:3 or 2:4 or 0:2. But it hurts when it ends up being 2:0 w/o any logic. Or a Stuka attacks infantry on open field and it is 0:0 altough prediction is - maybe - 0:4. Or a surrounded 3 unit infantry in a city survives 4 attacks from 4 different units, although already being hit by artillery and Stukas. And on the next turn a 10-tank-unit attacks an Opel Blitz with 0:1.
To conclude this: There has to be an undelying reliabilty in the results. I am not talking about fuming over a 1:3 result althuogh it should be 0:5.. This game never should be a simple looking-at-the-results-game.
BUT: It should be about battle logics. Randomness yes, but general battlefield and war basics simply have to be fulfilled. Right now, the game prefers an "attack with what you want" style -because you know you can and will get lucky. But simple war basics like entrenchment, soft on hard attack (and vice versa), type of ground, air superiority and structure of defensive lines should still be the main factors to decide battles, and remain untouched.
-
Amaranthus
- Senior Corporal - Ju 87G
- Posts: 96
- Joined: Wed Jan 11, 2012 1:51 am
- Location: Adelaide, Australia
- Contact:
Okay, I just read this whole thread, and didn't see this suggested, so I'll link here what I just wrote in another topic (because this is probably the most appropriate thread for it, even though I'm reviving a 2-month-old topic): viewtopic.php?p=294763#294763
The basic premise - what about averaged combat rolls like in CEaW-GS...
The basic premise - what about averaged combat rolls like in CEaW-GS...
Re: RNG (That's Random Number Generator)
Edit: You poll is offering two extreme choices, some thing in between would be appreciated.
Some randomness is OK for me, but the result generated are , sometimes, really frustrating as they are. There are such large deviations from the average that it is hard to find a correlation with the combat depicted. And the predicted average is often more a cause of confusion than anything else.
Smoothing the result would be really appreciated.
So far I play mostly in chess mode, so I don't bother to reload after poor result, even in case of rough defense.
Something like a -3/+3 variation would be fine for me, or a similar distribution of probabilities:
-3 : 5%
-2 : 10%
-1 : 20%
0 : 30%
+1 : 20%
+2 : 10%
+3 : 5%
Some randomness is OK for me, but the result generated are , sometimes, really frustrating as they are. There are such large deviations from the average that it is hard to find a correlation with the combat depicted. And the predicted average is often more a cause of confusion than anything else.
Smoothing the result would be really appreciated.
So far I play mostly in chess mode, so I don't bother to reload after poor result, even in case of rough defense.
Something like a -3/+3 variation would be fine for me, or a similar distribution of probabilities:
-3 : 5%
-2 : 10%
-1 : 20%
0 : 30%
+1 : 20%
+2 : 10%
+3 : 5%
Last edited by Delta66 on Sat Jan 19, 2013 5:00 pm, edited 1 time in total.
Re: RNG (That's Random Number Generator)
My observations after 130 MP games.
1) You do get used to high swing of randomness
2) Generally I don't care anymore. Because I learned to plan for the worse.
3) But it magnifies the power of artillery. With artillery, you remove bad outcomes.
4) I like the idea of Experience toning down Randomness.
5) I also like my idea of "Luck Token" Given to a Single Unit. This unit only has good results while attacking. This is historical. See the history of Tiger tanker Otto Carius. Himself says luck is a factor.
6) I still don't like seeing 15 dice rolls of Russian Conscript attacking tanks and by luck killing 2 units of King Tigers, while just before that a T34 full strength died 8 units killing only 1 Tiger unit.
So there are some specific instances where I really dislike the randomness which are totally historically/tactically IDIOTIC.
1) You do get used to high swing of randomness
2) Generally I don't care anymore. Because I learned to plan for the worse.
3) But it magnifies the power of artillery. With artillery, you remove bad outcomes.
4) I like the idea of Experience toning down Randomness.
5) I also like my idea of "Luck Token" Given to a Single Unit. This unit only has good results while attacking. This is historical. See the history of Tiger tanker Otto Carius. Himself says luck is a factor.
6) I still don't like seeing 15 dice rolls of Russian Conscript attacking tanks and by luck killing 2 units of King Tigers, while just before that a T34 full strength died 8 units killing only 1 Tiger unit.
So there are some specific instances where I really dislike the randomness which are totally historically/tactically IDIOTIC.
Re: RNG (That's Random Number Generator)
For people that don't know after combat you can press L to see the combat report details.
If you have Windows 7 you can use the Snipping tool to take a screenshot of only the combat result screen without the rest. Very convenient when taking screenshots of dozen reports.
Will look like that.
Taking hundredths of screenshots and later processing them helped me to understand why prediction seems to be off and more importantly how combat works. Prediction, kill chance and actual results are tied together. The lower the kill chance the more likely you will fall below prediction, but opposite is true too. High kill (>60%) chance usually result in above prediction results. Anything below 35% and it's become a gamble. However I only took 150 samples and I anything below 1000 sample probably means shit.
Anyway after finishing the eastern Grand Campaign twice I can tell you one thing. I never lost a battle because of the RNG. I played on Colonel and FM. When I didn't score a DV it was, except for one buggy scenario trigger, always me and a flaw in my strategy. Of course it hurts like hell when your 14 over strength Tiger get screwed by some infantry in open terrain and you just lost like 420 prestige in strength points, but still solid strategy always bring victory in the end. Even a extra loss of 420 prestige like this didn't ruin me on the next battle. At least not unless you play Manstein difficulty I would say, but I never have.
In true my own arrogance and impatience was a bigger thread to me than the RNG at the beginning. Also when thing go too smooth I start to become careless. Like instead of waiting an other turn I had to force it immediately. Instead to stun an enemy with arty/ strat bomber or waiting for backup to get the mass attack bonus, I often rushed and attacked right away and got angry if it ended badly. However when I followed solid old the pike and bow tactics I did well even when I got bad rolls.
Always suppressing the enemy before an attack or flanking for the mass attack bonus.. If he can't shoot back he can not screw you. Protecting your tanks with high ground defense SPG arty means usually the AI won't even try to attack you. But in the end it's all about just taking it easy. No matter if I win or lose. Playing the game alone is already fun. Wining is just a bonus.
Still I got the impression dice rolls fall way too often into the same range.
If you have Windows 7 you can use the Snipping tool to take a screenshot of only the combat result screen without the rest. Very convenient when taking screenshots of dozen reports.
Will look like that.
Taking hundredths of screenshots and later processing them helped me to understand why prediction seems to be off and more importantly how combat works. Prediction, kill chance and actual results are tied together. The lower the kill chance the more likely you will fall below prediction, but opposite is true too. High kill (>60%) chance usually result in above prediction results. Anything below 35% and it's become a gamble. However I only took 150 samples and I anything below 1000 sample probably means shit.
Anyway after finishing the eastern Grand Campaign twice I can tell you one thing. I never lost a battle because of the RNG. I played on Colonel and FM. When I didn't score a DV it was, except for one buggy scenario trigger, always me and a flaw in my strategy. Of course it hurts like hell when your 14 over strength Tiger get screwed by some infantry in open terrain and you just lost like 420 prestige in strength points, but still solid strategy always bring victory in the end. Even a extra loss of 420 prestige like this didn't ruin me on the next battle. At least not unless you play Manstein difficulty I would say, but I never have.
In true my own arrogance and impatience was a bigger thread to me than the RNG at the beginning. Also when thing go too smooth I start to become careless. Like instead of waiting an other turn I had to force it immediately. Instead to stun an enemy with arty/ strat bomber or waiting for backup to get the mass attack bonus, I often rushed and attacked right away and got angry if it ended badly. However when I followed solid old the pike and bow tactics I did well even when I got bad rolls.
Always suppressing the enemy before an attack or flanking for the mass attack bonus.. If he can't shoot back he can not screw you. Protecting your tanks with high ground defense SPG arty means usually the AI won't even try to attack you. But in the end it's all about just taking it easy. No matter if I win or lose. Playing the game alone is already fun. Wining is just a bonus.
Still I got the impression dice rolls fall way too often into the same range.
Re: RNG (That's Random Number Generator)
Removed, no. Toned down, absolutely. Some results are WAY off.
Re: RNG (That's Random Number Generator)
I think the dev's forgot to include a bell-curve.