Op fire weirdness and the "hot or cold" effect.

[IainMcNeil]

I am curious about the if 1 shot left and moral <80 maybe i am miss understanding the purpose of this line.

it seems to check if it is hit by moral it might just have low moral from last round, if moral last round is greater than this round
But anyway, im not trying to pick apart and have you defend every line of code, after a big loss you can't help but think the game was somehow bugged.

This means if your morale is less than 80 you do not hold back your last shot for a juicy target. The idea is if your morale is low you might get suppressed if the enemy fires so why hold your shot looking for a better target.

[Merr]

Ian,

Thanks for your time in making me (us) understand how the code works. I agree with you when it comes to spending time and resources looking for something that might not even exist.

I ran my own test on randomness by adding code to the TURN function. The TURN function is a free function that "turns your unit" to face a different tile.
I made it so that there is a 50% chance the unit will turn to face the new tile.
My conclusion is that everything is OK with regards to SOLO play ... meaning that if my unit didn't turn to face the new tile I would keep trying again and again.
Sometimes it took 9 attempts to change facing, sometimes it took 1 attempt ... etc etc.
I even switched units to see if the chance was "locked" into memory ... meaning if one unit passed the roll then would another unit pass too.

Below is the code change I used ... makes a good test (IMHO) ;

Code: Select all

FUNCTION ALL_TURN(me, tilex, tiley)
{
	if (Rand(0,1) == 0 )
	{
		AddVizUnitTurnToFace(me, tilex, tiley) ;
		if(GetUnitTurret(me) == 1)
		{
			// turn the turret to face the target as well.
			AddVizUnitTurnToFire(me, tilex, tiley) ;
		}
	}
	else
	{
	 // do not turn
	}
}

[Merr]

Also, since it took 9 attempts to change my units facing .... one might think that a 50% chance would mean it should change facing on the 2nd attempt, right?

Well, it goes back to what I said with regards to a 1d6 .... If you need to roll a "6" you have a 16% chance right?
One might assume that after rolling the die 6 times it would roll a 6 .... maybe, maybe not.

If you want to try a 1d6 attempt to "turn your unit to face another tile" then change the Rand(0,1) to read Rand(0,5).

Again, makes a good test.

[IainMcNeil]

Thanks for that - that's probability for you!

I dont know how we deal with player expectations without fudging the results to be more palatable.

[Merr]

Thanks for that - that's probability for you!

I dont know how we deal with player expectations without fudging the results to be more palatable.

Ian,

Yes, that's how I look at probability ... perhaps players want more absolutes?

I chose 50% in my test because it goes back to Opfire ... Beyond 3 tiles against a sneaking unit it's 50% (unmodified, not considering killchance).
So, in my example, if we used this as an Opfire, it took 9 attempts for my unit to consider opening fire. This is the "bad roll" or "heavy rolls" that folks are seeing, and it's just that, bad rolls.
Like I said, I was able to turn my unit with one attempt, sometimes two or three, and I stopped when it took 9 attempts.

IMO, I think if players want more fudging to meet their expectations then they can "mod" a scenario with the fudged numbers and post their change.
If everyone "likes" the change then perhaps Slitherine can then look into the matter (spend time and resources) to make it a global BA change.
On the other hand, not everyone has the time to fudge numbers.

If Slitherine fudged the numbers to change that 50% to 80% then we'd be complaining that Opfire is too much?

You realize (Ian) ... you'll never win the arguement

It's the typical Goldilocks story ; the player has a 33% chance the porage being served is too COLD ... after 9 servings, it's still too COLD!
Goldilocks had better odds, after 3 servings it was JUST RIGHT.

[pipfromslitherine]

We spend this morning reanalysing the way that the opfire logic works, and we think that (because there is only so much we can usefully feed back to the player) some of the internal logic needs to change subtly to reward good play more, and prevent people feeling cheated - and the reality is that most of this is making sure it feels right to as many people as possible.

We also found a bug in single player where the AI wasn't being penalised as much as it should be. So that fix should help make some of the brutal missions (coughBulgecough) a little less tough.

We're aiming to get this fix out with the MP cheat detection upgrade before the end of the week.

Thanks to everyone who let us know their feelings on this, and indeed any other, element of the game.

Cheers

Pip

[ssouthrey]

Also, since it took 9 attempts to change my units facing

Merr, in your tests, keep an eye on how many of these streaks you're getting and how long they are. If you do 100 d2 tests and you get 50% (50 black, 50 white) that seems great, but if you got 25 black, 25 white, 25 black, and 25 white then you do have a problem with the randomizer. I doubt there is a problem and I mentioned that I was pretty convinced it wasn't the random number generation. Since you're doing the test, I thought I would throw it out there for you to keep an eye on it. Thanks for doing the test too!

By the way. thanks Pip and Iain for reviewing the op fire logic.

[LOGAN5]

Merr, in your tests, keep an eye on how many of these streaks you're getting and how long they are. If you do 100 d2 tests and you get 50% (50 black, 50 white) that seems great, but if you got 25 black, 25 white, 25 black, and 25 white then you do have a problem with the randomizer.

see the problem is, 100 d2 tests will rarely return 50% because the 50% is based on infinity.. so you could really get 100 blacks in a row and nothing would be wrong with the program although it is unlikely.. But for example i have seen black come out 30 times in a row in roulette, thank god i was not involved in playing... there is a video in the other thread you should check it out , here is another link

http://www.youtube.com/watch?v=bY7aRJE-oOY

watch the whole thing if you like, but if you don't have 1 hour to waste on this just check out the math 101 section skip to about 18 minutes into it and check out what this guy has to say

[cptkremmen]

Single player seems fine to me, I have never had any objections to either the hot/cold effect or reaction fire in single player. For this reason single player has become my preferred means of playing.

Multi player appears to give more problems. Some of this i am sure is when you are facing a player better than you. it will look as though they are being luckier than you.

Still not 100% sure though. If I wanted to (which I don't) could I cheat in multi player? Could i go into any of the settings and tweak them so that fate appears to affect me more favourably? Do you have any variables stored in insecure files where they could be changed? I honestly doubt if much if any of the effect people notice is actually due to cheating, but just asking

I actually quite like the idea of deliberately cheating some of the figures (Slitherine not me!) so that although not right it gives the impression of being right. I believe the chance of hitting already goes up 10% after the first shot/hit which is a good idea, maybe a couple more adjustments in that direction would make people feel it is fairer.

Andy

[pipfromslitherine]

All of the data and scripts which feed into this are protected against cheating, even more so with the upcoming update.

I think that once 1.3.10 arrives things will be a lot 'fairer' feeling. While the initial thinking behind some of the logic was good, it obviously was frustrating people, and so it should now reward good play, while leaving in those situations where it will all depend on who is quickest on the draw - but the important thing is that the player will have good visibility on the factors which cause the behaviours.

Cheers

Pip

[cptkremmen]

Sounds jolly good pip.

When is 1.3.10 due?

If we are part way through a multiplayer game is this affected by a new release?

[pipfromslitherine]

Fingers crossed for final test success tomorrow.

Ongoing MP games shouldn't suffer anything more than perhaps some non-synced outcomes (e.g. it will be running different logic for the replay than the original turn). But only once post update.

I'm working on a better solution, but it's tough to come up with one that doesn't mean passing 10Mb savegames around...

Cheers

Pip

[pipfromslitherine]

The 1.3.10 update is now live. This should make the behaviour of reacting units more consistent. As always we still welcome all feedback.

Cheers

Pip

[djddalton]

I have a feeling that if your tactics/overall strategy are lacking in thought, it might come across as looking like bad luck. However, I have had a few incidences on this game where H&C seems to affected the whole game, no matter what you do...

Not sure if it is really a problem with the number generation. It would be awesome if someone could try out a time-based randomiser though (can you change the code in such a way on user maps?) to see if it makes a difference. Because if a computer is doing all the dice throws, there is bound to be some sort of horrible pattern in the results. For example...my mp3 player will always play 'random' with the same order of songs.

I think the frustration with this game might not be with the throws, but in fact with the bias over certain units. For example, playing BA France MP, all my probs fade quickly to 0 for AP kills on Germans at long distance, whereas my opponent can pop off my tanks at long range with ease. The result of this is that you must rely on very specific tactics to make your kills! Because there is 0 chance all the time, it looks like your opponent can just drive in without seeing a single thing so much as deflect. This combined with the fact that there is no force selection on mp yet (!) means that there is little scope for developing new strategies.

So it would be nice to see a few more 1%s and 2%s and 5%s in there, whereas the current models for probability kills seems overly simplistic. Or is it realistic? Anyone got a good spin on this?

[PirateJock_Wargamer]

The way I understand it is the Chance to Kill - the important factor - is based on the Chance to Hit & Chance to Penetrate and, racking my brain from the previous discussions, there is no random element to this. Randomness was associated with (i) deciding if there was reactive fire and (ii) the amount of damage caused.

I believe the way the game is modelled, the reason the Chance to Kill drops off to 0% is a function of the target's armour at the point of contact (front, side, top or rear) and the quality of the firing gun. If you are shooting at thick front armour with a low calibre gun and at distance there is no way you will be able to penetrate the armour, i.e. Chance to Penetrate, and so Chance to Kill will be also be 0%.

Can somebody confirm if I've got that right.

As it stands I'm OK with seeing 0% - and I certainly see a lot of them - when I'm trying to hit an Axis big cat with one of my peashooter ATGs. I think the current unit values are relatively accurate.

Cheers

[IainMcNeil]

Yes you have it right - chance to hit and penetrate are multiplied together to get a final result so if either are 0 the end value is 0.

[Merr]

I think the frustration with this game might not be with the throws, but in fact with the bias over certain units. For example, playing BA France MP, all my probs fade quickly to 0 for AP kills on Germans at long distance, whereas my opponent can pop off my tanks at long range with ease. The result of this is that you must rely on very specific tactics to make your kills! Because there is 0 chance all the time, it looks like your opponent can just drive in without seeing a single thing so much as deflect. This combined with the fact that there is no force selection on mp yet (!) means that there is little scope for developing new strategies.

So it would be nice to see a few more 1%s and 2%s and 5%s in there, whereas the current models for probability kills seems overly simplistic. Or is it realistic? Anyone got a good spin on this?

One factor, regarding BAF, might be the scale or range effectiveness. Since it's using late war range data, the early war guns are no good beyond a few tiles, which indeed, changes you tactics from late war strategy.

Range in BA's "range bracket" is .... 100m, 500m, 1000m, 1500m, 2000m ... So, perhaps using a rubber range for BAF may help?

example ... 100m, 200m, 300m, 400m, 500m. So, instead of having to be within 3 tiles to be effective, you are now effective within 7 tiles.

Now, a quick change would be adjusting the Range Bracket values ... located in CombatTools.BSF, FUNCTION GetRangeBracket (distance).
By stretching out the range it might help change the odds across the board.

Merr

[PirateJock_Wargamer]

I've got a feeling that changing range brackets between core game and expansions would cause confusion. Stick with one and get up close and personal ... as Corporal Jones would say "They don't like it up 'em!"

Cheers

[bones65]

Yes you have it right - chance to hit and penetrate are multiplied together to get a final result so if either are 0 the end value is 0.

there's always gotta be a slight chance, the'critical hit' from a.s.l. for example! .. i played one game maybe 20yrs ago, and took out a panther or k.t. (can't remember which,.. but never forgotten the moment!).. we both laughed sooo much, with an 81mm mortar!! we came to the conclusion the shell must have gone down the hatch.! i've heard of real life accounts of rifles shooting rivets in tanks and the rivets injuring , or setting off shells.. surely there could be a 1 percent chance, even if it only pops up every 10 games..but for the fun factor alone it would be rational to include it.,.. i've experienced the hot cold effect on both sides of the coin.. 'race for the town' is the biggest culprit.. i think the mp 3 player randomiser mentioned a little while ago is a good illustration of what could be happening here. . still love the game and play it almost every day, so this is constructive not destructive

[Jonesy1760]

Here are some suggestions I would like to see in this game...Inf. should not "surrender" to artillery fire...how about a retreat or fall back. Tanks should not be suppressed so easily. I have had Panthers and Tigers suppressed and then turn tail. Suppressed in place is more likely. Or not turning their ass to enemy!!! Tank fire while moving is far to accurate!!! Inf. are spotted in cover far to easily.


I will return with more suggestions as how this game can be improved