Random weather not random???

[olorin42]

Running the latest beta (1.08.07 exe file). I'm playing as the soviets against the AI. I finished the June 26 1941 soviet turn and told the computer to do it's turn. It finishes and i see Mud in most of Western Russia. I know this massively hurts the AI so i figure i'll go back and regenerate the turn to give it clear weather. I've tried 14 times and it keeps making that same weather zone mud but all the others clear. Consistent as heck. I saw something in the patch notes about a new random number generator. Is it working right? Weather is not being random - and yes the option is NOT picked for non-random weather. Feedback greatly appreciated.

[chaos45]

Weather is already generated by the Soviet players turn for next turn.

is only a % chance that the forecast will change and its a low % like 20-25% not sure when that is rolled.

So if you are already past the point of the rolls it wouldnt change ever as its already saved/locked in by your turn.

You would have to drop back an entire additional turn to make the game re-roll weather for the turn you are on is how I understand it.

[Mehring]

Something I notice with "random" weather is that while it's usual for every weather zone to experience mud in the summer of 1941, it's equally usual for no weather zone to have mud in 1942. This is so consistent as to make me wonder whether the "once per zone" rule is carried over from one year to the next?

[morvael]

chaos45 is right, weather for turn X is determined on turn X-1 in the Axis logistics phase.

Mehring, variables blocking mud in summer should reset during October turns. I will debug it, but IMHO it's working fine.

[morvael]

I can confirm that weather works ok, but I will reduce chance for mud in the summer as it was higher than before (on old random table) and the mud usually hit early.

[Mehring]

OK, thanks morvael. Strange though, it happens pretty much every game, mud in 3/4 zones 41, mud 0/1 zone 42. I don't recall a game with weather outside those parameters.

[loki100]

I think its the way the statistics are working out and that gives an impression of a lack of random element.

Crudely there are two rules in place - sampling with replication till you get mud (ie that you had no mud last turn has no bearing of this turn). Then certainty of no mud once mud has happened.

So to keep it easy, lets say each turn there is 10% chance of mud and the summer lasts 10 turns.

In theory you *should* get 1 turn of mud per zone (10*10/100), but in each turn there is only 10% chance. So the chance of mud in one zone from the start of the game (out of the 4) on T3 is (10/100+10/100)*4, which means there is quite a high chance that by T3 at least one zone will have triggered mud. Its odd that most games seem to imply this happens in the Ukraine though.

Once an area has triggered the summer mud, the odds per turn of mud somewhere becomes 3*10/100 and so on. So the chance of the last mud event is relatively low.

There are other ways to set up the randomised sequence.

In my game against vigabrand, we got mud in the south in July in 1942 (forget the actual turn).

[Mehring]

There is no way, other than freak chance or bug, that my games would almost invariably exhaust mud opportunities in 41 yet not have any in 42. It happens too often in my games for me to dismiss it as chance. If others players experience an even ratio of mud every year, though, looking at that bigger picture puts my experience into perspective. That said, an occasional instance of mud in 42 is not beyond my own experience either. But it's just that: occasional.

[chaos45]

So is weather to random now?

As I just got 2 zones of mud first turn of july and rumor is even more mud second week of July.....seems abit high on the mud count already for summer.....

Going to be very hard to conduct summer campaigns with almost for sure mud every week in at least one zone during a campaign season.....

[morvael]

There is still limit of 1 mud per zone per summer. It just usually happens early now.
Weather tables will be adjusted for next version so that dominant weather type will get 90-95% chance to happen instead of 80%.

[chaos45]

Yes, definately needs reduced....as now you are looking at for sure mud at least once every zone each month.

Basically means you can get a successful breakthrough in the summer and odds are your going to be stumped with mud the very next turn as at most you will only have 2-3 turns before each mud rotation.

Will make it much easier on the Soviets to defend early game and much easier for the Germans to defend late game.

Any ETA on next patch? As to me you mise well stop current games if you have 20% per turn per zone for mud each week over the summer- summer offensives will be pathetic after 1941.

[morvael]

No, you get it wrong. One mud per zone is allowed between 20th June - 9th October, so once it hits (early), it's gone for the summer. Lower chance will make it come later, but still the maximum of 1 mud per zone per summer will hold true.

No eta, nothing critical so I don't want to test Matrix and 2by3's patience by launching multiple hotfixes.

[chaos45]

If thats true, then I shouldnt see a single additional mud turn the rest of the summer.

As every Zone has been hit by mud in first 2-3 turns of summer...yep every zone right at the start of the summer....apparently July is now monsoon season.......

[sillyflower]

ORIGINAL: chaos45

As I just got 2 zones of mud first turn of july and rumor is even more mud second week of July

Only 2 mud zones on T3? You were lucky - I have just had all 3 plus a total sense of humour failure. No mud in north of course because that won't come before most of german army is in it.

This does not seem very random (one of Pelton's issues in the past) and I can't help but be slightly sceptical about the 'random' nature of early mud. I understand statistics (it was even part of my 2 original university degrees) and I know that the occurrence of a once in a million event does not mean that the probability was in reality higher than 1 in million, and that no one ever reports of the 999,999 times the almost inevitable event occurs.

That said reports of a problem especially in the south do seem to be surprisingly frequent ......................

@ Loki - I'm not sure what you mean by a 'high chance' of mud on T3 but it is not the case that mud is likely somewhere on T3 or any given turn. You have to look at the probability of all zones being clear and subtract that from 1 (or 100 if using percentages). The chance of mud is 1:11 or 9% The probability of all 4 zones being clear on any given summer turn is therefore 69%. If you ignore the north ( which is irrelevant in the early turns, the probability of the other 3 all being clear is 75%. The probability of all 3 being mud in the same turn is 1 in 1331 or 0.075%. 1 in 1451 if you add in the chance of north being clear.

As you say, once there has been mud somewhere, these probabilities change even when the probability remains constant for any particular zone.

[morvael]

20% is too high so mud comes early, it will be reduced to 5% in next patch.

[sillyflower]

Hooray

[loki100]

ORIGINAL: sillyflower

@ Loki - I'm not sure what you mean by a 'high chance' of mud on T3 but it is not the case that mud is likely somewhere on T3 or any given turn. You have to look at the probability of all zones being clear and subtract that from 1 (or 100 if using percentages). The chance of mud is 1:11 or 9% The probability of all 4 zones being clear on any given summer turn is therefore 69%. If you ignore the north ( which is irrelevant in the early turns, the probability of the other 3 all being clear is 75%. The probability of all 3 being mud in the same turn is 1 in 1331 or 0.075%. 1 in 1451 if you add in the chance of north being clear.

As you say, once there has been mud somewhere, these probabilities change even when the probability remains constant for any particular zone.

should have been clearer, what I meant was 'by' T3. I think you need to use Bayesian logic (to be far too technical) in that at the game start (T1) the odds of mud by or on T3 in one of the 4 zones are 2 (ie either T2 or T3)*4 (ie 4 zones)*base odds (currently 9% of mud in any one turn?). So 2*4*9=54% chance of mud in any of the zones in or before T3?

Now if you are in T2 and only interested in the chances for T3 (assuming no mud yet) its 4*base odds for a single zone).

So the chance of mud in turn x when you are in turn y are hard to predict (due to the rule that once mud happens it can't again), but the chance is a case of combining all the factors.

having said that, I've just done a test human vs human game with random weather, no moves etc, just press end turn. Got mud in the south on T3 ... so something is working oddly as that has happened in all my recent games and I think is common for others?

[morvael]

9% was with old random weather, now it's 20%, but will be 5%.

[loki100]

ORIGINAL: morvael

9% was with old random weather, now it's 20%, but will be 5%.

ah that does explain the almost certainty of some mud by T3 - 2*4*20 ie 120%, its not quite a given as each turn resets the odds but it is nearly impossible to avoid.

If the summer is 15 turns (if I recall October allows a second instance of mud even if it has already happened?) then your 5% may bring a new problem ---- 15*4*5=300%. In other words the norm will be 3 instances in the 4 regions over the summer. Would need to do the calculation properly but its going to come out in that region.

Depends on what you want? Also on modifiers etc (so if mud in some zones is more likely than the crude base rate?), but 5% may well set up game where 1-2 instances of mud is quite feasible and none at all in some zones is plausible. No big deal, I had that in 1.07 games.

[morvael]

The limit of 1 turn of mud per zone per summer stays in.