41 coaches online • Server time: 16:47
Forum Chat
Log in
Recent Forum Topics goto Post Creating a custom to...goto Post Secret League Americ...goto Post DOTP Season 4
SearchSearch 
Post new topic   This topic is locked: you cannot edit posts or make replies.
View previous topic Log in to check your private messages View next topic
AlcingRagaholic



Joined: Aug 02, 2003

Post   Posted: Aug 06, 2003 - 05:12 Reply with quote Back to top

If you need practice, make a team with no rerolls. It helps you learn to roll as few dice as possible...

Z
Dragoonkin



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 08:20 Reply with quote Back to top

Of course it could always be that the BLOCK DICE THEMSELVES (and I'm talking real ones, not electronic) aren't properly weighted and roll LESS Skulls than they should...so when you get a "proper" distribution of bad rolls (as Java provides) you're blown away by how many Skulls you SHOULD have been rolling and weren't.
Twahn



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 09:01 Reply with quote Back to top

I must admit, though I at first scoffed when people started saying that the dice were bad and needed fixing, I've been having a few games of last myself where one or the other team seems to roll a lot of the same number. Generally it seems, 1s (conveniently but truly). This even seems to extend to the kick offs where I've been experiencing a lot of repetitions of the ones found at the ends of the tables?
In a game just recently, my bashy team had incredible luck rolling POW after POW where my opponent could not buy a GFI or a dodge (with his elves). At least 50% ones. My bashes were first class with even the fair fights in some turns coming out POW POW POW.
This has been weighing on me because I've seen it happen a couple of times recently where I never used to see it happen before.
Especially concerning is the fact that POWs are actually 1s on the Block die (if you look at the script)...

Are the dice Gods suggesting we need more violence by giving us nothing but 1s?

_________________
Founder of the Southern Wastes League

DonTwahn's Dungeon = Cheap Miniatures! Check it out...
Vertigo



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 10:23 Reply with quote Back to top

Well i was really pissed off by my bad luck and by the good luck of my opponents -
so to prevent hearthattack i made a papernote of every 1/9 chance taken - and 1/6 chance taken
well this helps a lot. seeing your opponent burn 1/9 chances like candy and getting a good respect
for them in your own game - your luck turns Smile
Skar2



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 10:25 Reply with quote Back to top

I am pretty sure, that the rolls made by the computer are as good as the random generator and I think everybody will agree that there isn´t something in the programm code to change the odds for one team. So of course, the amount of rolls of 1,2,... will be statistical distributed. But every person who knows a little bit of statistic knows that it´s very unlikely to achieve a perfect distribution (like Bloodrunner told). Normally the amounts will differ and scatter around one expected value which is 1/6 of the rolls. The expected value within the standart deviation by making 600 rolls will be 100 +/- 10 (66%). Only 1 of 3 numbers should scatter more than this.

But anyway, I don´t know what´s the problem with giving a statistic at the end? Of course it´s work to implement that, but it could be interesting and gives you a small sign at the end of the game why you win/loose. It´s clear that there are games where nothing really works. But perhaps it is because your op has played so intelligent that you had to make lots of difficult moves? After the game you could see: "Today I had really bad statistics, so it´s not so bad to loose the game." or "I have lost, but my statistic says, that my rolls where ok. Why did I loose? It seems that I made some stupid mistakes..."

This is why a statistic at the end could be nice.
Chickenbrain



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 10:29 Reply with quote Back to top

And well losing strike due to bad luck sucks... but thats the game.

_________________
Join Themed Blood Bowl for the joy of Themed Teams.
WolfyDan



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 11:08 Reply with quote Back to top

Hmmm...

I do think that overall the numbers that come up are random apart from 2 points I have noted.

1/ Certain actions.
-- Certain actions always seem to fail for certain people, and this may be down to the individual computer/ OS whetever etc etc they are using. From personal experience I have never succeeded more than 4 GFI's in a row, and on average attain about a 30% success rate on a 2+. I'm not complaining, as I think I have onlt ever failed about 3 dodge attempts on the same figure (and I'd rather dodge than GFI), but I have seen a lot of people comment that certain actions are doomed to failure. Not overly sure where java get's it's numbers from, and whether this differs for action type, but I think it might be worth checking over.

2/ Rerolls - again I'm not sure exactly how the java generates rerolled numbers, but from experience I see that they have a VERY good chance of generating exactly the same number. I have even rerolled double POWs (on my own in test games) just to check it was a reroll thing, and not a '1' thing. The two examples I can think of that might cause this is either the random number generator 'failing' to pick up new numbers in certain sitautions (you have to love the witchcraft that is computer sience), or else the java random number generator has a prevalence of grouping the same numbers together. The second could not be spotted by looking at an average number of rolls, but again it is not as random as it should be.
EvolveToAnarchism



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 12:35 Reply with quote Back to top

Psikobunny, you stole my thunder. Sad For anyone who likes to complain about the random number generator, I was going to offer the following advice: "Take 2 courses and call me in a semester."

Preferably statistics and cognitive psychology.

As Always,
Evolve To Stuntydom

New Stunty League/Tournament
MattiasF



Joined: Aug 02, 2003

Post 3 Posted: Aug 08, 2003 - 12:46 Reply with quote Back to top

All block-rolls, skill-rolls, armour- and injury-rolls are in teh log. Can someone (that have more time than me!) write a script to statisticlly analyse the distribution. This could possibly end this discussion...

I know that distribution and luck are not the same. And I hope this analysis can prove that.
Logan



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 13:39 Reply with quote Back to top

Also another problem is that the rolls are independant, the odds of failing a 2+ with a reroll are 1/36, but if you fail the initial roll the odds of failure 'improve' to 1/6. This also means the odds of rerolling the same as the previous number are 1/6. Which is what you remember. Just for the record however the dice aren't all bad I manged to roll two sets of double 6's for armour and injury rolls against a friend yesterday. I may forget those double 6s however I bet he won't in a hurry esp as all he could roll were 1s. People remember the bad stuff and forget the good stuf and that I feel it what is wrong with the world today Laughing
So forget your double 1's and remember your sucesses instead like your ogre intercepting on a 6

_________________
God does not play dice, therefore he does not play bloodbowl and is missing out Laughing
Sinner



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 13:47 Reply with quote Back to top

Logan wrote:

like your ogre intercepting on a 6

and failing his bone head for 4 turns, thus not moving, loosing the game and staying on the pitch for 2 days, because he tries to remember what he has to do with this ball thingy he´s cradling in his hands Very Happy

Hey guys: everything is right, the number generator doesn´t want to fool you (only you and not your opponent, though), you just are as bad a coach as I am.

Cheers

_________________
Sinner
Darkie's Dreams - successfully cherrypicking any race, any coach, any rating, any number of DP since 20/09/2003 ... and still winning!
Christer



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 14:10
FUMBBL Staff
Reply with quote Back to top

WolfyDan wrote:
1/ Certain actions.


Java gets all dice rolls from the same number generator.
Now, there has been incidents where SkiJunkie has forgot to remove debug code, making all rolls of a particular type fail or succeed. Naturally, this sort of thing is _very_ quickly identified.

Another interesting point is that both you and your opponent are using _the same_ random number generator. That's what the dice sync messages are all about. Neither client trusts the other end, so all dice rolls are checked against their own random number generator to make sure the other end isn't cheating.

Your example of 4 GFIs in a row.. How many of those series are you doing? The statistical probability of succeeding 4 2+ rolls in a row is around 48%. Failing a series of those isn't at all unlikely.

WolfyDan wrote:
2/ Rerolls


The probability of rolling the exact same two-die block is 1/36. 1/9 for double pushbacks.

Again, the same random number generator is used (as you will see if you enable verbose dice sync). Ie, the random number generator is being called for rerolls and is being confirmed by the opponent's client.

I am confident that the random number generator is proper. However, this is also a psychological issue. You simply don't remember the rolls which go the way you expect.

-- Christer
Guest





Post   Posted: Aug 08, 2003 - 15:59 Reply with quote Back to top

A) % that you have to win in Lotto. (in Belgium)
B) % that you have to do 12 times 6 in a row with a dice.

A) 0.0000000265 %
B) 0.0000000459 %

Conclution:

You have more chance to throw 12 time 6 in a row with JavaBBowl than winning in Belgium Lotto. Wink

That's why I play JavaBBowl and Not to Lotto.

Very Happy Cool
deneb
Lord-Willi



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 16:27 Reply with quote Back to top

@ christer:
Sure the odds of rerolling the same double is 1/36 but you only use it on double skull or double 2 (both down) and combination 1 and 2.
That would anyway lower the odds to about 1/81 to fail a double block with a reroll (fail in sense of suffering a turnover). If that 1.x % chance hits me three or four times per Game (just counting blocks) than everyone would complain it is not usual, because i just make about 80 Blocks per Game...

Perhaps it would be easier to just count the actual dice-throws...
I won't need a percentage, just a number of throws i made like
1: 32
2: 19
3: 21
4: 12
5: 90
6: 0

or is this to difficult as well ?
SkiJunkie



Joined: Aug 02, 2003

Post   Posted: Aug 08, 2003 - 17:17 Reply with quote Back to top

Quote:

Especially concerning is the fact that POWs are actually 1s on the Block die (if you look at the script)...


POW is 6 and SKULL is one. This change quite a while ago.

I can add a number counter at the end, but unless I take the time to sync the number tallies over the network you are only going to see the distribution of the rolls made on your computer. So if 12 rolls are made in the game, 3 on your computer 3 on the other computer, you might see your computer rolled 1, 2, 4, 1, 5, 1. Hey that's wrong! Well the other computer rolled 6, 4, 2, 6, 2, 6. So it's not. Of course this is an overly simplified example.
Now another problem is the same generator is used for scatters, random player selection, and anything else that needs a random number. So rolling an 8 for scatter steals a 6 from the d6 statistic etc.

_________________
--SkiJunkie
________________________________
Java BBowl - http://www.geocities.com/javabbowl
Mirror - http://javabbowl.no-ip.org
Mirror - http://home.austin.rr.com/javabbowl
Display posts from previous:     
 Jump to:   
All times are GMT + 1 Hour
Post new topic   This topic is locked: you cannot edit posts or make replies.
View previous topic Log in to check your private messages View next topic