| | You are not authorized to post a reply. |
| | Author | Messages | |
Richard II
Commander
3663 Posts
    
 | | 04/10/2005 7:10 PM |
| | Originally posted by DoubtofBuddha I believe. | |
 | |
|
Richard II
Commander
3663 Posts
| | 04/10/2005 7:10 PM |
| | I dunno if I have anything definite, but it seems to me that the vassal dice bot is a little bit "streaky". One day it'll be all 15's and better and the next day it's 1,1,2,1(I actually did that with a dire bear). | |
| |
|
rgrayua
Sergeant
400 Posts
| | 04/10/2005 7:25 PM |
| | I think the dice-bot does a pretty good job; I've had games where it feels like I never rolled greater than 10 and ones where I consistently got good rolls. The 20s and 1s just tend to stick in your mind longer, depending on which end you were on. Honestly, I can't remember an inordinate amount of 20s rolled by it, at least not in my favor. [:D] | |
| |
|
Zippy
Underboss
1993 Posts
Whitewater, WI
| | 04/10/2005 8:02 PM |
| Try this:
1) Log the game
2) plot the rolls on a histogram
3) look at the histogram - it should be pretty flat indicating random distribution. The Vassal die roller is not truly random, but so close we cannot perceive the difference.
Keep in mind, humans learn by patterning, so we tend to see patterns even when there are no repeatable ones. "Random" sometimes seems like "trends", when in fact it's "fluke".
To make a histogram, list 1 through 20 down the page, put an X after it for each time that roll shows up (use graph paper and fill one square with each X:
Roll
1 XXXXXXX
2 XXXXXX
3 XXXXX
4 XXXXXX
...
20XXXXXXX
Post your histogram here for fun. Then we'll compile all of them into one giant histogram and see if there's really a problem with the roller or not! | |
There are 10 kinds of people in the world; those who understand binary, and those who don't.
Reference Thread, H/W List, Champion of the Catoblepas | |
|
Zippy
Underboss
1993 Posts
Whitewater, WI
| | 04/10/2005 11:36 PM |
| Here's some data:
1) A string of rolls pulled from Vassal: 9,5,10,5,5,7,14,11,17,1,3,10,2,5,12,8,14,7,8,19,6,5,8,7
We might say "I see a problem - Vassal's d20 is rolling WAY low!"
2) Another sequential string from same rolling session: 16,16,16,16
We might say "Whoa! This Vassal d20 thing is BROKEN!"
1) and 2) do not tell us there's a problem. They are TINY samples and happen all the time in random strings. Check a random number table sometime and TRY to find patterns - you will I guarantee if you want to find them!
Remember, the odds of rolling 20-20-20 are the same as the odds of rolling exactly 3-17-8.
You might think these are patterns, but from 426 rolls here's the histrogram (we expect one or 2 numbers with quite low freqency and one or two more often than the other numbers - it's part of random distribution, even totally flat frequency only happens at infinite test rolls!):
1 =23
2 =19
3 =19
4 =25
5 =25
6 =21
7 =18
8 =20
9 =26
10=23
11=17
12=12
13=23
14=20
15=19
16=27
17=19
18=28
19=24
20=18
Looks random to me, but we'd have to do a lot more rolls to verify. No one complains "I don't roll enough 12's in Vassal!" but that is what the data might be telling you. I won't go in to the details of other tests, but the dice roller tests OK as "quasi-random" by my calcs (it is close to random but not perfectly so, similar to real dice which are NEVER truly random).
I will entertain data discussion to the contrary, but I think we are all safe (except if someone hacks Vassal engine) thinking the Vassal d20 is legitimate. I was relieved to see the results since there was so much concern. | |
There are 10 kinds of people in the world; those who understand binary, and those who don't.
Reference Thread, H/W List, Champion of the Catoblepas | |
| Hero of Skirmish
doubtofbuddha
Commander
3371 Posts
| | 04/11/2005 9:31 AM |
| Fair enough.
My work involves a fair amount of statistics (and my background is in computer science, so I am aware that computers only generate pseudorandom numbers) so I was aware that my few samples on VASSAL were hardly statistically significant. It just seemed a bit off to me to see as many critical hits as I do when I play the game, when on tabletop it very rarely reaches the level that I have seen on VASSAL.
| |
I am not gone. | |
| 
Avatar of Skirmishes
tullywi
Sergeant
982 Posts
| | 04/11/2005 12:55 PM |
| I also went through and did 500 rolls.
- 5 of the numbers had 30 rolls or higher with a max of 32.
- 5 of the numbers had 20 rolls or less with a minimum of 19.
- 5 of the numbers had rolls between 25 and 29.
- 5 of the numbers had rolls between 21 and 24.
The distribution seems fairly even to me although I can't stand statistics (but love math).
To check on the dice bot after a wierd game, I'd suggest copying the chat window log and saving it to Excel. Use the filter feature to get just the dice rolls. Copy them and paste them into another Excel sheet. Use the Text to Columns feature to parse the rolls using fixed width. Count up the rolls with the Pivot Table. I've done this in several matches to see how bad my rolls were. I've usually discovered that mine weren't that much worse than my opponent, I just got bad rolls at the worst times. | |
| |
| Hero of Skirmish
doubtofbuddha
Commander
3371 Posts
| | 04/11/2005 1:01 PM |
| Honestly, I am not really sure I am going to use VASSAL again. I have some pretty decent players to practice against here, and I don't like how perceptions of the board differ between when I am playing on VASSAL as oppossed to in person. I just really wanted to play yesterday, and was kind of surprised at how many crits my opponent rolled, especially considering the last time I had played they were equally bountiful.
| |
I am not gone. | |
|
oldcoast
Sergeant
394 Posts
| | 04/11/2005 2:06 PM |
| I've been playing on Vassal regulary for the last 6 weeks or so and have also had many questions as to Randomness" of the dice roller I don't think the problem is the "Randomness" as applied to the average of mass rolls, but in the "clumping" of numbers, of say rolls of groups of 10. For instance in my tests I often find 6 or 7 out of every ten rolls being very "Low" ( 7,11,1,2,4,2,2,16,9,14) or High (19,9,13,4,10,12,14,20,19,17) this is hugely aggravating in Vassal games where you can consistently roll blocks 1,2,2,3 or while your opponent rolls, 16,20,19,17..I see this all the time.
Also as Zippy can attest I have recently see an inordinate number of double criticals (20,20) one each in the last several games and I actually rolled "2" 20,20 in my last game. (unfortunately one was an initative rolls followed by an attack roll). How many of you that play "in person" often see double 20's rolled for critical hits when rolling "real" dice?
Keep in mind an average Vassal doesn't go 500 rolls so some average calculations should be done based on the average number of rolls in a Vassal match (100-150?, 200?) Overall, it *is* what it *is*,
I don't think there is anything that can be done about it, just pray to dice gods when playing on Vassal like the rest of us.
| |
Check out The Old Coast my online Greyhawk campaign featuring all maxmini's members!, and.........Champion of Aspect of Zuggtmoy | |
|
Zippy
Underboss
1993 Posts
Whitewater, WI
| | 04/11/2005 10:26 PM |
| | In RPG, it's a guy named Neil who seems to get streaks of numbers, high and low, and we all laugh and fart about it. Then the monk triple crits a gargantuan spider and we all laugh and fart again, and drink some beer. It's just that the dice rolls are more rapid-fire in Vassal I think. It really sucks when dice control a game though, I totally agree with that! | |
There are 10 kinds of people in the world; those who understand binary, and those who don't.
Reference Thread, H/W List, Champion of the Catoblepas | |
|
PatEllis15
Commander
4463 Posts
| | 04/14/2005 10:51 AM |
| quote:
My work involves a fair amount of statistics (and my background is in computer science, so I am aware that computers only generate pseudorandom numbers) so I was aware that my few samples on VASSAL were hardly statistically significant. It just seemed a bit off to me to see as many critical hits as I do when I play the game, when on tabletop it very rarely reaches the level that I have seen on VASSAL.
Ah! Jesse you need to get some new dice! [:)]
quote:
Honestly, I am not really sure I am going to use VASSAL again. I have some pretty decent players to practice against here, and I don't like how perceptions of the board differ between when I am playing on VASSAL as oppossed to in person. I just really wanted to play yesterday, and was kind of surprised at how many crits my opponent rolled, especially considering the last time I had played they were equally bountiful.
I hope you'll reconsider that. I've seen the same thing, but It has happened for and against me. I played Zippy earlier in the week, and the dice rolls seems to be going ALL my way, right up until I mentioned that to him, and immediatley I couldn't hit, and he was smacking me around.
While playing with local solid talent cannot be replaced (vassal still takes 50% longer to play than face to face), I suspect that if you find yourself with a couple of hours worth of time at 10:00 at night that you can't just go grab an opponent. Yet Kiddoc, or Dagni, or LynchPT, or Zippy, or who knows who else, whom isn't as familiar with your style of play WILL be there. And seeing other people approach to tile placement, movement strategy, warband synergies, etc. is invaluable. Even if you do loose to a couple of critical's, who cares?! If you did loose to 2 crits in a face to face game, would you discount the game out of hand or just accept that lady luck is not on your side?
Come on back!
Pat E
| |
"Games evolve. Otherwise we'd still be pushing rocks around the dirt. What do you think the cavemen said when some dude showed up with sticks?" - Chairman7w | |
| Hero of Skirmish
doubtofbuddha
Commander
3371 Posts
| | 04/15/2005 3:36 PM |
| You have no clue how often I find myself playing miniatures at 10 pm in the evening, when I really should be sleeping. Especially since I have two other friends who are also prepping for Nationals. ;)
But yes, I will probably try it again, if for no other reason to find random minis players to talk to. :P
| |
I am not gone. | |
|
robbdaman
Underboss
2380 Posts
| | 12/30/2005 8:53 PM |
| I don't think it's so much of a problem that the dice gen isn't rolling enough of a variety over a whole game but rather that code is structured so that it creates strings as stated. Rolling four or five 1s in a row is not unheard of on Vassal and the odds for it doing so in real life is so insanely unlikely that it brings up a flaw in the dice gen. It is flawed because it can't truly be random as the code for it just would be more than a simple mod could handle. Sometimes it sucks sometimes it doesn't but it is flawed, but until someone comes up with something better it's our only option.
R~ | |
Champion of the Titan
******************************************************************************************************************************************************
Successful trades with: Tickparasite, Iyceman, Faragdar The Wise's friend, avrivah, Drakkengi, brucemc, Krush, maniacal_mini_monger, hung4treason, Gandy, NarlethDrider, Kunimatyu, etc, etc, blah, blah, blah..... | |
|
Avatar of Skirmishes
tullywi
Sergeant
982 Posts
| | 12/30/2005 9:57 PM |
| | Sorry everyone, this is a really old topic. I edited the original topic name and text forgetting it would show up here. | |
| |
|
| | You are not authorized to post a reply. |
|
| |
ActiveForums 3.7 | You must be signed in to participate in the
games. |