I'm new to vassel and I have about a 50/50 win loss ratio and I can't help but think that the dice app absolutely stinks. The games I've won I have felt the dice were mostly in my favour by a big margin and the games I've lost my opponents have said stuff like Getz the dice don't like you.

I have had a game where I had 20 shots and all but one was with 3 dice shots and managed to get 5 hits and then the next turn about the same. Granted not all the shots got modified but some did and it still rolled rubbish.

Is this just me or is this a common thing ?

It's just you.

You can copy everything in the chat log from vassal, and paste it into Sozin's Lady Luck site to see the the dice statistics from your game. it give you a ton of information, like expected dice results vs actual results. After reviewing a few matches you'll probably fond that the vassal dice are fine.

http://sozin.pythonanywhere.com/about

Everyone complains about the dice on Vassal, but there's no evidence that the dice are loaded in any sense--and people have looked pretty hard for that evidence, including me.

Basically, the dice are fine in a mathematical sense. You still might get squashed by the dice in a particular game,, but it's not Vassal's fault.

I believe it's still random, but I've wondered if there are times it just wants to fail.

Seen quite a few one sided games where red dice would seem to favor one over another, and same with green dice

While I do believe it is completely random, I wonder at times

When you lose the dice are against you and when you win they were hot, that's down to how our memory works which is very selective and totally biased.

You will of course remember bad rolls in a game you lost and because your mind alters your memory you'll remember it being worse than it really was.

What Vorpal said. People have done some studies on large numbers of dice rolls, it was a big discussion at one point. Since then, long time Vassal players have come to see the dice results are acceptable. I have a few thoughts on why the discussion comes up when people start using Vassal X Wing:

1. General Skepticism from using a new system, especially one that the player hasn't personally investigated. I definitely can understand this, as I have heard of enough supposedly "random" systems not actually being random. However, the studies seem to clear this up pretty well.

2. Vassal shows and stores the dice log in chat, as well as keeping the actual result visible for longer times than on the table. Heck, on the table, you might only see the dice result for a very brief time as players tend to roll the dice and pick them up pretty quickly, not giving your brain much time to memorize the result. On Vassal, you are reading the results and viewing the results for a much longer time, meaning you are memorizing the results better. This allows us to dwell on the positive and negative results better. I know I've had plenty of games on the table and on Vassal where dice were pretty skewed. I just remember quite a few more details from my Vassal games, and I can understand why. I don't really think there is any difference in tabletop vs Vassal, we just remember the vassal results better.

Hopefully that helps make you feel a bit better about it!

Putting in your dice results to Sozin's dice app is certainly a good idea too. From my experience, the dice results end up being more even than what I thought in the game. Out of the last 10 or so games I've played on there, only my most recent one ended up being heavily in my favor.

I still believe it's random, and I believe those who say they have ran tests and found no evidence of the dice being one sided.

Even so the other week I had a game and man did I have the worst rolls

Even my opponent was rooting for me lol.

It was so bad that it was amusing lol.

Even the lady luck results he said were the worst he had seen (I'm still not sure how that works)

I've had games where I felt pretty bad for my opponent where he can't catch a break, but it always comes back around.

Vassal dice are like any dice I guess. I've seen it on the table where I near threw my dice out the window.

It's the green dice I think are just plain terrible vassal or table

I personally get the feeling that computerized random dice rolls are fully random, but that they're differently random from player rolls. That is to say, player rolls are not truly random in the same sense that computer generated numbers are. You are influencing the dice when you're rolling them, by the way that you pick them up, hold them in your hand, etc. I wouldn't be surprised if dice rolling is less random with humans. The other thing is that random events have a tendency to go on streaks at times - lots of losses and lots of wins, and there has been some mathematical modeling of this, I'm sure, but I haven't familiarized myself with the literature. Nonetheless, I also wouldn't be surprised if humans have shorter streaks, owing to the input we're giving the dice.

So, it's partly a perception issue, but I think real dice rolling behaves differently from computerized dice rolling over long intervals.

Ok cool ,I'm happy others have questioned the dice too and it's just not me. Like with the real game I'll just keep plugging away

@OP: So basically, you're saying the the Vassal dice app is every bit as low-down, cheating, backstabbing, biased, unfair, demon-possessed as a set of real dice is?

I personally get the feeling that computerized random dice rolls are fully random, but that they're differently random from player rolls. That is to say, player rolls are not truly random in the same sense that computer generated numbers are. You are influencing the dice when you're rolling them, by the way that you pick them up, hold them in your hand, etc. I wouldn't be surprised if dice rolling is less random with humans. The other thing is that random events have a tendency to go on streaks at times - lots of losses and lots of wins, and there has been some mathematical modeling of this, I'm sure, but I haven't familiarized myself with the literature. Nonetheless, I also wouldn't be surprised if humans have shorter streaks, owing to the input we're giving the dice.

So, it's partly a perception issue, but I think real dice rolling behaves differently from computerized dice rolling over long intervals.

Human rolls and computer rolls, when both are performed properly, are both as random, and are the "same random". If there was anything different about them, Vegas would know.

Your part about streaks is misleading. Given a certain random event (die roll, coin flip, card draw, etc.) performed repeatedly, you can calculate many things regarding "streaks": the odds of finding a "streak" of a certain length, the average length of the longest streak, or how many rolls, on average, it'll take to get a streak of a certain length.

This is to say that you can answer questions like "how many times will I need to roll a die, on average, before getting 25 hits in a row?"

What you absolutely cannot use the math for is to say "I just rolled 7 hits in a row, I must be on a streak, this next roll is therefore more likely to be a hit!"

The "input your giving the dice" has no impact on the outcome, unless you're rolling poorly. Once the die bounces off the table a couple times, your initial "input" is lost in the chaos.

You could also argue that you are giving input to computer rolls. The initial seeding of the pseudo-random number generator used in many applications is often based on a timestamp. So, the exact moment the generator is initialized impacts the numbers generated. If the generator is initialized when you click "new game", you're influencing the outcome!

So, it's ALL a perception issue, computer dice and real dice perform identically over long intervals.

I personally get the feeling that computerized random dice rolls are fully random, but that they're differently random from player rolls. That is to say, player rolls are not truly random in the same sense that computer generated numbers are. You are influencing the dice when you're rolling them, by the way that you pick them up, hold them in your hand, etc. I wouldn't be surprised if dice rolling is less random with humans. The other thing is that random events have a tendency to go on streaks at times - lots of losses and lots of wins, and there has been some mathematical modeling of this, I'm sure, but I haven't familiarized myself with the literature. Nonetheless, I also wouldn't be surprised if humans have shorter streaks, owing to the input we're giving the dice.

So, it's partly a perception issue, but I think real dice rolling behaves differently from computerized dice rolling over long intervals.

Human rolls and computer rolls, when both are performed properly, are both as random, and are the "same random". If there was anything different about them, Vegas would know.

Your part about streaks is misleading. Given a certain random event (die roll, coin flip, card draw, etc.) performed repeatedly, you can calculate many things regarding "streaks": the odds of finding a "streak" of a certain length, the average length of the longest streak, or how many rolls, on average, it'll take to get a streak of a certain length.

This is to say that you can answer questions like "how many times will I need to roll a die, on average, before getting 25 hits in a row?"

What you absolutely cannot use the math for is to say "I just rolled 7 hits in a row, I must be on a streak, this next roll is therefore more likely to be a hit!"

The "input your giving the dice" has no impact on the outcome, unless you're rolling poorly. Once the die bounces off the table a couple times, your initial "input" is lost in the chaos.

You could also argue that you are giving input to computer rolls. The initial seeding of the pseudo-random number generator used in many applications is often based on a timestamp. So, the exact moment the generator is initialized impacts the numbers generated. If the generator is initialized when you click "new game", you're influencing the outcome!

So, it's ALL a perception issue, computer dice and real dice perform identically over long intervals.

Right, but what about streaks? And what about short intervals? It may all be a wash in aggregate, but it may not be in experience. If I have a program that flips a coin, and it gives heads the first ten times and tails the next ten times, alternating, over the long haul that's the same thing as a true random coin flip, but the behavior is vastly different. So, if online dice rollers do tend to "streak" differently, it's worth looking at.

Right, but what about streaks? And what about short intervals? It may all be a wash in aggregate, but it may not be in experience. If I have a program that flips a coin, and it gives heads the first ten times and tails the next ten times, alternating, over the long haul that's the same thing as a true random coin flip, but the behavior is vastly different. So, if online dice rollers do tend to "streak" differently, it's worth looking at.

"over the long haul that's the same thing as a true random coin flip, but the behavior is vastly different."

Apart from the fact that the average is the same, those two distributions are worlds apart!

I agree that yes, it could be worth analyzing VASSAL's random number generation is coded properly. (Just a note that this would likely be in the VASSAL source code, not the XWing module's.)

When I was young, I wrote a program that would draw a random line from the centre of a circle until it reached the circumference (equal chance of stepping up/down/left/right). Because of how I setup the random number generation, I noticed a weird repeating pattern in the lines it was generating. Not sure why I'm sharing this story anymore, just smile and pretend I said something funny.

Hahaha!

When I'm not studying quals, I'll try to get through some actual mathematical papers on this kind of thing.

Right, but what about streaks? And what about short intervals? It may all be a wash in aggregate, but it may not be in experience. If I have a program that flips a coin, and it gives heads the first ten times and tails the next ten times, alternating, over the long haul that's the same thing as a true random coin flip, but the behavior is vastly different. So, if online dice rollers do tend to "streak" differently, it's worth looking at.

One of the tests I did on Vassal's dice roller was to look for correlations between adjacent rolls (that is, whether it's a "streaky" generator) and to look for average streak length. Even over the short term, its behavior was appropriate.

Also, the Vassal source code uses java.security.SecureRandom for dice calls, so finding streaks in an implementation would be a really big problem.

Thanks, Vorpal! That's helpful. I've seen a lot of random dice generators online that do tend to streak more than I would rolling the dice in person, so it's something I've thought of. Never dd a mathematical evaluation though, so I appreciate the effort.

Additionally, in an effort to increase randomness in the x-wing vassal dice the module curator(s) did 2 things.

1) The dice have 16 faces instead of 8. basically all face types were doubled.

2) Each die is "unique" Die 1 might have its first face be a blank, and Die 2's first face is a focus.

Since each die has a unique ordering of its appropriate faces, any streaking output by the random number generator (RNG) is evened out. A series of three 1s from the RNG instead of being 3 blanks, results in [blank] [boom] [focus]

Additionally, in an effort to increase randomness in the x-wing vassal dice the module curator(s) did 2 things.

1) The dice have 16 faces instead of 8. basically all face types were doubled.

2) Each die is "unique" Die 1 might have its first face be a blank, and Die 2's first face is a focus.

Since each die has a unique ordering of its appropriate faces, any streaking output by the random number generator (RNG) is evened out. A series of three 1s from the RNG instead of being 3 blanks, results in [blank] [boom] [focus]

That doesn't actually help much if you ask me.

It only helps if your RNG is likely to generate the same number multiple times (once it rolls a 4, it's more likely to roll another 4). It doesn't help if, after the RNG rolls a 4, it's more likely to roll a 6. And both of those cases are problematic.

Basically, streaking (getting multiples of the same result in a row) is a specific instance of the more generic problem of having previous results affect future results.

IIRC there are 350k rolls in ladyluck; the last time I checked, the distributions looked good. But if someone wants to really analyze 'em for reals (I'm more a data dude and less a stats dude), send me a PM and I'll give you dataset with all the dice results of the 1,800+ games recorded in ladyluck.

IIRC there are close to 1MM rolls in ladyluck; the last time I checked, the distributions looked good. But if someone wants to really analyze 'em for reals (I'm more a data dude and less a stats dude), send me a PM and I'll give you dataset with all the dice results of the 1,800+ games recorded in ladyluck.

How is the dataset formatted? If I don't have to do a bunch of cleaning, then running it through SAS to compute some descriptive statistics would be a fairly trivial task.

It is pretty normalized. I'll post a link to the dataset once it finishes downloading :-) In terms of data cleaning, the only issue is that there will be some dupes (sometimes both players submit the games to ladyluck); the dataset has some hints on whether or not something is a dupe.

Vorpal, here's a link to the dataset. I was off by an order of magnitude in my estimate; its 100k+ rows :-)

A visualized datamodel is here; in plain English, here is the idea: games are composed of a unique id ('game_id'), a date/time it was played ('game_date'), and dice throws. A dice throw has a unique id ('throw_id'), a player throwing the dice ('player_name' ), and a set of numbered dice ('dice_num'). Each dice rolled has a type of roll - ATTACK, DEFEND, REROLL, or CONVERT, a color - RED, or GREEN - and a face result ( HIT, FOCUS, etc etc).

If I throw some attack dice, and you throw some defense dice, then both those throws share the same attack set ('attack_set_num'); a visualization of attack set concept can be found in the game tape tab of a ladyluck game.

Duplicates can be disambiguated by the game_date-player_name data pair; for example, below you see Paul and starslinger72 both submitted their TC Aces league game at the same time (3/23/2015: 1:53). There aren't a ton of duplicates, so they can either be ignored or cleaned out by looking for games that were played on the same day, have the same set of players, and same count of throw_ids.

Vorpal, here's a link to the dataset. I was off by an order of magnitude in my estimate; its 100k+ rows :-)

Thanks. I don't know when I'll get to this--counting results is easy, but I have to figure out what stats will be meaningful, figure out logic for removing duplicates, and then actually run the code--but when I do I'll put the results up here... and maybe even at TC, too.

...1,800+ games recorded in ladyluck.

Off topic post of the day: that's pretty cool!