|
I got some dice from GameScience.com and I made a graph to see if what he said is true. guess which one is the new set.
Data.
I'm not the sharpest bulb in the forest, so forgive me. But what am I looking at in that graph?
I'm not the sharpest bulb in the forest, so forgive me. But what am I looking at in that graph?
I think:
He made 10 rolls with every die. He divided the results into those from 1-10, and those from 11-20.
So the green die rolled 4 times in the 1-10 range, and 6 times in the 11-20 range. And so on. So the only ones that came up exactly even were the clear and the rainbow dice.
Is this a correct interpretation?
I'm not the sharpest bulb in the forest, so forgive me. But what am I looking at in that graph?I think:
He made 10 rolls with every die. He divided the results into those from 1-10, and those from 11-20.
So the green die rolled 4 times in the 1-10 range, and 6 times in the 11-20 range. And so on. So the only ones that came up exactly even were the clear and the rainbow dice.
Is this a correct interpretation?
If it is, it doesn't mean anything... Need to toss them a heck of a lot more than 10 times each to have any significant results.
|
If it is, it doesn't mean anything... Need to toss them a heck of a lot more than 10 times each to have any significant results.I'm not the sharpest bulb in the forest, so forgive me. But what am I looking at in that graph?I think:
He made 10 rolls with every die. He divided the results into those from 1-10, and those from 11-20.
So the green die rolled 4 times in the 1-10 range, and 6 times in the 11-20 range. And so on. So the only ones that came up exactly even were the clear and the rainbow dice.
Is this a correct interpretation?
Indeed, and the cone/bars need to go away. To avoid creating a misleading graph, bars should be the same width for their entire length, which is why rectangles are so popular. A 50/50 cone gives a lot more volume to the bottom portion, despite the fact that they should be the same.
RPG Superstar 2010 Top 16
|
Catman,
There isn't a hard-and-fast threshold for how many rolls are appropriate. What you're trying to do is check to see whether an run of actual data, from rolling real dice, matches up with a hypothetical model (1/n probability of each fact of an n-sided die). The standard mathematical tool to do that is something called a chi-square test.
As I said, the more rolls, the better, but for an n-sided die, you'll get trustworthy results if you make n times n+1 rolls. (For a 6-sider, 42 rolls. For a 12-sider, 156 rolls.) And keep track of your actual counts per result. (So, none of this low-high stuff. If you're checking a 20-sider, and it rolls "15" five times for every "16", the low-high data won't reveal that bias.)
There's a pleasant little formula to use, or you could just drop the results into this on-line calculator and it will determine just how suspicious you should be of the results.
(This is my gripe with Lou Zocci's convention presentation. Who cares if the other guy's dice aren't perfectly symmetric? The only way to tell how fairly they roll is through a chi-square test.)
If it is, it doesn't mean anything... Need to toss them a heck of a lot more than 10 times each to have any significant results.
Yeah, I was going to find out if I was reading it right before I mentioned that.
|
My players are all now reinterested in Zocci dice after I explained Gamescience to them, and we went to meet Lou in person.
For those who may not know, this guy was once the only person distributing D&D dice in the U.S. He is a legendary contributor to the industry, and a pleasant guy to talk to.
As for the statistics, 10 rolls of anything aren't a large enough sample. Gamescience's test used many thousands of rolls.
For those who may not know, (and this is not an attempt to persuade) the typical 90s and 00's dice we've been using are twice-polished. This means they go through a tumbler to smooth edges so they "roll". They roll much farther, and in many ways are more comfortable to hold and sometimes more fun to roll.
Lou isn't saying tumbled dice are bad. However, he does explain (and you can watch this easily on YouTube) that the precision edges of Zocci dice are akin to those used in Las Vegas. In gambling, for example, there is a reason dice must be precise, "to give you equal access to all sides."
In the double-tumbling process of other dice, some sides can become unevenly worn, as compared to other sides of the same die. Thus, some dice do indeed roll better than others, or in some ways skew the results.
Now, Gamescience, or Chessex precision edge dice, are designed not for comfort or appearance (although many prefer these) yet they do tend to give better "equal access to all sides."
In a joking way, you might say that for circumstances when you really, really need to make that saving throw, don't trust your character's life to any average die. A Zocci die will give you equal access to all sides. And, thus, from a probability stand point might actually improve your chances to make your save. That is, with double-tumbled dice, there is some minor influence to the sides of the polyhedral.
So, just in case anyone was wondering, just thought I would supply that information. What Lou says doesn't mean there's anything wrong with using whichever dice you prefer. Yet, it is an interesting selling point to avid gamers. I use Zocci dice exclusively now, and actually prefer the classic feel to the dice, and appreciate some of the minor science involved, even if the difference is minor. They look great, and even the chipped caster cuts provide an endearing and charming appearance. The translucent high-quality material is also a big draw for me.
Just my 2cp.
-Pax
I like dice with sharper edges, but that's mostly because I'm so tired of the smooth ones rolling straight off the table on anything more than a gentle roll. Drives me crazy.
