Neal Litherland |
LazarX |
That's why I use gamescience precision dice!
Dice roller app would be more randomized, but too slow for actual play use in my opinion.
Dice roller apps and software are actually not randomised AT ALL. They simply work on a seed number which is usually how many seconds have passed since last midnight.
LoneKnave |
4 people marked this as a favorite. |
You may as well say that real dice aren't randomized, since they fall based on the forces affecting them, which you can control.
Also, dice rollers that use screen swipes/shaking the device as the random element exist (or they could potentially anyway, I haven't looked into their code) and should be about as random as rolling real dice. Also same speed to use.
justaworm |
Protoman wrote:Dice roller apps and software are actually not randomised AT ALL. They simply work on a seed number which is usually how many seconds have passed since last midnight.That's why I use gamescience precision dice!
Dice roller app would be more randomized, but too slow for actual play use in my opinion.
Sort of true. There is no such thing as true "randomization" in a random number generator, but there are way more options and science behind picking the system clock as your seed.
Atarlost |
LazarX wrote:Sort of true. There is no such thing as true "randomization" in a random number generator, but there are way more options and science behind picking the system clock as your seed.Protoman wrote:Dice roller apps and software are actually not randomised AT ALL. They simply work on a seed number which is usually how many seconds have passed since last midnight.That's why I use gamescience precision dice!
Dice roller app would be more randomized, but too slow for actual play use in my opinion.
This isn't actually true. On most linux implementations, /dev/random uses environmental noise as an entropy source to re-seed a pseudorandom number generator every time it approaches its period. Apple's posix derived OSs including iOS only have pseudorandom number generators, but wikipedia doesn't mention Android one way or the other.
If you have a linux laptop at the table for SRD reference you have easy access to a truer random number generator in the laptop than in your dice cup. There should be more than enough noise coming off the wireless adaptor to provide all the entropy you'll ever need.
Charon's Little Helper |
If you have a linux laptop at the table for SRD reference you have easy access to a truer random number generator in the laptop than in your dice cup. There should be more than enough noise coming off the wireless adaptor to provide all the entropy you'll ever need.
That implies someone having a linux laptop.
Why would someone ever do that to themselves? :P
rknop |
1 person marked this as a favorite. |
I love my Linux Laptop!
In any event: the whole "computer random numbers aren't really random" is a gigantic red herring.
Sure, yes, if you feed the pRNG the same seed, you will get the same sequence of numbers.
However, think: for game purposes, what do we really want out of our "random" numbers? True actual randomness? Or, really, unpredictability, and uniform distribution? The latter two are what we're really after. And, computer pRNGs are going to be way (*) better at this than the actual dice most of us use. So, the fact that pRNGs aren't purely random, the unpredictability of the seed means that this has absolutely zero implication for RPGs, and should not cause anybody any concern whatsoever when it comes to their use.
(*) Well, "way". The truth is, the number of times most of us are rolling our actual dice, they're just fine. We're not really going to see the differences. The differences we think we see are our brains evolved tendency to over-interpret noise in the environment and find patterns.
rknop |
1 person marked this as a favorite. |
I have to admit to being very surprised at the result of the linked study though. 29% of rolling a 1 on Chessex dice? I'm extremely dubious of that result; it violates my reasonableness-meter. I'd want to look very closely at the methodology of that study, and see a replication somewhere.
19% maybe I could believe. Not 29%.
Undone |
I have to admit to being very surprised at the result of the linked study though. 29% of rolling a 1 on Chessex dice? I'm extremely dubious of that result; it violates my reasonableness-meter. I'd want to look very closely at the methodology of that study, and see a replication somewhere.
19% maybe I could believe. Not 29%.
Actually while my experience isn't quite so bad it's definitely right. I no longer use pipped d6s as a result of it.
Mudfoot |
That implies someone having a linux laptop.
You can use a Raspberry Pi instead. A wifi dongle and a few LEDs gives you a near-as-dammit perfect RNG smaller than your dice bag for maybe $40. You could write a daemon to read a comparator on a noise diode and write to /dev/random to make it even better if you're desperate.
I've bought my 7-year-old son a laptop for Xmas. It comes with Win8, but I'm not cruel so I'll probably put Mint on it instead. The only meaningful problem I've found with Linux recently is PF Online doesn't support it, and my son is too young for that.
Steve Geddes |
I have to admit to being very surprised at the result of the linked study though. 29% of rolling a 1 on Chessex dice? I'm extremely dubious of that result; it violates my reasonableness-meter. I'd want to look very closely at the methodology of that study, and see a replication somewhere.
19% maybe I could believe. Not 29%.
Me neither. It's hard to tell without the actual data, although the study sounded pretty thorough.
The reference to spring semester raised a red flag - I'm not in the US, but does that include early April?
BigDTBone |
Protoman wrote:Dice roller apps and software are actually not randomised AT ALL. They simply work on a seed number which is usually how many seconds have passed since last midnight.That's why I use gamescience precision dice!
Dice roller app would be more randomized, but too slow for actual play use in my opinion.
Good dice apps will take the last 3 digits of a MHz clock. It doesn't yield a perfectly even distribution (ie 6 doesn't divide into 1000 an even number of times) but it is a very good mean.
This method yields a result very close to random.
Oladon |
rknop wrote:I have to admit to being very surprised at the result of the linked study though. 29% of rolling a 1 on Chessex dice? I'm extremely dubious of that result; it violates my reasonableness-meter. I'd want to look very closely at the methodology of that study, and see a replication somewhere.
19% maybe I could believe. Not 29%.
Me neither. It's hard to tell without the actual data, although the study sounded pretty thorough.
The reference to spring semester raised a red flag - I'm not in the US, but does that include early April?
Note that later on in the (original) thread, Yade (the original poster) describes his rolling method and admits that he suspects it to have significantly influenced the rolls. He eventually disappears from the thread, but in the meantime a number of other posters attempted to reproduce his results and failed. The general consensus is that something about his rolling method (a giant box with a compartment for each die) unduly affected the results.
Here, on the other hand, is a study of two d20s (note: that's way too small a sample to draw conclusions about "Chessex" vs "GameScience") that's well-executed and well-documented, as well as published in its entirety (all data included).
BigDTBone |
1 person marked this as a favorite. |
So, after reading the write up from the OP, I think the guy was just an idiot. I have severe reservations about his ability to design and conduct a valid survey. He even goes so far as to express his complete ignorance about the entire field of statistics at the end. This study practically begged for a chi squared test but instead we get, "well uh gosh it's possible to flip a coin 100,000 times and not get heads, but that doesn't mean that heads doesn't exist."
Blarg.
Astral Wanderer |
Well, by exploiting a minimum of physical conditions, you can roll all 6s you want (or any other number, and on any kind of die). Just try letting a die fall always in the same way, starting from always the exact same position and facing, on the same surface at the same distance. The less variables you introduce (mainly the small shakings a hand does naturally), the more you'll get the same result.
BigDTBone |
Well, by exploiting a minimum of physical conditions, you can roll all 6s you want (or any other number, and on any kind of die). Just try letting a die fall always in the same way, starting from always the exact same position and facing, on the same surface at the same distance. The less variables you introduce (mainly the small shakings a hand does naturally), the more you'll get the same result.
Which is why if you are testing randomness then you should minimize that sort of bias by insisting that valid rolls must travel a minimum distance from the throw point, dice should be thrown from a cup, dice should be required to bounce off both a horizontal and vertical surface before coming to a rest. Also, the die should have adequate room to roll without bouncing off too many surfaces. Hard surfaces are preferred to felted surfaces. The rolling surface should be a level as possible and should be resistant to environment vibrations.
The test in the OP only had a few of those going for it.