Was: Claw Poll - Now: Dice vs RNG

[Darkson]

No thanks.

[dode74]

age old programming arguement. you cannot program a random die roll.

Dice: -> http://www.dakkadakka.com/wiki/en/That% ... is_of_Dice

The PRNG:
Edit: Well there was a rather good dissection of the PRNG here but the rather prickly original author has "asked" that it be removed. However, the distribution was shown to be 16.667% or better across the board.

---------------------------------

You will note that the distribution of the PRNG is better than that of the dice. Any argument about the "randomness" of a PRNG is based on something being able to predict the PRNG in some manner because there is enough information to do so. You could also predict dice if you had enough information, but you don't, and unless you have made some sort of predictor then you don't have enough information to predict the PRNG. An RNG doesn't have to be truly random, just random enough to be unpredictable. Dice are simply an RNG we have insufficient information about.

[Daht]

There is a difference between random number generator, and random dice roll.

Even the broadest rng generator is a simulation of an actual random roll.. thats why they use numbers like 4 bazillion rather than say, six.

In running through that 4 bazillioin number, any streak of a given number will be balanced by the end. Any true random sample will not work out exactly equal, just (if big enough) close to it, with repeated test yeilding generally varied but small differences.

The actual noticable effect is minimal to none, but some TT players are a superstitious lot, prone to favorite dice, throwing dice away for bad behavior, or even stealing someone else's dice for a year!

[dode74]

There is a difference between random number generator, and random dice roll.

The actual noticable effect is minimal to none

Philosophical question: if you can't tell that there is a difference, is there one?

[spubbbba]

I thought that unless you use casino ones then real life dice are not all that random either.

The standard ones GW supply are certainly supposed to roll more ones that 16.66% of the time according to some tests i've seen on other sites.

[Jani74]

There is a difference between random number generator, and random dice roll.

The actual noticable effect is minimal to none

Philosophical question: if you can't tell that there is a difference, is there one?

There is no need to bring in philosophy into this. All that matters are exact sciences. We're geeks, after all.

[burgun824]

I thought that unless you use casino ones then real life dice are not all that random either.

The standard ones GW supply are certainly supposed to roll more ones that 16.66% of the time according to some tests i've seen on other sites.

This is true. Standard dice will roll more of a particular type of number than another. They are naturally unbalanced by the way they're made. Unless you're using precision (casino) dice the odds will not be 16.6666666 of rolling a number on a standard die. They will always be skewed towards one and less from another.

This is why I've said I wish they made precision block dice. I'd buy them in a heart beat.

[RogueThirteen]

There is a difference between random number generator, and random dice roll.

The actual noticable effect is minimal to none

Philosophical question: if you can't tell that there is a difference, is there one?

There is no need to bring in philosophy into this. All that matters are exact sciences. We're geeks, after all.

There is no such thing as an 'exact science' independent of philosophy (and certainly no such thing as an 'exact' science). Even the 'hardest' of the sciences (physics, chemistry) are rife with assumptions that are fundamentally philosophical and are justified on philosophical grounds. For instance, any explanation in physics is underdetermined by an infinite amount of alternative explanations, and deciding between them is often a philosophical endeavor, not a scientific one.

[dode74]

What RogueThirteen said. Science is just philosophy with added numbers. The PRNG is at least as good as any standard dice at giving you a fair roll.

[burgun824]

What RogueThirteen said. Science is just philosophy with added numbers. The PRNG is at least as good as any standard dice at giving you a fair roll.

If programmed properly it's better then a standard dice. But I think this is where "conspiracy theorists" have a hard time with it. They just simply don't trust that it hasn't been programmed to F you. I think it stems from how streaky they can "seem" with how they spit out your die results. On Cyanides game when I play it definitely feels like I have a stream of rolls that result in 1 and other times I can't seem to be stopped with a group of 5's & 6's. Overall I'd say they roll around 16.6666 for all results but I could see how it comes down to the feel of how they run together at times.

[RogueThirteen]

Another thing that might make people suspicious is that I've heard the computer's rolls (for an AI opponent) seem to depend on the difficulty setting. For instance, on Hard apparently the computer makes many more of it's rolls than it does on Easy. I've only played on Hard very briefly, so have no idea if this is true or not. I do know, however, that the AI opponent seems to almost never roll less than a 6 when a dodging (anecdotal, of course, but I've regularly seen AI Gutter Runners roll four or more '6's in a row to get past defenders, also rendering Diving Tackle worthless most of the time). Additionally, and also anecdotal, I'd estimate that I'm knocked down three times as often by a 2-die-MY-choice block thrown against me by the AI than I actually get to pick an "attacker down" on a 2-die-MY-choice block thrown against me.

Again, small sample size and purely anecdotal, but it does seem like the AI rolls unbelievably well sometimes (though it's still incredibly stupid) and lots of players seem to have noted that AI die rolls depend heavily on the game setting. Does anyone know if this is actually the case?

Because my guess is most complaints or suspicions about the die rolls are leveled against the AI, and NOT against the die rolls while playing other human opponents.

[Daht]

Philosophical question: if you can't tell that there is a difference, is there one?

That depends on how many ladders you walked under on a given day.

[Daht]

This is why I've said I wish they made precision block dice. I'd buy them in a heart beat.

Precision dice are overrated, in that once you use them a few times they are no longer precision dice (smallest wear at all strips them of their 'precisionness') most casino dice don't make it thru a single shift of play before going to the trash/recycle bin. They are mostly a product of casinos not letting folks use outside dice, and being 'approved' by the gaming commision for use on gambling tables.

At the price of block dice GW could double their revenue on gamers maintaining sets of precision block dice!


The real difference in RNG and actual dice is actual dice are actually random, RNG is not. No human brain is going to precisely know a 4 quadrillion long string of results well enough to predict a given sequence in the RNG, so it's apparently random, and where you fall in the string can affect how many 1's and 6's you get in a given game..

In terms of results there is very little difference, if any that a person could honestly differentiate.... just like there is no mechanical difference playing with unpainted 2nd edition human plastics with posted-note numbers, and a professionally painted custom team of minis.

[dode74]

The real difference in RNG and actual dice is actual dice are actually random, RNG is not. No human brain is going to precisely know a 4 quadrillion long string of results well enough to predict a given sequence in the RNG, so it's apparently random, and where you fall in the string can affect how many 1's and 6's you get in a given game..

Dice are as deterministic as any physical process. With enough information and computational power you could predict real dice rolls quite easily. Nothing outside of the quantum is "really random," and even that is far from certain.
If you can't predict it (and you can't), and the distribution is as expected (and it is), then there is, in terms of using it as a substitute for dice, at least as good as dice. If I were to present you pairs of strings of numbers, one string die-rolled, the other RNG produced, you would not be able to tell the difference. Differentiating between two things which you cannot actually tell the difference between based on some undefined notion of what is and is not random is splitting Schrodinger's cat's whiskers.

[Darkson]

Didn't someone once post a 2 strings of numbers, and ask people to pick the rolled list from the RNG list, and the majority got it wrong (the RNG list loked more "random", so the majority assumed that was the dice roll list).

I can't remember where I saw it posted, else I'd link to it.