Dice with character...when is using them cheating?

[MNblockhead]

I actually hate playing on VTT's because (as least on the ones I've used), the results of die rolls are incredibly wonky. Multiple repeated numbers, 8d6 lightning bolts that deal 17 damage, attacks made with advantage that miss multiple times- it's a mess, and the only reason it isn't a complete disaster is that it effects the NPC's as well, lol.

One of my tech-savvy friends claims that it's because generating true random numbers is a hassle and the easier method is just have programs poll the Windows time clock for numbers. I don't know for sure, I just know that I'd rather play with dice than not.

It’s likely that the VTT’s pRNG is more random than your physical dice. The human mind can’t grok true randomness.

Foundry explains how it randomizes its dice rolls and why they should be as random as physical dice. I've pasted their explanation below. The last paragraph is very interesting to me. I don't have enough technical knowledge to ensure I'm reading this right. But it sounds like a set of random numbers are pregenerated and cached when a user connects to a session. So, let's say you are rolling poorly in a session. Could you just log out and re-log in the hope the your next set of random numbers will be better? Since they are all random, it probably doesn't make sense to do so, maybe you are logging out just before a string of good numbers. But I'm curious.

There is also a module to integrate Random.org's true random number generator for Foundry dice rolling. Foundry-TrueRNG/README.md at master · kidfearless/Foundry-TrueRNG. I'm not sure I see much need to bother with this. The argument is that Foundry's method uses pseudorandom number generator (PRNG), which means it produces sequences that appear random but are ultimately deterministic if the seed is known. Random.org (TrueRNG): Uses atmospheric noise to generate numbers, which is genuinely random and non-deterministic. For game purposes, I don't think it matters--though that seems a bit hypocritical for a guy who bought "precision" dice. I guess I'm more about form over function after all.

Mersenne Twister PRNG​

How does Foundry Virtual Tabletop handle randomization?
Foundry utilizes a Mersenne Twister pseudorandom number generator for all of its dice rolls. It was originally developed in 1997 by Makoto Matsumoto and Takuji Nishimura to rectify most of the flaws found in older PRNGs.

It is fast, reliably random over long periods of usage, and easily implemented, which has led to its widespread use in numerous programming languages. In fact, this pseudorandom number generator is the most widely used general-purpose PRNG in existence, and widely viewed as the most reliable for use in dice and other gaming impelementations.

The Mersenne Twister utilizes a seed number that the internal mathematics use to determine the set of random numbers generated. In Foundry VTT, this seed is set at the time each user connects, giving them a unique set of rolls for the rest of that session.

Advanced Dice | Foundry Virtual Tabletop

The official website and community for Foundry Virtual Tabletop.

foundryvtt.com

Finally, this relatively short video gives a good explanation of all of this in less time that it takes to read this post:

 

[Joshua Randall]

You’d be replacing one very large set of random number with a different one, so sure, if that makes you feel better.

 

[Ralif Redhammer]

In my experience, people that are going to fudge their dice rolls aren't going to bother with weighted dice. They'll just roll and make up a number to their liking.

One thing I've realized that pre-3e editions of D&D did right was that sometimes you wanted a high number, sometimes a low number. Rolling a d10 for damage, a higher number is better. Rolling d100 for your Thief abilities, you wanted lower. By unifying those mechanics, even if it makes sense, you made it so that you always want higher numbers.

 

[Duke_]

One thing I would like to say about the (in?) famous Lou Zocchi video is that Mr Zocchi is a salesman selling his product. Such lines as "Your characters are dying because you're using dice that cannot make saving rolls" (5:18 of the linked video) are classic FUD manipulation. Having spent years of my early professional life on both sides of the sales dynamic, this is a huge red flag that the person is trying to manipulate you.

It put so much distrust of his claims into me that I took a few hours and did a little experiment. I no longer have the spreadsheets, but I chi-square tested GameScience, Chessex, Koplow, and Q-Workshop dice using the formulas in that old Dragon magazine article "Be Thy Die Ill-wrought?" (A summary can be found here.) Unsurprisingly, over the recommended sample size of 10 x (numberOfSides), dozens of dice displayed no bias no matter who made them, whether or not they were tumbled, etc.

TL;DR buy dice because they look cool and/or are readable, not because they'll meaningfully impact your roll distribution over the course of a gaming sesion. Because, well, they won't.

 

[MNblockhead]

One thing I would like to say about the (in?) famous Lou Zocchi video is that Mr Zocchi is a salesman selling his product. Such lines as "Your characters are dying because you're using dice that cannot make saving rolls" (5:18 of the linked video) are classic FUD manipulation. Having spent years of my early professional life on both sides of the sales dynamic, this is a huge red flag that the person is trying to manipulate you.

It put so much distrust of his claims into me that I took a few hours and did a little experiment. I no longer have the spreadsheets, but I chi-square tested GameScience, Chessex, Koplow, and Q-Workshop dice using the formulas in that old Dragon magazine article "Be Thy Die Ill-wrought?" (A summary can be found here.) Unsurprisingly, over the recommended sample size of 10 x (numberOfSides), dozens of dice displayed no bias no matter who made them, whether or not they were tumbled, etc.

TL;DR buy dice because they look cool and/or are readable, not because they'll meaningfully impact your roll distribution over the course of a gaming sesion. Because, well, they won't.

True, but I could listen to Lou's self-interested tirades all day. And I always like the gem-like look and feel of game science dice, though the burrs from the molds always bothered me. Not because I was worried about the rolls but it just blemished the look.

 

[Joshua Randall]

If you kept track of all your dice rolls over an entire mult-year campaign, into the thousands, you might notice some “not sufficiently random” results if and only if you studied them with proper statistical tools (such as chi-square). But to make this scientifically valid you’d have to track EVERY die roll without fail. Including the rolls you make while bored. Or the ones that fall off the table. (They may not count in your game, but they count in your scientific experiment.) Every. Roll. Without. Fail.

Per die.

Instead what most people do is “remember” the rolls from a single session, which might be… 20 rolls for a given specific d6 die, let’s say? And they remember the 6s and 1s but not the 2s, 3s, 4s, 5s.

 

[Emerikol]

I'd say more cheaters have special ways of rolling the dice than are using cheaters dice. Buy a dice tower that everyone uses and few cheaters will be happy.

I'm for fair dice though and if I knew certain that dice were unfair then I'd dispose of them. That is hard to verify. I wouldn't buy cheaters dice.

Cheating will also get you banned at best at my game table.

 

[UngainlyTitan]

If you kept track of all your dice rolls over an entire mult-year campaign, into the thousands, you might notice some “not sufficiently random” results if and only if you studied them with proper statistical tools (such as chi-square). But to make this scientifically valid you’d have to track EVERY die roll without fail. Including the rolls you make while bored. Or the ones that fall off the table. (They may not count in your game, but they count in your scientific experiment.) Every. Roll. Without. Fail.

Per die.

Instead what most people do is “remember” the rolls from a single session, which might be… 20 rolls for a given specific d6 die, let’s say? And they remember the 6s and 1s but not the 2s, 3s, 4s, 5s.

This and very much this. People cannot eyeball randomness, this has I believe, been tested. In that screens with random dots were shown to people and they identified them as non-random but also identified non-random images as random.

Back 30 plus years ago when I remembered more of my maths degree than I do now, after a discussion like this I took my random selection of d20'ies and did a set of 300 rolls for each of them, manually tabulated the results and tested for randomness, I had forgotten the name of the test (thanks @Joshua Randall and @Duke_ ) for reminding me and discovered that except for 2 they were all random. I am not sure for what confidence level but knowing me it would have been at least 95% and the 2 that were not random where not that bad.
I concluded that for a ttrpg game they were good enough and never worried about it again.

 

[Joshua Randall]

On the topic of cheating: there was a guy in our Living Forgotten Realms (the 4e organized play program) at the FLGS who claimed he had bad eyesight. He would roll his d20, then quickly pick it up "to look at", while rotating it so a suspiciously large number of high die results were facing him.

It was impressive how smoothly he could do this. I often wondered if he practiced in front of a mirror for hours and hours to get the moves down. Consider that the icosahedron has so many faces that you can only plausibly rotate it to, say, one "circle" of them as you pick it up. (You can't rotate it from the 1 to the 20 -- that would take too long and be too obvious.)

So he had to know, based on what he REALLY rolled, which way to rotate it to get to a high face like 17 or 18 or 19 or whatever.

We eventually called him out on this and he stopped doing it.

 

[Umbran]

But it sounds like a set of random numbers are pregenerated and cached when a user connects to a session.

That's not typically how it is done. A typical random number generator starts with a "seed", which it then folds, spindles, and mutilates until you cannot recognize it, and hands it back to you, usually in the form of a value between 0 and 1. It then takes the results of the first round, and feeds it back into the generator for the second, and third, and so on.

The sequence of numbers is not actually random - if you give the generator a seed today, and the same seed next week, it will hand back the exact same sequence of numbers. But, they are scattered over their range in such a way as a human can't really tell the difference. In mathematics, we call these pseudo-random numbers.

You don't usually cache the results - you run the fold, spindle, mutilate process every time you need a new result. The generators are quick, by human standards.

In many consumer applications (like, say, videogames), the program will grab its seed from the CPU clock of the machine it is running on. There can be limitations to such choices, but they are usually sufficient for our needs.

So, let's say you are rolling poorly in a session. Could you just log out and re-log in the hope the your next set of random numbers will be better? Since they are all random, it probably doesn't make sense to do so, maybe you are logging out just before a string of good numbers. But I'm curious.

They say the seed is set at connection, so logging out and back probably would do so. In other cases, it will depend on the details of coding. You might need to log out, kill your browser session, restart your browser, and log back in to get the code to pick up a new seed.