A YouTuber Makes A One-In-Nine-Million Dice Roll! (Or So He Says)

This YouTuber did an experiment where he rolled each of six dice in sequence, with the goal of getting the same number on each roll. For those not familiar with Dungeons & Dragons, we use dice with different numerical values: one four-sided die (a d4), one six sided (d6), a d8, a d10, a d12, and a d20. His goal was to roll each of them in sequence with the goal of getting the same number (4) on each roll. His Instagram followers crunched the numbers and gave him the odds, something like one in 498k or thereabouts, and told him to plan on doing this over the course of several days. As it turns out, he pulled it off after about 3.5 hours. The video begins with his successful d4 roll and then, when he gets his 4 on the d20, he loses his shit. Then, for shits & giggles, he rolls another d20 and gets another 4!!! Odds of this entire sequence were like one in nine million or something. Pretty impressive.

I’m not going to claim that his rolls were “random” in the sense that would be required of a rigorous scientific experiment. His dice are obviously not machined to the precision that would be required of dice in a Vegas casino, for example, and his methodology didn’t necessarily ensure pure randomness. If he submitted his results to an academic paper it likely wouldn’t survive peer review. But impressive nevertheless.

Looked like some were being spun, not rolled. No way to know if the die or the surface were magnetized. Were there independent observers who could verify this was in one session, examined all the equipment, etc?

Well, no, but I doubt any chicanery was involved. Off-the-shelf equipment and unscientific methodology, yes, but fraud? Probably not. He’s done similar “challenge” videos with his dice that he’s live-streamed and they’ve taken 8-9 hours. His channel isn’t about “Look at this cool thing I did!”

According to this video his original challenge was to roll the highest value on each. He added the “roll all fours along”. And the probability of rolling all of a predetermined sequence on these dice is just 1/460k.

I’m not going to watch more to see exactly what his setup was and where the 1 in 9 million come in, but I suspect it came from just going “Oh, I got all of these. I’m going to roll the d20 again just because.”

He says his followers calculated the expected time to succeed was originally 5 days, but I don’t know if that was just for the 4-6-8-10-12-20 challenge or if it was adjusted for the 4-4-4-4-4-4 addition, which doesn’t make it all that remarkable that he happened to do it in 3.5h. That is if you think they made all the correct assumptions in the first place when coming up with 5 days.

I was confused on this. The guy claims that the odds off 4-4-4-… are the same as 4-6-8… because it’s still one of n numbers on each die. I’m not skilled enough at math to determine if there was a difference.

I think the “five days” calculation was something like 45 seconds per sequence (picking a number out of my ass) times 468k rolls.

This guy has a YouTube video series in which he does nothing but roll dice? And people subscribe to this? How much money does he make?

(Now I’m trying to imagine what silly thing I can do on YouTube to make easy money.)

No. He has a YouTube series where he makes D&D content. His second channel (where this challenge took place) is content like this. As to how much money he makes, he seems to live a lower middle class lifestyle in suburban New Jersey, based on the shots of his apartment and his ride.

The math checks out. Any fixed sequence of single results has the same odds. 4-6-8-10-12-20, 4-4-4-4-4-4, 1-5-2-1-3-9, they’re all 1/(4681012*20). But if you start allowing more than one sequence your chances go up.

That should have been 1/(4*6*8*10*12*20), but pairs of asterixes were interpreted as formatting code.

Isn’t every dice roll (with that set of dice) a one-in-nine-million roll?

One in four hundred and some thousands. And only the ones where which die rolled what matters. If you only care about the set of six numbers combinations with many low numbers have higher probabilities. E.g With 4,4,4,3,3,3 an of the dice could be any of those digits. With 4,4,4,10,10,10 the last three would have to be rolled with the d10, d12 and d20.

4-4-4-4-4-4 is the same odds as 4-6-8-10-12-20. But 4-4-4-4-4-4 OR 3-3-3-3-3-3 OR 2-2-2-2-2-2 OR 1-1-1-1-1-1 OR 4-6-8-10-12-20 is five times likelier than 4-6-8-10-12-20. This is a key point in statistics: You can’t just look at the odds of any one event; you have to look at the odds of that event and any other event that’s at least as significant.

Or, since “significance” can be imprecise to pin down after the fact, if you want to be rigorous in determining the probability you specify the set of successful outcomes before you start.

Very improbable events like that occur continuously. Grab 40 d20s and roll it. The particular result is a quite improbable event.

He also isn’t going to roll all the dice on every attempt. If the first one doesn’t come up four, he just rolls it again. Once he gets a four, he rolls the next one, and if it’s not a four he starts again. The same with the rest. That speeds up the process.