its actually the reason i stopped watching the video. insulting people for not agreeing with you is never a good look. it wont convince anyone to come to your side, and i dont believe someone who has been lied to is an idiot.

@@arcanealchemist3190 That is true...but people should be more concerned about being wrong than about not being insulted. If I tell anyone my theory how I think Baseball works..everyone who actually understand the game will probably think. I'am an idiot and maybe even tell me...especially if I come from a position that all the people who claim that Baseball doesn't work like I think it does are payed liars. And If I listen to them I might actually learn how it works..and If I don't I stay ignorant. I agree that insulting people isn't. helpful...but it also isn't helpful to pretend someone is smart or has a good point or understands something if he clearly doesn't.

@@arcanealchemist3190 The interenet has not helped with the number of insults either. Its good and healthy that you turned off at that point. But some people enjoy a well time piece of vitriol and see the humour in it, and KZfaq likes divisive content, so I'm going to be content with my self-restraint that 90% of my content is not tempted by the algorithm.

Yes, the psychology of randomness is very interesting. I spent the weekend teaching Steve Jackson's OGRE, a game in which players feel a strong psychological pull to play poorly (by increasing the odds ratio on attacks, which makes that single dice roll more likely to succeed but reduces the overall effectiveness of their attacks).

@@Lessons_from_the_front And i really enjoy playing games with dice and randomizers, but I have no interest in gambling, which is for me an entirely practical thing. I don't mind losing. I hate losing anything of value, even money.

depends on if you are looking at the expected faces, or the expected value. imagine the faces are random arbitrary symbols and not digits, do you still think that three wugs and three forps is equivalent to one roll of each?

That would only be significant if those are the only 6 rolls. Otherwise you get xkcd 882 style publication bias problems.

Say everyone follows your reasoning. If 1000 people play the game, on average 1 will wrongfully conclude that their friend is scamming them. So, when human activities are involved, i think the better question is not the statistical significance, but rather the amount of people experiencing it. This still results in a significance threshold, but a very lower one. As Matt Parker says in his "how lucky is too lucky" video (min 29:00), we can define the measure of how many times it will happen if all of humanity did that activity once each second for a year, and you can see that a lot of wrong conclusions would be reached. Also, for human activity, a threshold p-val of 5% is very high: it means that 1 out of 20 people will reach the wrong conclusion. I also want to cite the bayes theorem P(A|B)=P(B|A)*P(A)/P(B) where A="dice are rigged" and B="6 times 6". P(B) = 1/(6^6)= 2e-5. Say we give a prior P(A) that 1 in 1 million friends would rig the dice, and say that rigged dice have P(B|A)=50% probability of giving all sixes 6 times in a row. That leaves us with only a 10% probability that your friend is scamming you

@@gioelechristille4650 Yeah, im aware that the nature of the significance level, with the 6 dice rolls about 16 out of 1000 games will get a false positive if you use a strict yes-no test. However it still is somewhat significant, more so for 10 dice rolls, so to say "its almost certainly random chance" is, just like blindly (and genuinely) accusing your play partner, not a reasonable thing to do. That doesnt mean you cant get suspicious and start taking a closer look at the dice-rolls in the future (should you genuinely suspect your friend is cheating at this casual game with no stakes, its another story if something is actually on the line), or at least reasonably complain in good spirits that youre being unlucky (everyone gets unlucky sometimes as the significance level shows). Like I said I only really calculated the probabilities because I was curious, and I fully agree that 6 or even 10 games arent enough to accuse your friend. It might be enough for you to get suspicious about your friend, if you deem their character or the circumstances to be compatible with cheating, and make you pay close(r) attention in the future, but thats about it.

Yes, that is to do with a fundamental problem with the thresholds for statistical significance. By definition applying a 5% threshold will get you the 'wrong' answer 5% of the time - which is why statistical significance is only a good heuristic in certain circumstances (and why publication bias is such an issue). Unfortunately statistical significance is a widely misunderstood concept even amongst people who engage in professional research, so its not surprising it gets misunderstood in wider contexts.