Statistical Analysis

[Anonymous No.12476730]

Hello fellow zoomers. You have probably heard of very famous eceleb Dream. Dream is an autistic Minecraft player who does speedruns. In one of his recent speedruns he got some insane drops that are seemingly not possible, well technically they are as possible as winning two lotteries in a row whilst only purchasing two lottery tickets in your entire life. Here's the video that goes into more advanced math only stat PhD students are familiar with. So if there are still some boomers on this board who are disattached from current trends, you should check it out.
https://www.youtube.com/watch?v=-MYw9LcLCb4

Any statisticians who can point out if there was something wrong with the math in the video? Dream claims some stat PhD he hired says their calculations are wrong, or maybe not the calculations or the distribution that was used, or the entire approach, or the interpretation.. whatever. How would you approach this problem? Would it be any different from the autistic speedrun mods?

[Anonymous No.12476762]

>>12476730
all of the math in the paper is undergrad-level

[Anonymous No.12476798]

Brainlet here, and to preface this I don't give a shit about Dream so I'm not defending him I just want to know how this works. Can someone who actually knows about statistics explain how his chances were significant when they're technically just as likely as any other drop? So his drop chances were 1 in several trillion or whatever(watched it a few days ago so don't remember exact numbers), but also getting any other combination of those drops is also 1 in several trillion right? It's like how getting 30 heads in a row is astronomically unlikely, but doing it 30 times and getting HTHHTTTTHTH etc. is just as likely as getting 30 heads in a row and the only reason we care about the all heads result is because we're giving it meaning as humans. So couldn't it be the case that he just got drops that we arbitrarily deem significant?

[Anonymous No.12476801]

>>12476798
>>12476730
literally binomial distribution super easy stats

[Anonymous No.12476816]

>>12476730
>more advanced math only stat PhD students are familiar with.
no it's not

[Anonymous No.12476825]

>>12476801
The question is, is it legitimate to use binomial distribution here? As in are all the requirements for binomial distribution met? What about the fixed number of observations?

[Anonymous No.12476850]

>>12476798
Disregard this I'm retarded. Getting the drop is far less likely than not getting it it's not like a coin flip. This is what I get for not playing minecraft since alpha

[Anonymous No.12476853]

>>12476825
if you watch the video he corrects for some of the biases and talks about the objections dream brings up.

[Anonymous No.12476864]

>>12476853
But is the 262 really fixed number of trials? I mean they are investigating how likely is it for him to get this amount of drops in 262 trials. That would be fixed number of trials. But aren't we interested in how many trials will it take him to get this amount of drops?

[Anonymous No.12476887]

OHHHHHHH DREAM

[Anonymous No.12476898]

>>12476864
They are just testing the probability of what happened in the last 262 barters he had because he was extremely "lucky" in those streams where they happened.

[Anonymous No.12476906]

>>12476898
Well yeah, but they are testing it via binomial distribution and if the requirements are not met, you can get some flawed results from it. There is for example negative binomial distribution which doesn't require a fixed number of trials, would it change the odds to the point where it gets to the realm of being possible?

[Anonymous No.12476913]

>>12476887

[Anonymous No.12476960]

>>12476906
Binomial distribution models the probability of getting k heads out of n coin flips, given the probability of a head is p. You can use it to compute the probability of getting 42 heads from 262 coin flips given the probability of a coin flip is 0.0473. However, getting k heads out of n coin flips tends to be tiny as n gets to be large, thus you usually measure the probability of getting at least k heads which is P(>=k heads out of n tries: p) = P(k heads out of n tries : p) + P((k+1) heads out of n tries: p) + ... + P(n heads out of n tries : p). An argument against the binomial distribution is that the estimation for p is wrong due to real-world factors (ex. stopping once you get a head instead of arbitrarily). The paper goes into how they correct for this. However, even if you compute the binomial distribution with the expected probability of drop from illumina's runs (p=20/323) you still dream’s streams (getting 42 heads in 262 coin flips) still have dismal probabilities.

[Anonymous No.12477003]

>>12476730
The problem is that they dont account for data skewing from lucky runs and they also dont address the idea that when you get the necessary amount of drops you stop collecting. I thought an exponential distribution would have been more appropriate. Regardless the paper reads like shit and is likely written by a high school student.

[Anonymous No.12477010]

>>12476960
I know what binomial distribution is. But when you set up an experiment, let's say you want to know how many heads are you (likely) going to get in 262 trials, then it's fixed number of trials. Then you check for other requirements and thus can use binomial distribution. But when you are investigating the probability of getting k heads in 262 trials, it's not a fixed number of trials and you can't use binomial distribution here, can you? It all comes down to what I mentioned earlier, there is a difference between investigating how many heads you are likely to get in 262 trials and how many trials will it likely take for you to get k heads.

[Anonymous No.12477048]

I just saw a video of a guy saying he did 160 unsuccessfull trades in a role, and that is way more rare.

[Anonymous No.12477075]

this is a great thread, so have a bump

[Anonymous No.12477088]

does anyone have an idea how they deal with bias due to the fact that he'd stream the good runs and not bad runs (as opposed to bias due to stopping on a good trade, which they do discuss in the vid)? is it even possible to correct for that?

[Anonymous No.12477111]

>>12477088
How can you only stream good runs if they're live?

[Anonymous No.12477117]

>>12477111
don't they also include just vids he's uploaded?

[Anonymous No.12477183]

>>12477075
This is a shit thread simping for some eceleb. Meanwhile the fucking cock sucking jannies take down the thread discussing the science of race.

[Anonymous No.12477197]

>>12476730
WHO THE FUCK GIVES A SHIT ABOUT FUCKING MINECRAFT?
PLEASE END YOUR LIFE

[Anonymous No.12477206]

>>12477010
I don't understand what you mean here. When they took the sample, they sampled 42 heads out of 262 coin flips. If you wanted to compute "how likely it is to take 262 trials given you got have 42 heads", I assume it would be \frac{p(42 heads in 262 trials)}{\sum_{i=42}^{\inf} p(42 heads in i trials)}. I suppose it would be interesting to compute this as well. Personally, I feel the binomial distribution already gives a good estimation of the likelihood... unlikely events happen all the time; the question then is "how unlikely is too unlikely?" and how reliable a sample size of 262 is...

[Anonymous No.12477295]

>>12476730
No, it was pretty cut and dry. Not very complicated stuff.
t. guy who's most complicated math class was high school algebra
>>12476887
>today i try to beat minecraft while two of my friends try to fuck me in the ass

[Anonymous No.12477319]

>>12476850
>>12476798
You're also a brainlet besides the p≠0.5 because the binomial distribution's probability mass function includes, obviously, the binomial coefficient to account for the number of different permutations by which you can acquire a given number of successes. The binomial distribution doesn't look at the probability of a certain sequence, but of a specific number of successful trials.

Now, you have the inkling of intuition for a problem that actually is real: The probability of getting exactly k successes isn't what's important, but the probability of getting k successes or better. However, the youtube video mentions this, so it's not applicable. They seem to have correctly used the cumulative distribution function.

[Anonymous No.12477327]

>>12477295
Well Drean claims to have hired a PhD in stat and apparently he is somehow gonna prove his innocence. Why can't /sci/ answer how is this possible, is this PhD a fraud?

[Anonymous No.12477334]

>>12477183
I would much rather read a thread about some speedrunner getting impossible odds than about your "science of race". Fuck off retard

[Anonymous No.12477341]

>>12477327
>is this PhD a fraud?
If he says Dream didn't cheat then yes. But considering he's getting paid and Dream doesn't seem too keen on owning up to it I think we can expect him to act like a lawyer more than a statistician.

[Anonymous No.12477365]

>>12477341
But isn't it hurting his entire reputation by doing this? Imagine sticking up to some Minecraft fag and then getting shit on from the actual mathematicians who are gonna point out how retarded you actually are. Bet it's gonna be some pajeet PhD anyways.

[Anonymous No.12477378]

>>12477365
Yes, he would be better off just admitting it and acknowledging that he did something stupid but rarely does that happen with anyone.

[Anonymous No.12477410]

>>12477319
This doesn't make any sense. You've just defined successes as a sequence like "drop drop drop drop drop" but the ideal sequence of "successes" could easily be "drop, not drop, drop, not drop, not drop" in which case what Dream did wasn't special at all. You can't have probability of .5 or it's arbitrary.

[Anonymous No.12477464]

insanely unlikely events do happen. a woman got beaned by a meteor that fell through her ceiling-the odds of that happeningto her are probably just as bad as against this guy. but it's still really fishy.

[Anonymous No.12477521]

>>12477464
That's true, but it's not one single instance of absurd luck, it's many many instances of better than average luck over something like twenty plus hours. One of the other things that makes most people believe he cheated is that he has cursory knowledge of making simple mods for the game and when asked to show his mod folder he deleted it and then said he deletes and recreates it daily.

[Anonymous No.12477525]

>>12476730
that's it im living /sci

[Anonymous No.12477543]

>>12477525
>guys this isn't the white race superiority threads I came for
>why are you discussing actual math on this board?

[Anonymous No.12477552]

>>12477521
that's sketch. he could have volunteered to have his internet history and computer looked at by digital forensics people

[Anonymous No.12477562]

>>12477552
Now that would be actual proof.

[Anonymous No.12477570]

>>12477334
Your the retard because you're sympathetic to brown people. Enjoy Brazil 2.0 in US in 15 or so years.

[Anonymous No.12477715]

>>12477552
He actually submitted his log files and loaded mods after he got his best run and they had nothing wrong with them. Its still possible to cheat but it would have been much harder.

[Anonymous No.12477835]

Does /sci/ really think this is legit proof? The paper's absolute trash, no references, how are people accepting their conclusions lmao? Dream is clearly innocent but I'm too lazy to prove it.

[Anonymous No.12477914]

>>12477835
You don't need references for what a binomial distribution is. The paper's essentially a back of napkin calculation - it's not a definitive proof that someone cheated, but some rough estimates plugging numbers into year one undergrad stats equations.

[Anonymous No.12477937]

>>12477552
>letting the whole world know you watch pornography with llamas and peruvian dwarves because of minecraft

[Anonymous No.12477966]

>>12477937
>literally the luckiest human being to ever grace the earth
>cares more about a shitty non-world record speed run time
>cares about his llama porn
Does not compute

[Anonymous No.12477972]

>>12477914

[Anonymous No.12477977]

>>12477835

[Anonymous No.12478592]

>>12477715
without an independent audit it's all his word.

It's not impossible that he's innocent, but it's pretty telling that he conveniently deletes his mod folder super often. like some on man...

[Anonymous No.12478606]

>>12477543

[Anonymous No.12478618]

>>12476730
>what is RNG

[Anonymous No.12478970]

Does he get quads every time he posts here?

[Anonymous No.12479000]

>>12476798
>>12477410
The point is to compare p(sequence | chance) against p(sequence | cheating to do better), you bloody retard. The sequence is much more likely given cheating. Other arbitrary sequences aren't because there's no fucking point in trying to cheat to get them.

[Anonymous No.12479011]

>>12477966
Listen here, you fuck, you don't tell a computer scientist what computes or not.

[Anonymous No.12479015]

>>12478618
The RNG gods have smiled upon him.

[Anonymous No.12479044]

This was every trade dream made in 6 consecutive streams. The enderpearl is the needed trade.

[Anonymous No.12479047]

>>12479044
And broke down by stream?

[Anonymous No.12479059]

>>12479047
Here is the xlsx
anonfiles com
slash ndRfje07p1
slash Dream_Stream_Barter_xlsx

not spamn niggerasdfasfasdfasdf

[Anonymous No.12479060]

>>12476798
K heads

[Anonymous No.12479081]

>>12476730
Dream, you cheated. Fuck off already

[Anonymous No.12479503]

>>12479044
what happens if you take the difference between the number of ender pearl trades that he got and what was expected, and sum it onto the other trades he got? Does the total resemble chance more closely? I'm asking because it looks like he underperformed relative to chance in other categories.

[Anonymous No.12479554]

>>124795030
good catch, if you add up the bars on the right, you wont get 262 trades. i think it’s because of the stopping rule where they couldn’t distinguish all the drops on the screen since he didn’t pick them up but just restarted the game if he didn’t see ender pearl there. thus some drops remained uncounted, to be fair they should have made a separate bar for “remaining drops” or something, because this is really shitty comparison. they might talk about it in a paper, i haven’t read it so idk

[Anonymous No.12479557]

>>12479554
meant for
>>12479503

[Anonymous No.12479607]

>>12479554
Okay, so in the best case scenario to accurately compute the true deviation from chance we'd have to subtract uncounted drops from the ender pearls, right?

[Anonymous No.12479702]

>>12479607
technically yea, but how would that help?

[Anonymous No.12479721]

>>12477327
PhD in his head? He may say he consulted the sheepels idol, the Elon Musk himself and i can say i consulted his mother last night.

Also, if the lucky run COULD've happend it doesn't mean that HIS run was valid.

[Anonymous No.12479822]

>>12479702
It would account for bias due to optional stopping. But I could be wrong (it wouldn't be the first time)

[Anonymous No.12479841]

>>12479044
>n=6

[Anonymous No.12479935]

>>12479841
that's not the N you tard

[Anonymous No.12479959]

>>12476730
I am not watching 15 minutes of spooky music and transitions just to present the binomial distribution
I fucking hate these "scene youtubers" who larp and overcomplicate stuff

[Anonymous No.12480442]

>>12479959
So set it to 2x speed and skip forward at certain parts so it only takes like 6 minutes.

Brainlet. If you don't watch all your youtube videos at at least 1.5 speed you are objectively wasting your time and ruining you life.

[Anonymous No.12480723]

>Hey shitass, wanna see me cheat my speedruns?

[Anonymous No.12480762]

>>12480442
I already watch videos at 2x speed
this was barely worth skipping through 10% at a time

[Anonymous No.12482080]

>>12480762
>b-but my time is so valuable
>on a Croatian embroidery forum

[Anonymous No.12482207]

Believing video game player pros dont use an edge. Yeah normie they play games just like you.

[Anonymous No.12482218]

>>12477117
No he streamed 6 speedruns live, and in those 6 runs he had stupid luck.
He could've MAYBE proved his innocence if he said that the streams were prerecorded, but he didn't claim that at all.

[Anonymous No.12482243]

>>12477835
>Does /sci/ really think this is legit proof?
I'm only a retard with a B.S. in stats, but imo statistics don't typically "prove" things, they only give statistical evidence for a claim.
The study the mods did don't "prove" that Dream cheated but it pretty much proves that there is an extremely significant difference in Dream's trade luck compared to everyone else's. Dream needed to somehow explain why his luck was so different. Other than him cheating, the only thing I can think of is that he prerecorded 6 good runs out of like 100, making the study the mods did biased. But it would probably be easy to prove that the runs were actually live

[Anonymous No.12482322]

>>12480723
Kek

[Anonymous No.12482455]

>>12477835

Supreme bait, here's a (You).

[Anonymous No.12482763]

>>12476730
Statistics doesn't define reality.

Anybody who believes that should quit this universe and go somewhere else.

[Anonymous No.12482769]

>>12476730
Collapse.

[Anonymous No.12482808]

>>12476730

Technically speaking, the mods didn’t really need to apply stats techniques perfectly. As long as their methods downplay Dream’s luck, but still come out with bonkers odds, then they did find pretty strong evidence of him cheating.

[Anonymous No.12482824]

>>12482243

In Physics, experiments can’t prove things, they only offer statistical evidence.

At some point, you kind of have to decide on a probability cutoff that will cause you to reject your null hypothesis.

[Anonymous No.12482834]

>>12476825

Yes, because advancing the game state only relies on looting a minimum number of Ender Pearls and looting a minimum number of Blaze Rods (each pair of Ender Pearls and Blaze Rods combine to create Eyes of Ender, which are the keys to the boss).

[Anonymous No.12482927]

>>12476825

Actually, no, it would be correct to apply a negative binomial distribution to each of his runs because he stops after getting a sufficient number of Ender Pearls, and Blaze Rods.

The only problem is that the math would be a bit finicky because we would need to use Arithmetico-Geometric series, and some monomial-to-finite-series expansions.

[Anonymous No.12482942]

>>12477197
its math being used in pop culture
whats the issue?

[Anonymous No.12483040]

>>12477570
Niggers we don't give a shit about your /pol/ spam thread number 8337

[Anonymous No.12483041]

This isn't looking good Dreamers

I changed the perspective on the analysis by treating the number of trials as the random variable, and the number of Ender Pearls as a constant, but its damning.

Trading 42 Ender Pearls, while only getting at MOST 220 failures only has a 1 in 500 TRILLION chance at happening.

And that is a HUGE overestimation, because Dream's runs have the Ender Pearls more evenly distributed among the individual runs.

[Anonymous No.12483044]

>>12483041

*A huge underestimation of the odds because Dream's Ender Pearl distribution was more even.

[Anonymous No.12483059]

>>12483041
OH NO NO NO DREAMERS WE WERE TO COCKY

[Anonymous No.12483242]

>>12482080
Yes.
>yet you live in society
>curious

[Anonymous No.12483255]

>>12476730
fuck all this boring math shit all you have to do is watch other speedrunners and see that they always get less than half of dream in 263 trades even if you started counting from a ender pearl trade.

[Anonymous No.12483515]

>>12478606
>Why yes, I check catalog every once and in a while and search ctrl+f "IQ". How could you tell?

[Anonymous No.12483529]

*puke*

[Anonymous No.12483855]

>>12483529
he makes like millions and yet only donates couple thousands lmao? I reckon this is just another high IQ move by Dream to gain some retarded leftist followers so he can make more money off them

[Anonymous No.12483987]

>>12483529
Totally unrelated who cares
>>12476730
I don't think anyone here understands the game well enough to say whether the autistic speedrun mods are correct in the way that they have approached this. Anecdotally I've been accused of cheating by speedrun admins who were more knowledgeable but completely wrong so I guess we'll have to wait and see what this PhD guy has to say.

[Anonymous No.12485114]

>>12483041
So the correct way to calculate the odds would be to take each attempt, calculate the probability of that attempt by using negative binomial distribution and then take the product of calculated probabilities which would yield the actual result?
So that way you also include the stopping rule bias into the calculation correct?

[Anonymous No.12485746]

>>12483040
NOOOOOOO YOU NEED TO POST RACEBAIT IQ THREADS ONLY!!!! HELP! HELP!!! IM BEING GENOCIDED!!!!!

[Anonymous No.12485876]

>>12476850
No no no don't be mistaken, it wont do you any good, even for coin flips you're still full of shit.

[Anonymous No.12485884]

>>12483987
>Totally unrelated
lol

[Anonymous No.12486066]

>>12476730
>Dream claims some stat PhD he hired says their calculations are wrong
>some stat PhD he hired
>hired
wow no fucking shit mate, you think he'd hire someone who tells him that the calculations are right? lmao
this is no different to paid oil, tobacco, etc. shills

[Anonymous No.12486301]

>>12479000
Trips of truth

[Anonymous No.12486424]

>>12485114
>>12482927
>>12485114
I find it really odd that the paper they published said they accounted for "stopping bias" by dividing by 90 or some shit
When you can just use the negative binomial formula and calculate the probability of a success on the nth trial
I literally never heard of minecraft speedruns before this happened but this seems really strange

[Anonymous No.12487480]

>>12486424
I haven't read the paper, can't be bothered really, but people keep claiming that they explain it there. I am talking about the first bias - the stopping rule. The 90 you are talking about is a different thing, it was the fact that they took this exact pair of drops (pearls and blaze rods) while there might be a bunch of other different drops he could have been unlucky/lucky with (e.g. flint drop and iron drop from golem), they are basically saying that the fact that they took that exact pair is biased. I fail to see how though cause the probabilities are given by the game code and if he clearly altered either of them then how would that be connected to his other drops (obviously except for other piglin drops which would be logically lowered by the weight he added to the pearl drop). Plus pearls and blaze rods are literally the only drops that are very RNG dependent in these speedruns. Anyone feel free to correct me if I'm wrong.
Probably the thing that represents Dream cheating the most are the graphs of # drops (=pearls/blaze rods) x # attempts (= gold traded/blazes killed). If they investigated other runners and their drops are within the plausible range and Dream's drops are far above the 99.9th percentile. then he's obviously cheating. Perhaps they should have added more runners to the graphs, or just created some average of several runners, because there is one other runner, Illumina, who they claim to be the luckiest they could find.

[Anonymous No.12487806]

>>12476730
I am unconvinced. Stats can't be used to prove an election is rigged, why pretend it counts for minecraft speed runs?
Also, why are they only sampling such a few number of drops. They should go through every available run. They only did enough analysis to convince themselves that cheating took place and stopped there.
Next is the problem of the nature of this. Trials by public opinion are braindead clickbait fodder. There is essentially a debate where Random n responds to Dream statement x. No expert gives a fuck what somebody responds with. Analysis is done on things as they stand with no weight given towards what someone may say in response.
I don't mind this video, they just have too much certainty for only analyzing a small amount of the data available. All together, the only take away I have is that they have shown that this event can happen given minecraft environment as they have shown it. Doesn't seem like cheating to me.

[Anonymous No.12487880]

>>12487806
the issue is that the runs they investigated were from 6 consecutive streams, before that Dream had apparently taken a long break from speedrunning and after those 6 streams he stopped doing speedruns since he set his PB
by your logic if they decided to sample all his runs from that version he would get better odds since he likely didn't cheat before the break
the video is certainly very biased but in Dream's favour, they literally "make up for bias" by giving him extra odds
there have been many people simulating his results and not a single simulation ever got close to his luck
i also suspect you are Dream himself or one of his retard shills, i seriously recommend you to do some better research

[Anonymous No.12487883]

>>12487806

Taking all available runs is a pretty dumb way to analyze the data, because its easy to add a bunch of fake runs with deliberately bad rng.

The right way to analyze the data would be to define a bernoulli trial that is 0 when Dream gets unremarkable luck, and is 1 when he gets insane luck, and to take a binomial distribution of said trial. However, there should also be a standard of removing runs that are quit due to gameplay mistakes, or are otherwise quit early. The quantity of unremarkable runs should scale with the amount of time he speedruns the game.

[Anonymous No.12487932]

>>12487880

If the runs were consecutive, then the mods’ decision to only take those six runs is perfectly fine. Picking a singular cutoff date for the data pool isn’t really cherry-picking data.

[Anonymous No.12488082]

>>12486424
the dividing by 90 was to account for unintentional p-hacking, it was not the only adjustment for bias they did

[Anonymous No.12488861]

>>12479000
Bayesed and redpilled

[Anonymous No.12489386]

>>12487932
Afaik it's not just 6 runs, but 6 streams which likely include a much higher number of runs

[Anonymous No.12490299]

>>12487880
This is you just in agreement with a shitty unconvincing youtube statistician. I am guessing you are not familiar with the subject at all. They presented a weak case and then you tell me to go do research to prove their case. That isn't going to happen.
You have fallen in with a mob of twitter retards over minecraft speed runs. You sell out cheap.

[Anonymous No.12490363]

>>12490299
>I like that trade!

[Anonymous No.12491095]

>>12490299
>t. Dream

[Anonymous No.12491839]

>>12483855
He smart . The proplems is he is not that smart

[Anonymous No.12492544]

>>12485746
>shitlibs think MAGApedes care about race and IQ
lel

[Anonymous No.12492584]

>>12476730
The math is probably right, but if it is the only thing they have against him then it really means jack shit. This happens multiple times and people fail to see just how weak statistical evidence can be. Rare events occur, that is just what is to be expected, finding a rare event is proof of nothing, bring actual evidence he cheated, a proposed mechanism or something. That is why science based on statistics is bullshit.

[Anonymous No.12492758]

>>12492584
but...1 in 7.5 trillion, that cant happen, its big number, lottery is 1 in trilion, he cheated

[Anonymous No.12494393]

>>12492758
And? people win the lottery all the time.

[Anonymous No.12494975]

>>12494393
The probability of winning lottery increases with the number of lottery tickets you buy though. Dream’s odds of getting that many drops only decrease as he does more runs.

[Anonymous No.12495005]

>>12494393

Think about it like this:

Dream won the lottery of getting randomly banned.

[Anonymous No.12495066]

>>12476730
Ok so I read their explanatiln and really yu caan simplify the whole thing. It is completely sufficient to present the following to make a damning argument: Killing a blaze has a 50% chance of dropping the rod. Over the course lf killing ~300 blazes he got about ~200 rods. Then with the binomial distribution you can quote easily check that getting 200 heads in 300 coin flips is neigh impossible so he's in slme real trouble.
Now of course on reality you can't really just plug in the binomial distribution because theoretically dream could have played with the intention of stopping as soon as he gets to a 2/3 success ratio over the last 300 blazes killed and just play for billions of games until this happens, but since theres an easy upper bound on how many blazes he could have killed on streams in speedruns (100000 or whatever) you can give him the benefit of the doubt and assume he tried to get 200 rods over the last 300 blazes by killing up to 100000 blazes. but even then his chances are fucked so yea it looks pretty bad.
And in addition he had even better luck om the pearls.
I really enjoy dreams videos but as of now I believe he cheated and you really can't make an argument with complicated statistics here. He got 200 heads in 300 coin flips on stream.

[Anonymous No.12495070]

>>12495066
and it is also important that this is not a single, crazy unlikely event, but a long sequence of consistently higher than expected probability outcomes

[Anonymous No.12495106]

>>12479044
damning.

[Anonymous No.12495107]

>>12495106
Something tells me you've never gotten lucky.

[Anonymous No.12495108]

>>12495107
Actual retard.

[Anonymous No.12495157]

>>12479841
no you idiot n is the sum all bars, the total number of trades. It is >150 Id estimate, at least >100.

[Anonymous No.12495159]

>>12479959
read the pdf faggot

[Anonymous No.12495588]

>>12495066
That's why they should have used negative binomial distribution instead of binomial. Not like it would boost Dream's odds in any way, he still cheated and I remember he promised a response like 2 weeks ago, what is that faggot doing?

[Anonymous No.12495613]

>>12492584
Rare events cant happen, but it's not one instance of extreme luck, it's several. The statistical evidence shows there's a significant difference in Dream's trade luck. That doesn't prove he cheated, but he needed to provide a decent explanation for that luck which he didn't and I can't think of any other reason his luck would be so good other than cheating.
You can't say that he for sure cheated based on the statistical evidence alone, but writing off the statistical evidence as "dumb luck" is even more retarded

>Doctor 9/10 of the patients who took this treatment died
>Well those 9 patients were just very unlucky! We don't need to do any more research

[Anonymous No.12495647]

>>12495613
No he doesnt need to provide jack shit that's up to the accusers. Obviously this is minecraft retardedness so It doesn't matter but if the rules of the community aren't tight enough to prevent shit like this well that's a problem with the system. Im not saying it is luck or that it's anything but banning him only because of statistics is retarded too as it is not direct evidence.

And yes, proper medical sciences doesnt consist of statistical evidence, well it became that way and now we have a reproducibility crisis. But to actuañly3 be scientific you need both a testable MECHANISM and statistical evidence. You can probably find 1 or 2 proper homeopathic studies that show it can work on statistics alone. Same with parapsychology.

[Anonymous No.12495721]

>>12495647

>minecraft speedruns are stupid and shouldn’t be taken seriously

>banning a cheating speedrunner is stupid because the mods need to have more serious standards

cope

[Anonymous No.12495729]

>>12495721
I'm saying what I think they ideally should do but also I don't give two flying fucks about retarded minecraft autists. It's not like I'm spaming their twitter or anything you retard. Fuck off
>>>/v/irign

[Anonymous No.12495779]

>>12495729
I think statistical evidence is more than sufficient in this case. Consider those groups of people who gather around outside of malls and ask passerbys to take bets on a coin toss (or cup shuffling or whatever). If you watch them play 30 times and they somehow got 24 wins in a row on a H/T coin toss, it would seem quite fair to assume they’re doing something shady, and you would only be wise to refuse to play.

But then there’s people like (You) who immediately appear from the nearby bushes and start screeching "YOU DONT HAVE ANY EVIDENCE THEYRE CHEATING! YOU CANT PROVE IT YOU DONT HAVE ANY TESTABLE HYPOTHESES OF A MECHANISM THEY WOULD USE TO CHEAT"

and maybe you’re right, we can’t explain how they are cheating, hey maybe they’re just the luckiest group of humans ever, but I would rather assume they're just cheaters.

[Anonymous No.12495806]

>>12495779
>but I would rather assume they're just cheaters
A ok, the point of a fair justice system is that we don't decide based on what our "gut" tells us. I personally feel that he cheated, but well that should be irrelevant to a proper investigation. Compare this with autists that show a speedrun is spliced, the latter has clear direct evidence and always reproduce what the cheater does, this only shows something odd. I agree that if you have something this odd it is worth investigating.
>but he probably already deleted bla bla and changed whatever
I'm not saying take his computer to a forensics, but if you simply gave what particular mod or tweak he could have used, and reproduce the results then I would consider the ban fair. Again, it is the same as just saying someone cheated if he won the lottery because it is unlikely. You haven't specified certain important priors, That's why bayesian statistics should be thought to everyone by force.
https://en.wikipedia.org/wiki/Prosecutor%27s_fallacy
But again I'm here because I'm board I seriously don't give two shits about the minecrafta virgin asppie comunity

[Anonymous No.12496175]

>>12495806
The justice system does not "PROVE" anything. It too deals in matters of probability and uncertainty. Odds this out of wack would be enough to convict anybody of murder.

>But your honor, you haven't disproven the 1 in a trillion chance my client might not have murdered that old woman! What if his DNA was at the crime scene because.... a bird put it there! You haven't disproven that astronomically unlikely scenario!

Not how it works buster.

[Anonymous No.12496273]

>>12476762
not only undergrad level, but I learned all this in my sophomore year as an industrial engineering student

[Anonymous No.12496280]

>>12476730
>Hello fellow zoomers
TADR you're a massive faggot

[Anonymous No.12496290]

>>12476850
MMAHMAW LOOK AST ME IM OLD PALYER GOD IM AS SO ODOOODLLLDD I PAYED ASLPHA ALPHADA DA MINCEFAT NOTCH LETS MET SUCK YOUR FAT COCK MAMEMEMS AIM SO OAOLALD

[Anonymous No.12496296]

>>12477327
>misspelled PhD
discredited already

[Anonymous No.12496319]

>>12496273
The math in the paper is taught in american highschool stats classes.

[Anonymous No.12496321]

>>12477327
>also my dad works at nintendo and he says I didn't cheat!

[Anonymous No.12496585]

>>12496290
Lmao calm down, retard. Most zoom zoomers probably started playing Minecraft well after it’s official release, or maybe in late beta.

[Anonymous No.12496606]

>>12477327
It's probably gonna be some Applied stats nigger who learned mathematical modeling last semester and will purposefully try to confuse everyone by using some wacky models.

[Anonymous No.12496628]

>>12477977
gosh i wish that were me

[Anonymous No.12496780]

>>12476730
I agree that the probability of exactly 42 events each having their own probability of 0.0218 and total trials 263 is effectively zero. An outcome being improbable does not make it impossible to happen. But certainly is enough to say that outcome is unlikely to happen by chance. The degree of unlikeliness is sufficient to call the probability zero, few would object.
>only math PhDs
It's first year stats but ok

[Anonymous No.12496791]

>>12477410
What are you even talking about? The point is that getting HHHHHT has the same probability as HTHTHT but getting any permutation of five heads and one tail is significantly less likely than three heads and three tails, and the binomial distribution is describes the probability of getting a specific number of successes (as in, individual trials, in this case coin flips), not of getting any particular sequence.

[Anonymous No.12496836]

>>12496175
>What is beyond a reasonable doubt

[Anonymous No.12496845]

>>12496836
>one in a trillion chance of innocence is REASONABLE doubt
No.

[Anonymous No.12496849]

>>12496845
https://en.wikipedia.org/wiki/Sally_Clark
You and the fat autistic mods don't understand shit.

[Anonymous No.12496850]

>>12496836
Is this being defended in a court of law or for an online speedrun website?

[Anonymous No.12496854]

>>12496850
Again, I'm just saying why it could be unfair, not that I think that this should be taken to the supreme court lmao. Just that retarded mods don't understand statistics.

[Anonymous No.12496874]

>>12496854
if someone turned in a model with results reproducible 1/trillion times I would throw the model out
if it was deciding whether or not they should die then maybe I would hesitate about doing so but that's not that scenario

[Anonymous No.12496895]

>>12496874
Again, the problem is that there are no proper priors and so that probability alone cannot be interpreted as the probability of him cheating. Obviously trillion is really fucking high and that would mean cheating in speedrun must be really rare or difficult for the probability of him beig guilty to be low. This could be solved if they just tried to investigate how you could potentially cheat, if it's easy and if it could not be detected by whatever anti-cheat shit the site has. Maybe there already is but that should also be key. If no one has a fucking clue how he cheated either he is a genius or well, he didn't cheat.

[Anonymous No.12497017]

>>12496849
There was about a 50/50 chance of that happening to at least one person in England at that time.
There's a 0.00156% chance of this happening to anyone a single time out of all of Earth right now.
If every single person on Earth started doing this to find out how long it would take for it to happen even once it would take 101 years, 5 months, 10 days, and 16 hours for there to be just a 50/50 chance of this having happened at all.

[Anonymous No.12497173]

>>12496895
I think your the problem stems from your lack of knowledge regarding this particular game.

Cheating in the way he most likely did would not require some genius level hacking or difficult technical challenge, it is trivial to reproduce. Minecraft is a single player game. There is no anti-cheat, all this guy had to do was change a few numbers in notepad to increase drop rates, which is almost certainly what he did. Since he deleted the relevant files without letting anyone see it, it seems pretty clear that the obvious conclusion is that he cheated.

It’s like someone immediately flushing a bag into the toilet when the police arrive, and when they find traces of drug residue on surfaces all over, he says "oh well that could be a coincidence, I swear I didnt hide any evidence, that bag was empty".

Sure, we cant prove without a shadow of a doubt, but anyone with a bit of sense will understand that he is lying.

[Anonymous No.12497187]

>>12497173
Ok that's all I needed, if it is really easy there is no reason to believe the priors will change drastically. Deleting that specific shit is also damming evidence. Just saying that is much more crucial for me than pure statistics.

[Anonymous No.12497244]

>>12497187
Fair enough. It seems to me that statistics can be very misleading, perhaps due to the non rigorous nature of this subject. I have to say, reading things like this >>12496849
And seeing the reproducibility crisis occurring in fields that rely heavily on inferential statistics makes me very critical of the usefulness or soundness of the methods that are commonly employed.

Is stats actually reliable? Or is it largely pseudoscience?

[Anonymous No.12497265]

all odds are 50/50
either it happens or it doesn't

[Anonymous No.12497269]

>>12497265
PHYSICISTS!!!!!!!!!!!!!!!!!!!!!!!!!!

[Anonymous No.12497378]

>>12497244
Statistics in itself is not the problem I think, but how people learn it. For example, one of the most common mistakes made by people is thinking conditional probability "commutes", that means that given assumption A the probability of B is the same as given assumption B the probability of A. In this example what they calculated was, assuming he is innocent (I.e taking the typical distribution in the game code) what is the probability he got those results and that was horribly low. No this is not a probability on his innocence, but rather just showing how odd was that he got that. So what is the probability that he cheated assuming he got those results? Well this literally is the question that benefits from evidence or other arguments. And there is no clear relation between these two a priori. Just consider, what is the probability that a random dog has four legs, compared to the probability that a random four legged animal is a dog? Or the probability that an astronaut is Russian or American vs the probability that a random Russian or american is an astronaut. In science, this is a common fallacy especially when they calculate p-values, as they usually want to test a hypothesis by assuming certain distribution and computing the probability that they got that data, it only tells you that an odd event happened or your hypothesis is wrong, is up to you to explain why other hypotheses are a better fit. Well this is one type of problem but there are plenty of them and it will get worse now that people are drinking the datascience cool-aid. I think in general the rule is that if someone only has statistical evidence for something, then you should take it very skeptically.

[Anonymous No.12497423]

>>12497378
> I think in general the rule is that if someone only has statistical evidence for something, then you should take it very skeptically.
Great post and great heuristic. Thanks for the insights anon.

[Anonymous No.12497460]

>>12497378
That would be more relevant if this was a one off event, but it wasn't

[Anonymous No.12497463]

>>12497378
Wasn't it over 6 days not once

[Anonymous No.12497469]

>>12497378
>>12497463
And for 2 different probabilities of blaze rod and ender pearl drops which do not affect each other in the game's code

[Anonymous No.12497513]

>>12497460
>>12497463
>>12497469
The probability they calculated was his overall trades if I read it correctly, this means when it happened doesn't affect. I believe they cover this at the beginning. They corrected for some biases, but probability was in essence out of a pool of n events how many of them were wins, which takes out order or consistency. But again, because they calculated the probability assuming the distribution was bionomial (i.e. the distribution in an unmoded version) then the probability they found was, assume he is innocent, then what is the probability he got what he got. Now something I glossed over but it seems much better are the graph of the data were it compares to other runners, this I believe is crucial because as you can see other runners seem also to be systematically above others but he is way above. That is, how "lucky" other top runners have been.

[Anonymous No.12497536]

>>12497460
Alright, how do you know that the computer didn't bug out during the stream? Oh, you don't. Has anyone done tests on what happens when java idle or even running minecraft without interaction happens?
I saw someone that game bugs out when too many worlds were previously generated. In theory this should be unaltered data.
This doesn't stand up as evidence in the slightest. Only bandwagoners are this stupid

[Anonymous No.12497613]

>>12482763
Statistics is what glues patches of understanding together.
It the best tool we have against the unknown.

[Anonymous No.12497631]

>>12497265
actually all odds are 100%
because only the things that happen actually happen

[Anonymous No.12497650]

If numbers fail, get more numbers.

[Anonymous No.12497772]

>>12497536
If you read the paper they actually did consider things like that. For example they went over the way Minecraft uses java to generate random numbers, actually in a bit of detail about the pros and cons of linear congruential generators, and how it’s worked around in Java’s implementation using different bits or whatever. They also discussed for example if some glitch caused his system time to not update, since part of the random generator is based on sys.time() or whatever. They even went over previous versions where it was possible to game the random generator so that it would behave in a predictable way

[Anonymous No.12498719]

>>12476730
Dream responds to accusations.
https://www.youtube.com/watch?v=1iqpSrNVjYQ

We got too cocky anti-Dream bros

[Anonymous No.12498727]

>>12496849
you are retarded
Sally Clark occurred due to the events not being independent
the mods have already proven that the blaze rod and ender pearl random generations are independent events

[Anonymous No.12498739]

>>12498719
This is gonna be interesting. I hope he releases his PhD advisor’s name if he is actually real,

[Anonymous No.12498741]

>>12498719
Here is the report documenting the mistakes of the mod team.
https://drive.google.com/file/d/1yfLURFdDhMfrvI2cFMdYM8f_M_IRoAlM/view

This isn't looking good for us honestly. I really thought we got him. Fuck.

[Anonymous No.12498764]

>>12498739
dream quickly mentions how they only took 6 luckiest streams
>speedrum Minecraft
>take a break
>start speedrunning again
>only do 6 streams
>get 1 in a 7.5 trillion lucky
>say mods didn't take previous runs before break
>me thinking this was a serious response
that's how I know his PhD was some shitskin brailet

[Anonymous No.12498766]

>>12498764
yeah lmao, I've never gotten the whole "but you didn't include my previous streams" shit
yeah no fucking shit your previous streams weren't included because you weren't cheating in them and the only reason you want them included is to muddy the statistics and lower the probabilities

[Anonymous No.12498775]

>>12498741
counter-paper written by /sci/ when?

[None !DNiTE6Xixg No.12498802]

>>12498719
>>12498739
>>12498741
I think we're gonna do it, Dream bros!
We just need to find more excuses fast enough to prove that Dream is innocent!
Please like, subscribe, and share all his videos on facebook/twitter.
Reply to all haters and people asking questions in the comment section before reporting them.
Also please buy at least 3 free products from his merchandize page:
https://dreambranding.com/
I know it's asking a lot but if you could install nord VPN and rewatch all Dreams videos with a different IP, it would really mean a lot!


#IHadADream
#MiraclesCanHappen
#ItsJustABug
#NordVPN
#Blacked.com

[Anonymous No.12498820]

>>12498766
Just finished the video and checked the report. Lmao, that nigger who wrote it didn't even bother to include his name so that someone could check if he is real PhD or a fraud. It was apparently some employee of a company named Photoexcitation.
https://www.photoexcitation.com/
Their website looks like it was made by a high schooler, there's no contact on who is managing it whatsoever. Pretty much just "pay us so we can disprove what anyone about you says by using some flawed logic".
Truly bs.

[Anonymous No.12498828]

>>12498820
lul, to be fair though, I wouldn't put my real life info otherwise I would be spammed with bullshit from 9 year olds seething

[Anonymous No.12498836]

what is it with this retarded argument that the chance that any minecraft player could have gotten this lucky is relevant
isnt the chance of streaming speedrunner being this lucky what one should look for

[Anonymous No.12498844]

>>12498836
NOOOOO STOP MAKING SENSE WE ARE TRYING TO GET MAD OKAY????

[Anonymous No.12498858]

>Autist makes hundreds of thousands while math cucks discuss degrees f certainty all the whilst giving him even more exposure and money

Sage.

[Anonymous No.12498866]

>>12498836
I-I-It's the same f-figure!

[Anonymous No.12498998]

>>12498820
>Domain registered in May of this year

[Anonymous No.12499003]

>>12477003
>The problem is that they dont account for data skewing from lucky runs
Yes they do.

>dont address the idea that when you get the necessary amount of drops you stop collecting
That's irrelevant.

[Anonymous No.12499012]

>the odds aren't 1 in 7.5 trillion
>they're actually 1 in 10 million
that really showed them

[Anonymous No.12499014]

Streamer guessing 7:
https://www.twitch.tv/xqcow/clip/EntertainingVivaciousShieldSquadGoals?filter=clips&range=24hr&sort=time

[Anonymous No.12499050]

Huh, why is this here? I just googled "adults playing video games" and it seems alarmingly common. Do these men have no content in their lives? I am mildly confused.

[Anonymous No.12499114]

>>12499050
Yet you're down here hanging out with all us manchild losers. Do you know the average age of this site is mid teens to late twenties? Get a grip

[Anonymous No.12499181]

>>12476730
https://www.youtube.com/watch?v=1iqpSrNVjYQ
DUDE HE JUST RESPONDED
STATISTICS BTFO
https://drive.google.com/file/d/1yfLURFdDhMfrvI2cFMdYM8f_M_IRoAlM/view

[Anonymous No.12499290]

>>12499181

>[code comments] to test the hypothesis that the probability of ender pearls was somehow boosted, calculate the number of gold barters needed for a variety of ender pearl probabilities from the nominal 20/423 up to 100/423

>i have an approximate model for the number of pearls given (code snippet below) that matches the observed distribution and was suggested by a contributor that wishes to remain anonymous

The “anonymous contributor” is the biggest fucking troll

[Anonymous No.12499296]

I think he cheated
my reasoning:
low chance
He seems like the kind of guy who would do that
He's butthurt that he got caught

[Anonymous No.12499300]

>>12499290
I don't get it, can you explain?

[Anonymous No.12499301]

>>12499012
Given how many people play Minecraft every single day it's not that unlikely.

[Anonymous No.12499303]

>>12499296
>He seems like the kind of guy who would do that
Why? His speedrun videos weren't getting many views anyway.

[Anonymous No.12499306]

>>12499300

Just take an introductory statistics class, or just look up what “Hypothesis Testing” is.

[Anonymous No.12499343]

>>12499303
He's completely obsessed with winning all the time
It's obviously all consuming for him
Reminds me of myself

[Anonymous No.12499363]

>>12498727
That was one issue, another issue was the prosecutor fallacy.

[Anonymous No.12499367]

Reddit seems to think that dream is still BTFO
(Reddit is much smarter than 4chan so we should listen)
https://www.reddit.com/r/statistics/comments/kiqosv/d_accused_minecraft_speedrunner_who_was_caught/

[Anonymous No.12499477]

>>12499181
Can someone who understands math and statistics explain is this rebuttal good or bad?

[Anonymous No.12499494]

>>12499367

Summary: Whoever wrote that is either deliberately manipulating numbers in favor of Dream or is totally clueless despite having working experience with statistics. Familiarity with the concepts is clearly there, but they are misapplied in absurd ways.

The abstract has problems already, and it only gets worse after that.

The original report accounted for bartering to stop possibly after every single bartering event. It can't get finer than that.

Adding streams done long before to the counts is clearly manipulative, only made to raise the chances. Yes you can do that analysis in addition, but you shouldn't present it as main result if the drop chances vary that much between the series. If you follow this approach Dream could make another livestream with zero pearls and blaze rods and get the overall rate to the expected numbers. Case closed, right?

one in a billion events happen every day

Yes, because there are billions of places where one in a billion events can happen every day. It's odd to highlight this (repeatedly). All that has been taken into account already to arrive at the 1 in x trillion number.

Ender pearl barters should not be modeled with a binomial distribution because the last barter is not independent and identical to the other barters.

That is such an amateur mistake that it makes me question the overall qualification of the (anonymous) author.

Dream didn't do a single speedrun and then nothing ever again - only in that case it would be a serious concern. What came after a successful bartering in one speedrun attempt? The next speedrun attempt with more bartering. The time spent on other things in between is irrelevant. Oh, and speedrun attempts can also stop if he runs out of gold without getting enough pearls, which means negative results can end a speedrun. At most you get an effect from stopping speedruns altogether (as he did after the 6 streams). But this has been taken into account by the authors of the original report.

[Anonymous No.12499498]

>>12499494
The lowest probability will always be from all 11 events.

No it will not. Toy example: Stream 1 has 0/20 blaze drops, stream 2 has 20/20 blaze drops. Stream 2 has a very low p-value (~10-6), stream 1 has a one-sided p-value of 1, streams 1+2 has a p-value of 0.5.

Applying the Bonferroni correction and saying that there are 80 choices for the starting position of the 20 successful coin tosses in the string of 100 cases gives 80/220 = 7.629 × 10−5 or 1 in 13000. But reading over https://mathworld.wolfram.com/Run.html and performing a simple Monte Carlo simulation shows that it is not that simple. The actual odds come out to be about 1 in 6300, clearly better than the supposed ”upper limit” calculated using the methodology in the MST Report.

Learn how to use a calculator or spreadsheet. The actual odds are 1 in 25600. They are significantly lower than the upper bound because of a strong correlation (a series of 21 counts as two series of 20). The same correlation you get if you consider different sets of consecutive streams. The original authors got it right here.

For example, the probability of three consecutive 1% probability events would have a p-value (from Equation 2 below) of 1.1 × 10−4. The Bonferroni corrected probability is 8.8 × 10−4, but a Monte Carlo simulation gives 70 × 10−4.

From the factor 8 I assume the author means 10 attempts here (it's unstated), although I don't know where the initial p-value is coming from. But then the probability is only 8*10-6, and the author pulls yet another nonsense number out of their hat. Even with 100 attempts the chance is still just 1*10-4. The Bonferroni correction gets better for small probability events as the chance of longer series goes down dramatically.

[Anonymous No.12499502]

>>12499498
Yet another edit: I think I largely understand what the author did wrong in the last paragraph. They first calculated the probability of three 1% events in series within 10 events. That has a Bonferroni factor of 8. Then they changed it to two sequential successes, which leads to 10−4 initial p-value (no idea where the factor 1.1 comes from) - but forgot to update the Bonferroni factor to 9. These two errors largely cancel each other, so 8.8 × 10−4 is a good approximation for the chance to get two sequential 1% successes in 10 attempts. For the Monte Carlo simulation, however, they ran series of 100 attempts. That gives a probability of 97.6*10-4 which is indeed much larger. But it's for 10 times the length! You would need to update the Bonferroni correction to 99 and then you get 99*10-4 which is again an upper bound as expected. So we have a couple of sloppy editing mistakes accumulated to come to a wrong conclusion and the author didn't bother to check this for plausibility. All my numbers come from a Markov chain analysis which is much simpler (spreadsheet) and much more robust than Monte Carlo methods, so all digits I gave are significant digits.

[Anonymous No.12499506]

>>12499494
>>12499498
>>12499502
well /sci/ did reddit outsmart you? or is reddit truly the place where retards act smart

[Anonymous No.12499517]

>>12499494
>>12499498
>>12499502
why did u copy-paste here the most upvoted comment from the thread?

[Anonymous No.12499523]

>>12499517
he wanted someone to gift him 4chan gold
what a sneaky sneak

[Anonymous No.12499531]

>>12499517
To see how /sci/ would respond to it.

[Anonymous No.12499536]

>>12477327
>>12496606
This prediction was 100% correct. It was somebody in "astrostatistics" whatever the fuck that is

[Anonymous No.12499664]

>400k likes
If he is a cheater he's a clever one

[Anonymous No.12499673]

I dunno guys this Dream guy seems very convincing...

[Anonymous No.12499789]

>>12499664
This is as far as he needed to take it for his fans to believe him.
At this point he can't respond to the new criticism from the guy on reddit since it would be the same as admitting to cheating, so he's probably gonna sweep this under the rug and it won't make a difference either way

[Anonymous No.12499816]

>>12498820
The real issue is the incredibly mental gymnastics and abuse of statistics going on in that paper. He mixes rhetoric and rigor without any reservations. "Well I figure we can just choose to drop the last data-point and now the odds look less extreme", also his choice to use a separate stopping algorithm which in my opinion requires a convoluted and unwieldy way of considering things, and he himself even admired mostly binomial distribution got more or less similar results.

So he chooses to use the "barter stopping" process whenever it looks better for Dream, but when it didn’t make a huge diffference he decided not to use it. What kind of inconsistent bullshit is this? This guy is a charlatan. No surprise he wouldn’t put his real name on it.

[Anonymous No.12499833]

>>12499306
>note that the methodology is not in strict keeping with the premise of hypothesis testing
lmao

[Anonymous No.12499837]

The Speedrunning community and Mods win if this is dropped: Dream is giving them money to develop better anti-cheat measures.

Dream wins if this is dropped: His reputation is untarnished and even improves because everyone sees he's a philanthropist helping fight cheaters.

It's a win-win situation. They have a good thing going on. Stop investigating.

[Anonymous No.12499893]

>>12499816
>>12499833
Did you see this?
>These probabilities are
also similar to the probability estimated by the MST Report, with the most direct comparison to their naive
estimate of 5.65 ×10−12. As expected, using the Barter Stopping criterion increases the probability, though
some of the difference may be attributable to the Bayesian modeling method as well.
However, this probability does not account for the fact that these streams were chosen for investigation
specifically because they seemed low probability. That is, 3 × 10−10 is not the probability that Dream
modified the ender pearl probabilities.
LMAO
who tf is this brainlet

[Anonymous No.12499904]

>>12499893
>Applying the Bonferroni correction and saying that there are 80 choices for
the starting position of the 20 successful coin tosses in the string of 100 cases gives 80 220 = 7.629 × 10−5 or 1 in 13000. But reading over https://mathworld.wolfram.com/Run.html and performing a simple MonteCarlo simulation shows that it is not that simple. The actual odds come out to be about 1 in 6300, clearly better than the supposed ”upper limit” calculated using the methodology in the MST Report.
Isn't this wrong

[Anonymous No.12499947]

>>12498836
Yes. And they did that. Dream still got massively better luck than other speedrunners, like Illumina, who himself had moderately above average luck.
>>12499012
If so then it actually is more than enough to dismiss this as misuse of statistics because that's well within the possibility of happening.
The issue is whether they actually proved his odds were 1 in 10 million or if they just did some sleight of hand with the numbers.

[Anonymous No.12499957]

>>12499947
They did a slight of hand with the numbers read the posts above
This is proof applied stats niggers should never touch mathematics

[Anonymous No.12499997]

>>12499837
ok Dream, whatever you say

[Anonymous No.12500038]

>>12476730
if you want to just ignore any stats questions with this you can literally just look at the code and see how drops are generated, then simulate it? then zero arguing over stats even if it is trivial

[Anonymous No.12500062]

>>12476730
codemonkeys is this code good or is it shit?

[Anonymous No.12500084]

>>12500062
it's fine. too trivial of a snippet to say much of anything about the skills of the coder.

[Anonymous No.12500094]

>>12500084
He has more in the last two pages of the report.
https://drive.google.com/file/d/1yfLURFdDhMfrvI2cFMdYM8f_M_IRoAlM/view

[Anonymous No.12500105]

>>12500094
I haven't bothered to verify that what his code is doing is correct, it's possible to have fine code that simply doesn't do what the author purports. But the code looks fine.

[Anonymous No.12500111]

>>12500105
Thanks anon

[Anonymous No.12500148]

>be dream
>pick a whole number between 1 and 100 trillion
>was it 7?
>there was just 1 in a 100 trillion chance i could have guessed it
>its far less than my odds of getting those exact pearl trades
>mfw I've just proven myself to be innocent and not cheating
this is actually dream's retard logic

[Anonymous No.12500161]

>>12500148

[Anonymous No.12501013]

>>12499904
yes, it's 1 in 25600 not 1 in 6300

[Anonymous No.12501149]

>>12499947
You actually think luck is a thing and then expect anyone to even finish reading your post. This is a giant clickbait generator for retards to feel comfy thinking they are on side with mathematics man. If anything, I bet this whole debacle is clout farm by Dream et al.

[Anonymous No.12501240]

>>12499816
>>12498820
>>12498766
>>12498764
>>12498739
>>12498719
>>12476730
why doesn't this dumb cunt just admit he uses mods and also scripts videos and move on? drop the submission, it's not even first place WITH cheating, and continue making fake videos for kids and making a shit ton of money. is he so bored with the money and so narcissistic that he has to do this? or is it calculated drama to keep himself relevant, perhaps minecraft vids are doing too well again? i don't get it

[Anonymous No.12501245]

>>12501240
Because he knows it's not about who's right and who's wrong. It's about public perception.

[Anonymous No.12501265]

>second opinion came out
>it's a meager 1/100,000,000 chance
Oh ok
>>12501240
Speedrun people take this ultra seriously and basically all of his runs can be invalidated because of this.

[Anonymous No.12501271]

>>12501265
That is just because speed run judges are retards.

[Anonymous No.12501362]

>>12501240
>so narcissistic
Narcissism is common among the subset of people who seek fame.

[Anonymous No.12501403]

>>12501265
did you see the video, he said that the odds weren't 7.500 trillion but it was off by 7.4999 trillion which is still statistically significant, this is one out of 100 million. his whole video goes over lies and how u can manipulate stats for your advantage then at the very start of the video that's what he does. the hypocrisy from this guy holy shit

[Anonymous No.12501539]

>>12501403
yeah he also employs non-statistical arguments throughout the video which he's criticised in the past
also remember when the mod team offered to hire a third party to investigate it and he said no because it will be biased in favour of whoever hires it
OMEGALUL GUESS WHAT HE DID

[Quickly No.12501640]

>>12501240
the only reason I'm giving him a pass is because his handle is literally Dream
Something that isn't real.
>>12501539
Hired an astronomer instead of a defense attorney?

[Anonymous No.12501701]

How does the Java RNG work? Is the scrambling deterministic?

[Anonymous No.12501756]

>>12476730
This is how you do a speed run on youtube.
You play the game normally up to point X then you run through until you get killed.
You play the game normally up to point Y then you run through until you get killed.
You play the game normally...
Now you go into your editor and cut/paste together segments from X Y Z and then you post it all to youtube with the title 'I BEAT THE ENDERDRAGON WITH ONE SPEEDRUN'

However I think this guy actually is a good player and can probably speed run the whole game if he wanted, you only hate him because he figured out the youtube algorithm so he gets a bunch of views.

[Anonymous No.12501797]

>>12501701
They (the mods) go over it in the video and explain in detail in their paper, but basically the RNG is constantly scrambled such that it's impossible Dream could have been unintentionally manipulating it or it "just glitched out."
Furthermore there's two anomalous sets of RNG in question, one from a drop table and another that's a simple 50/50 "it drops or it doesn't," and the RNG behind both works completely differently and is completely independent from the other. The latter case also only fell under scrutiny because the first anomaly was noticed, which destroys Dream's shitty argument about the samples being cherry-picked.

[Anonymous No.12501897]

>>12476730
the PhD guy he hired said Dream had a 1/1,000,000 chance of getting as lucky as he did

[Anonymous No.12501926]

Why does he care what the informed segment of his audience thinks? He doesn't need them. He's going to get millions of kids' views on his videos no matter what

[Anonymous No.12501936]

>>12501897
No he said that the entire Minecraft speed running community had a 1/100,000,000 chance of *anyone* getting that lucky, which is obviously far worse for Dream.

[Anonymous No.12501955]

>>12487806
>I am unconvinced. Stats can't be used to prove an election is rigged, why pretend it counts for minecraft speed runs?
You'd be surprised. Ever wonder why sports follow rigid sex divisions but not the military? It's because winning is important in the former.

[Anonymous No.12502014]

>>12487883
this entire argument is only valid if you consider the "cheating switch" to be some independent effect outside of Dream's own control. He obviously has the (alleged, in any reasonable scenario) ability to turn cheats on and off. There is a clear and independent anomaly (the pearl trade rates) repeatably observed over those streams.

Further, you fail to consider that a continuous region of time can be defined which contains only those 6 streams and no others. These 6 are not just an artificially-assembled sample of his 6 best runs. The fact that these runs were consecutive is itself important.

I'm guessing you know all of this already and are just posting in bad faith, which appears quite common in this thread. But I will give you the benefit of the doubt.

[Anonymous No.12502036]

>>12502014
>There is a clear and independent anomaly (the pearl trade rates) repeatably observed over those streams.
Don't forget the simultaneous anomaly of his Blade Rods dropping at a roughly 70% rate (215 drops / 311 kills), when they're meant to drop 50% of the time, something that only fell under scrutiny because of the pearl bartering anomaly and operates on a completely independent RNG system of its own from said bartering, with no relation to each other than being the two primary progression factors necessary for a speedrun
Luck sure works in mysterious ways!

[Anonymous No.12502065]

>>12501756
Retard

[Anonymous No.12502289]

>>12501701
Java LGC RNG is pretty dogshit and is indeed deterministic. But it would be pretty hard to exploit even if you start with the same RNG seed because so many world events use RNG. The only way to really do it would be recorded gameplay input. But at that point you might as just well increase the drop chances.

[Anonymous No.12502373]

It's weird seeing a verified particle physicist get brigaded by dream stans.

[Anonymous No.12502395]

>>12501240
lel
You basically proved he has no reason to cheat and you don't even realize it.

[Anonymous No.12502406]

>>12502395
explain.

[Anonymous No.12502451]

>>12502014
Basically a bunch of butthurt minecraft speed runner simps came here to validate mathman youtube only to be btfo because it is flimsy argument that relies on questionable analysis.
>>12502036
They aren't completely independent, they are running on the same computer on likely the same instance of java. Only an absolute degenerate thinks different calls imply independence. I actually do shit with my life, maybe some CS cock smoker knows more but I bet Java acts unexpectedly if you leave it running especially with no ECC for extended periods of time.

[Anonymous No.12502500]

>>12502406
?
Just read his comment. Dream has zero reason to cheat . His speedruns were the least watched videos on his channel. And the one in question wasn't even uploaded to YT.

[Anonymous No.12502517]

>>12502500
Psychopaths are like 1% of the population odds of him faking his motivations are higher then him not cheating.

[Anonymous No.12502555]

>>12502500
>no motive i can think of therefore can't be him
you're not this retarded right? tell us you're just trolling

[Anonymous No.12502570]

>>12477543
>white race superiority
See, anyone advocating for the validity of IQ differences in races implies the superiority of asians, not whites.
>hurrdurr racsism
just doesn’t fly as an argument

[Anonymous No.12502571]

>>12502500
thanks for outing yourself as an underage retard

[Anonymous No.12502575]

>>12502555
Who are you quoting?

[Anonymous No.12502639]

>>12494393
now compare the number of people playing the lottery and people playing minecraft speedruns

[Anonymous No.12502786]

>>12499502
>>12499498
>>12499494
Lol Reddit btfo /sci/ yet again
Holyshit this place is pathetic

[Anonymous No.12502848]

>>12499494
>>12499498
>>12499502
How will he respond to this? He's banning people mentioning it and his appeal to PhD authority only worked on his younger fans.

[Anonymous No.12502850]

>>12502848
there's also 0 proof he actually hired a phd astrophysicist, he only mentioned a company name which if you look up on google there's barely anything on it. all it took was phd and harvard in the same sentence and his younger audience were hooked

[Anonymous No.12502854]

>>12502850
Yeah it's a wix site that was probably set up a few days before his response video, he probably just paid some bum off a paper site a few hundred for this shitty paper.

[Anonymous No.12502884]

>>12502850
The company's website is literally filled with anime gay porn and dick pics right now. kek

[Anonymous No.12502926]

TODAY GUYS IM GONNA TRY TO BEAT MINECRAFT WHILE A HUNDRED 2B2TARDS CALL ME A NIGGER,CAN I BEAT THE GAME BEFORE YOUTUBE HEARS THEM SAYING IT AND BANS ME? WE ONLY HAVE A ONE IN 12 TRILLION CHANCE OF DOING IT GUYS!

[Anonymous No.12502989]

>>12497244
statistics is reliable. The reproduction crisis stems from mostly fraud, eg faking the numbers, intentional or inintentional mistakes and simply not reporting negatives. The APA (psychology) recently implemented a policy that requires researchers to pre-report intended studies, this alone wil almost certainly significantly reduce failure of replications.

[Anonymous No.12502996]

>>12497265
>what is a priori information?

[Anonymous No.12503051]

>>12499999

[Anonymous No.12503061]

>>12497265
Why does this obvious truth make midwits seethe so much?

[Anonymous No.12503072]

>>12502884
I can't find that stuff.
Any screenshots?

[Anonymous No.12503189]

>>12503072
Lmao they disabled the comments kek.
pic related is one example, the rest was pretty much like /b/ catalog

[Anonymous No.12503216]

>>12503189
Kek
Nice

[Anonymous No.12503260]

>>12502848
he himself said it
he won't

[Anonymous No.12503285]

>>12501926
No idea. Probably legit narcissism. It’s like arguing about the existence of Santa Clause, or that WWE is actually real and not scripted at all. No one cares, adults who watch wrestling enjoy it either way and it’s pointless trying this hard to convince a bunch of kids because they’ll believe whatever you say anyways.

[Anonymous No.12503289]

>>12502451
Again: this was discussed in the paper.

[Anonymous No.12503325]

His response is pretty shit

[Anonymous No.12503362]

Statistics are not science.

[Anonymous No.12503410]

>>12503325
Nice. Who made this?

[Anonymous No.12503433]

>>12503410
I first saw this image posted by a /vg/tard. I don't know if it's him who posted it

[Anonymous No.12503445]

>>12503433
>tfw /sci/ is being outdone by /v/

[Anonymous No.12503714]

>>12503445
/sci/ has been outdone for years. This place is a wasteland and we just argue over whether numbers are real or not.

[Anonymous No.12504210]

>>12495157
its 262 which is more than enough

[Anonymous No.12504215]

>>12486424
you didn't read the paper then

[Anonymous No.12504224]

>>12492584
midwit post

[Anonymous No.12504314]

>>12501936
Oh come on, what's several orders of magnitude between friends?

[Anonymous No.12504321]

>>12502451
Blaming lack of ECC is laughable horseshit. The odds of random bits getting flipped causing just precisely the sort of corruption that would be necessary to cause interactions between two separate PRNGs but leave the rest of the system seemingly intact is absurd. I agree with the other anon, you are not arguing in good faith if this is the sort of scenario you put forth.

[Anonymous No.12504326]

>>12502517
>Psychopaths are like 1% of the population
Even higher when you limit consideration to those who seek fame (as all ecelebs do by definition.)

[Anonymous No.12504513]

>>12504321
One again, no evidence to support anything you are saying. As shoddy as the statistician. Bits changing and uncorrected is only one easy and guaranteed cause of problems. And that is if there was even a problem and not a fluke or hack, which has not been determined either. Did he have his minecraft open for weeks at a time? It isn't just Java that is questionable. Literally every component and software can fuck up. But you are already dick riding stats man because he brave.
Here is a trick for problem solving, though it won't help sub zero IQs. Identifying difficult problems is done by eliminating every area which is not a problem first. They have failed to even look.

[Anonymous No.12504518]

>>12504513
You're really doubling down on hardware malfunction to explain his luck? Fuck off retard.

[Anonymous No.12504524]

>>12504518
You actually think luck exists. Draw the line and see if luck helps you.

[Anonymous No.12504551]

>>12502926
top kek anon

[Anonymous No.12504701]

>>12504513
>Le statsman didn't take le rotational velocidensity inti account so his analysis is biased
>Le software can fuck up because of le cosmic interferences changing the value of single bits in memory

Extremely intelligent post. You're way too smart for this board.

[Anonymous No.12504769]

>>12504701
Kek based BTFO'd that fucking retard.
>how do you know I murdered this person? Sure there is video evidence showing me doing it, but how can you rule out the possibility that it’s a result of faulty hardware!?!? That surveillance footage isn’t solid evidence! You didn’t even rule out the possibility that a beam of cosmic energy interrupted the pixels and made it look like me!!! REEEEEEEEEEEEEEEEE

[Anonymous No.12505313]

THIS IS THE MOST AUTISTIC THREAD I'VE EVER SEEN

[Anonymous No.12505319]

>>12502065
No that's exactly how fake speedruns are made in Minecraft.
You play the game 100x and record each time then splice it all together with 2 guys shouting OMG and OH WOW YOU ACTUALLY DID IT overtop of the edited video.

[Anonymous No.12505396]

>>12503325
brainlet

[Anonymous No.12506004]

>>12483041
The Negative Binomial Distribution is for the number of successes given a fixed number of failures[math]\left[P(X=k)=\left(\begin{matrix}k+r-1\\r-1\end{matrix}\right)(1-p)^kp^r\right][/math]. You're trying to calculate the number of trials for a fixed number of successes[math]\left[P(X=k)=\left(\begin{matrix}k-1\\r-1\end{matrix}\right)p^r(1-p)^{k-r}\right][/math].

[Anonymous No.12507443]

apparently the statistician got doxed and he graduated from harvard and got phd from someone else, i wouldn't know tho.

[Anonymous No.12507446]

>>12507443
where did you hear this?

[Anonymous No.12507532]

>>12507446
Seems like it's common knowledge in Minecraft Java Edition Speed-running Community Discord. The author leaked it himself somehow, many chat mods seem to know it but I couldn't be bothered to ask for it so I just left the discord.

[Anonymous No.12507546]

>>12476730
https://statmodeling.stat.columbia.edu/2020/12/24/dream-investigation-results-official-report-by-the-minecraft-speedrunning-team/
Columbia University professors of statistics are weighing in now LMAO

[Anonymous No.12507554]

>>12507532
Lmao wtf? Who is he? What color is his skin?

[Anonymous No.12507578]

>>12507532
Come on anon please ask them for it

[Anonymous No.12507594]

>>12507554
Only things I know are, he's an assistant professor and he has been cited over 10.000 times, he's an astrophysicist, graduated from Harvard and has a PhD from another university. Seen it from one of the channels at the discord from a chat mod called PhD of eating cratons or something, he and some PhDDreamBadSad something guy knows, the bearer of the information isn't only them tho I don't think, they might have said some new things considering they're children and seem to be bragging about it, I didn't care and just left the discord after looking at the discord logs.

[Anonymous No.12507611]

>>12507594
please find out who he is for us

[Anonymous No.12507616]

>>12507594
Shouldn't be hard to find, not that many astrophysicists have 10k + total citations.

[Anonymous No.12507697]

>>12507616
>>12507611
Even though I wonder who the guy is, I don't wonder it enough to beg kids to tell me his name or check every Harvard graduate's positions and citation counts but I'm pretty sure you can deduce it from citation count, Harvard graduation and PhD from another university facts tho I don't know any of these are the truth but more than one person seems to be saying it so they're most likely true, you can even find more information digging today's chat logs of the above-mentioned moderators and some others, I haven't spent that much time looking at chat logs and they've probably said more stuff.

[Anonymous No.12507710]

>>12507697
I'm looking into it right now, so far no one fits the criteria. I'm assuming he has 10-20k citations since if he had 20k+ they would probably say he was cited over 20k times rather than over 10k times. I'm gonna check if any of them could be our guy.

[Anonymous No.12507730]

>>12507697
>>12507710
I've only found some vaguely similar individuals. Purdue, UDubs. Can't find anything exactly the same.

[Anonymous No.12507750]

>>12507697
>>12507710
>>12507730
It's Neil deGrasse Tyson

[Anonymous No.12507751]

looking at the logs i found that its been once publicly posted in speedrunning discord on #coding-math channel but is now deleted so some regulars there might know it aswell, not only mods know it so people can ask regular not-moderators there aswell.

[Anonymous No.12507757]

>>12507751
how do you get access to that discord

[Anonymous No.12507761]

>>12507757
https://discord.com/invite/jmdFn3C

[Anonymous No.12507785]

Why would they not use the negative binomial distribution?

[Anonymous No.12507822]

>>12507730
Somehow only this guy looks like he could be the one we are looking for
https://www.linkedin.com/in/francois-fressin-69414368
>didn't graduate from Harvard but was a Harvard postdoc
>works (or worked?) at Harvard Smithsonian Center for Astrophysics
>literally has in his bio "10,000+ citations"
>data scientist
not sure tho, I only looked through astrophysics department, Dream's expert might be from somewhere else

[Anonymous No.12507848]

>>12487806
Election fraud can be proven with stats which is what Edward Solomon has done on his YouTube channel. But in establishment controlled institutions weight given towards what someone may say in response is the controlling factor to if something is done about fraud.

[Anonymous No.12507855]

>>12507822
directly quoting from discord,
"they have degree(s) from harvard"
"They have a separate degree from Harvard for physics and some Astro stuff"
so you might be right about that. the guys seems to be potentially owning www.photoexcitation.com aswell

[Anonymous No.12507859]

>>12507848

[Anonymous No.12507924]

>>12507855
He also has a Wix page

>francoisfressin dot wix dot com slash francoisfressin
Photoexcitation was also made using Wix
Pretty sus but could be coincidence, any other clues from discord?

[Anonymous No.12508134]

>>12507546
topkek
when is harvard going to weigh in and disown their graduate

[Anonymous No.12508136]

>>12504769
I know you are butthurt. Maybe it was when Dream wiped his dick clean of your excrement and the last drips of cum from its shaft on your frothing tongue. Did it leave a bad taste or what? The person green texting REE... is usually the apoplectic moron. Gather your strength bottom bitch for more is coming to those fragile pouting lips of yours.

[Anonymous No.12508163]

>>12507924
from what I've gotten from that discord server, dream's manifesto author is:
>assistant prof
>PhD not from Harvard
>Harvard affiliation (apparently graduated from Harvard)
>PhD from other prestigious university
>apparently no background in statistics (idk why he claims to be astrostatistician then)
>over 10k citations
>"focuses on planets and space shit mostly" astro
>separate degree from Harvard for physics and some astro stuff
he apparently leaked his own name somehow, mods on discord know but refuse to share it obviously

[Anonymous No.12508245]

>>12508163
Is there a way to look up most commonly cited individuals by field?

[Anonymous No.12508367]

>>12508136
>being so upset you write this retarded fag shit
Dream just admit that you got BTFO

[Anonymous No.12509165]

>read the report by this "astrophysicist"
>he says a barter can give 4-7 ender pearls
>mfw it's 4-8
>mfw this fucking """"""astrophysicist"""""" couldn't even double check his own work
>mfw I found out his website didn't even fill in the FAQ properly
The literal STATE of Dream

[Anonymous No.12509276]

>>12508245
yes, you can do that on google scholar pretty much
by searching through researchers

[Anonymous No.12509301]

>>12508163
if you search the term "dox" in the mentioned discord many non-moderator people seem to know it, if desperate, some anon can go ask them, apparently it never spread outside the discord.

>>12508245
you can sort researchers by publications at app.dimensions.ai but too many people to check by hand.

[Anonymous No.12509375]

>>12509301
ik, but nobody wants to leak his name or anything

[Anonymous No.12509440]

>>12507822
could also be this guy:
https://www.astro.princeton.edu/~gbakos/CV/gbakos_cv_letter.pdf
>assistant prof
>Princeton
>graduated from Harvard CfA and from other Hungarian uni
>10k + citations
is this him?