Coin Flip Experiment

[Camacha]

So the real experiment is whether KSP players can be trusted

[The Jedi Master]

Heads, but with a dirty coin. That's a variable.

[K^2]

That's a cool experiment to do, I one thought of making you guys thinking a number between 1 and 10, and vote the result. Everyone will know it'll come out as a nice Gauss curve, but it'll be a cool experiment to do too.

It wouldn't. If these were fair random guesses, it'd be a uniform distribution. (In reality, as ZetaX pointed out, there would be some patterns there.) In order to get a Gauss, you need to add a bunch of random numbers together and do a histogram. Say, make a hundred pools of 5 d6 each. Roll them all, add up the numbers in each pool, and plot the histogram. That will be pretty close to Gaussian.

[Camacha]

I guess the point of doing 1 to 10 is that it is not random and will not be evenly distributed. I think people will have certain predispositions of wanting to be not too far from (...) but not too close to (...), with some added or even overpowering noise from lucky numbers and such.

People are not random number generators, so the answers they produce will be influenced by all kinds of psychological and cultural phenomena.

Edited June 9, 2014 by Camacha

[HeadHunter67]

This is a bad experiment, since every coin flip happens under different conditions that are impossible to validate.

Regardless of "conditions", the odds of either heads or tails on every individual coin flip is 50/50. Let's not pretend that reporting bias or dishonesty is unique to informal experiments.

[StuntMax]

I'm having a little wee problem here... Euro coins have no heads on either face, this makes all the coin flipping thing pretty confusing.

Luckily, I found around an old italian pre-euro coin which has a proper head and a proper tail.

If you ever take a trip in euroland and have to flip a coin, be sure to agree beforehand on which side is to be considered "head"....

PS. I KNOW, it is usually considered "tails" the face with the face value of the coin...

[Jenteb07]

I'm having a little wee problem here... Euro coins have no heads on either face, this makes all the coin flipping thing pretty confusing.

Luckily, I found around an old italian pre-euro coin which has a proper head and a proper tail.

If you ever take a trip in euroland and have to flip a coin, be sure to agree beforehand on which side is to be considered "head"....

PS. I KNOW, it is usually considered "tails" the face with the face value of the coin...

The belgian euro coins do show a face on one side of the coin

[goncaloeaguiar]

Head is considered where the number is, for example, the big 1 for 1 €.

Tails is by exclusion the other side.

[ZetaX]

That's contradictory to what it is called in germany: Kopf und Zahl (Head and Number). Thus heads is supposed to be everything but the number, i.e. either the number is tails or heads=tails, with the latter sounding more absurd.

[StuntMax]

Head is considered where the number is, for example, the big 1 for 1 €.

Tails is by exclusion the other side.

This is strange... i've seen quite a bit of coins (travelled a lot in europe before the EU), and in almost all of them the "head" was engraved on one face, and the value of the coin on the other (I think the only coin I've seen with the head and the value on the same face is the US quarter).

I conducted a small research on a small control group (i.e. I asked my office colleagues) and all of them agree that the side with the number is the "tails" ("Croce" - "cross" here in Italy), thus the other side is the "head", even if there's no actual "head" engraved on it.

This makes me define a new principle: "Playing heads or tails in a foreign country may lead to a brawl"

[einsteiner]

So the real experiment is whether KSP players can be trusted

Is it really that hard to flip a coin?

[Tommygun]

Is it really that hard to flip a coin?

Have you seen how many crash videos are posted on this site?

Just saying.....

[RogueMason]

Well, I got heads and brought the whole thing to a tie, 42-42.

[WinkAllKerb'']

none:

this or non stop fallin' / levitate

[Chiboko]

*Mildly off topic*

Most people when asked to come up with a random set of figures would create something like the following: HHTHTHHTTHTTTH

Such an even distribution is in fact unlikely and a truly random series would look something like this: HHHHTHHTHTTTTT

Similarly when asked to place random dots on a screen the majority of people are likely to distribute them more or less evenly.

In reality dots placed randomly on a screen will have areas with large groups of dots in a small area.

Starting at the 762nd decimal place Pie repeats the number 9 six times in a row.

TLDR: True randomness has clusters

Edit: In a strange twist I flipped a coin and it landed on heads, this caused the result to be a perfect 50/50 split with 43 on each side

Edited June 10, 2014 by Chiboko
Damn

[xcorps]

I don't understand why you included Pi as part of randomness cluster proof.

So the real experiment is whether KSP players can be trusted

This thread doesn't test that KSP players are untrustworthy, it only tests the people who answered the poll MIGHT be untrustworthy. Except winkallkerb. He posted a picture.

Euro coins have no heads on either face

Further proof that the EU is a farce hoisted upon its own petard.

No, studies show it will accumulate at 7.

Link?

[Chiboko]

Pi has no discernible pattern and is such a good example of a random sequence that companies in need of some random numbers (Useful as test data for programs to see how they handle various data inputs) often just use a section of Pi because it is more reliable than having an employee make one up (When creating a random sequence of digits most people have at least one number they use too much and one that they use too little)

[ZetaX]

Because no company can afford to call a random number generator¿

[xcorps]

Pi has no discernible pattern and is such a good example of a random sequence that companies in need of some random numbers (Useful as test data for programs to see how they handle various data inputs) often just use a section of Pi because it is more reliable than having an employee make one up (When creating a random sequence of digits most people have at least one number they use too much and one that they use too little)

Thanks. (too few characters so I have to type stuff extra)

[PakledHostage]

Because no company can afford to call a random number generator¿

Or do a Google search for a random number generator website such as RANDOM.ORG, for that matter...

[Chiboko]

Due to the nature of the logical functions used by computers AND OR NAND NOR ect... they are incapable of generating truly random numbers, most RNGs use a complicated algorithm to generate a series of seemingly random numbers but with a large enough sample size patterns and trends begin to show up. This is fine for most purposes but when an extremely large sample size is needed or if the program being tested is of great importance RNGs are considered to be not random enough.

(This information is a few years old and may be out of date)

[PakledHostage]

This is fine for most purposes but when an extremely large sample size is needed or if the program being tested is of great importance RNGs are considered to be not random enough.

That's why RANDOM.ORG uses atmospheric noise to generate its random numbers. I recall seeing other online random number generation sites that claim to use quantum effects.

[phoenix_ca]

Pi has no discernible pattern and is such a good example of a random sequence that companies in need of some random numbers (Useful as test data for programs to see how they handle various data inputs) often just use a section of Pi because it is more reliable than having an employee make one up (When creating a random sequence of digits most people have at least one number they use too much and one that they use too little)

I don't care that this is technically against the rules. It's too good to pass-up:

Data sure loves the number seven.

[Chiboko]

I never thought about using noise to generate random numbers despite having used it to generate a random height map in the past *curtseys*

[Camacha]

This thread doesn't test that KSP players are untrustworthy, it only tests the people who answered the poll MIGHT be untrustworthy.

Which realistically is about the same, assuming enough people answer. That is not good enough if you want to use it in reseach for a scientific article, but hey, we are not, so we're good.