0628: "Psychic"

[frog42]

but... 5 people explained it... in this topic... on this very page...

I am not 5 people! Argh! Basic math! J/K

[neoliminal]

1st.: Humans are incapable of creating random numbers. They are particularly bad at it, in fact. Ask people for random numbers and you'll find clustering base on culturally significant numbers.

2nd: Because BOTH parties in the comic are human, we can assume that there is a much higher than 1% chance of there being a match. There are a couple of assumptions in that, like that the first party didn't generate his answer through a truly (pseudo) random means.

So Randal fails.

HOWEVER: If two different entities create truly random integer in the range of 1 and 100 (inclusive), then the chance that the two numbers are the same is 1%.

fin.

[Agent Foxtrot]

1) We have LaTeX. You can show any mathematical formula you want on these forums. For example,

\Gamma\left(z\right)={1\over{z}}\prod_{n=1}^\infty\frac{\left(1+\frac{1}{n}\right)^z}{1+\frac{z}{n}}

2) You're assuming that he also randomly picks a different number each time. Sure, that further decreases his probability of success, but why wouldn't he just pick 43 every single time? Out of 100 girls, at least one of them is generally bound to pick 43 as well. Therefore, P=0.01.

Grr.... Despite your fancy math, you still FAIL. He can pick any random number he wants and the girl can pick any random number she wants. The probability will remain 1%!!!! EVERY TIME there are 100 numbers to choose from and ONE is right. Probability doesn't get much easier than this. 10,000 possible outcomes with 100 being positive. 100/10,000 = .01!!!!

Relax, chief. Easy on the coffee. Or anabolic steroids. Or whatever. You just said exactly what I did. 0.01 = 1%.

[tzvibish]

1) We have LaTeX. You can show any mathematical formula you want on these forums. For example,

\Gamma\left(z\right)={1\over{z}}\prod_{n=1}^\infty\frac{\left(1+\frac{1}{n}\right)^z}{1+\frac{z}{n}}

2) You're assuming that he also randomly picks a different number each time. Sure, that further decreases his probability of success, but why wouldn't he just pick 43 every single time? Out of 100 girls, at least one of them is generally bound to pick 43 as well. Therefore, P=0.01.

Grr.... Despite your fancy math, you still FAIL. He can pick any random number he wants and the girl can pick any random number she wants. The probability will remain 1%!!!! EVERY TIME there are 100 numbers to choose from and ONE is right. Probability doesn't get much easier than this. 10,000 possible outcomes with 100 being positive. 100/10,000 = .01!!!!

OK, I guess that answers my second question.

And now that I think about it, it answers the first one too. If there are 100 girls all picking random numbers, then out of that group of tests, there are 10,000 guesses. 100 of those will be correct guesses. So the probability stays close to 1% no matter how many tests there are. mmmmmmm, limits.

http://ruleroftheinterwebs.blogspot.com
-featuring xkcd comic strips!!

[Agent Foxtrot]

1st.: Humans are incapable of creating random numbers. They are particularly bad at it, in fact. Ask people for random numbers and you'll find clustering base on culturally significant numbers.

2nd: Because BOTH parties in the comic are human, we can assume that there is a much higher than 1% chance of there being a match. There are a couple of assumptions in that, like that the first party didn't generate his answer through a truly (pseudo) random means.

So Randal fails.

HOWEVER: If two different entities create truly random integer in the range of 1 and 100 (inclusive), then the chance that the two numbers are the same is 1%.

fin.

I'm inclined to agree with you, but different numbers are significant for different people. The number 32 is particularly significant to me, but most others, not so much.

Unfortunately, a rigorous study would require an extremely large sample size, so we'll never know.

[neoliminal]

1st.: Humans are incapable of creating random numbers. They are particularly bad at it, in fact. Ask people for random numbers and you'll find clustering base on culturally significant numbers.

2nd: Because BOTH parties in the comic are human, we can assume that there is a much higher than 1% chance of there being a match. There are a couple of assumptions in that, like that the first party didn't generate his answer through a truly (pseudo) random means.

So Randal fails.

HOWEVER: If two different entities create truly random integer in the range of 1 and 100 (inclusive), then the chance that the two numbers are the same is 1%.

fin.

Oh, I forgot to mention (and this will cause all kinds of follow up) that if the first person DOES pick a truly random number and the second person is Human, then the odds ARE 1 in 100 (1%) even though the Human is prone to clustering when choosing a number "at random".

[frog42]

Relax, chief. Easy on the coffee. Or anabolic steroids. Or whatever. You just said exactly what I did. 0.01 = 1%.

J'accuse!!! You were close. It's the alcohol. Not everyone is on the sunny side of the world right now

[neoliminal]

I'm inclined to agree with you, but different numbers are significant for different people. The number 32 is particularly significant to me, but most others, not so much.

Unfortunately, a rigorous study would require an extremely large sample size, so we'll never know.

Actually I think we could get away with a sample size of... just pulling this out of thin air here... 32.

[frog42]

1) We have LaTeX. You can show any mathematical formula you want on these forums. For example,

\Gamma\left(z\right)={1\over{z}}\prod_{n=1}^\infty\frac{\left(1+\frac{1}{n}\right)^z}{1+\frac{z}{n}}

2) You're assuming that he also randomly picks a different number each time. Sure, that further decreases his probability of success, but why wouldn't he just pick 43 every single time? Out of 100 girls, at least one of them is generally bound to pick 43 as well. Therefore, P=0.01.

Despite my drunken stupor and apology, you still FAIL. In number 2, you say that him also picking a random number further decreases his probability. Which is what I was saying was wrong. Apparently I need at least 500ml more of beer for you to get the drop on me, sir! To the combini!!!

[OCEAT]

I think it might've been better if there was another panel where the girl said, "You want to see X". The guy responds, but the girl cuts him off, saying "It wasn't a question." One more panel, no dialogue.
I dunno, has that trope been overused? Good comic anyway; I'll try this sometime.

[Caffeinated]

Try this one (works about 40-50% of the time).

So you've got a number now? Two digits, both odd, both different?

Spoiler:

37

Now, think, don't say, a tool in a workshop.
Now, think, don't say, a colour.
Now think, don't say, a vegetable.

Hrm....

Spoiler:

Why are you smashing carrots with red hammers?

If you get 2 or 3 out of 3, bet you can do it again

Spoiler:

This time you're sawing blue potatoes.

Works best on children under 13 or so

The first one, honestly, freaked me out quite a bit. Then I was highly disappointed in myself for proving I was quite the average predictable human.

I also tried it over an instant messenger, and it did not work in that case. Her number was '91'. Not sure if it's not as effective that way, or if she was just an exception in any case. I definitely want to try this one out some more.

The second almost worked. I was Hammer, Purple, Carrot. I know I chose that because they are all six letter words. With the number six already in my head, I had to continue the pattern. Not sure if I would have said 'carrot', though. That one I might have thought of just because someone else mentioned it earlier.

Now I have a cousin who does a trick like this, with guessing girl's middle names. He gets a damn high percentage of them right, apparently, though I've only seen the trick in action once, where it DID work. He refuses to explain how, but I think I have it reasoned out.

When he does the trick, he does his little psychic spiel, and asks them to give their first and last names. Then he tells her her middle name. His answer when I watched him do it was 'Lynn'.

I know Lynn is one of the most common middle names for girls, so I'd say the 'trick' is just using the top few common girls' middle names, Lynn, Anne, etc, and picking which one fits best within the first and last names, and therefore the most likely to be chosen, which would give you a pretty high chance of getting it right.

I think being the only one sober at the time is probably the reason I'm the only one able to understand this... or my friends are just logically-challenged.


Oh, and I also thought 'That's one more than 42!' when I read the comic, for the record.

[jeszjesz]

My response to a correct guess would be to say "Do it again", which I suspect would rather spoil the fun 99% of the time.

[Max2009]

Another good one...
"I'm so impressively intelligent (and slightly psychic). Want me to prove it?"
"Sure"
"I want you to think of a number. Any number. Minus 6, a billion and twelve, 76.9, doesn't matter. Got one? Now, what is it?"
"Uh.... <some number>?"
"Wrong!"

Thanks, Hitchhikers Guide!

I did that one to someone just last week, and it worked surprisingly well.
Me: You have no idea how many orders of magnitude my intelligence is over yours.
Gullible person: Oh yeah? Prove it!
Me: Pick a number. Any number you can think of. It doesn't even need to be a rational number.
GB: What's a rational number.
Me: Never mind. Just pick any number.
GB: OK, 93 and a quarter.
Me: wrong.
GB: Wow. How did you do that?

[sircheeziusbreezius]

Someone before mentioned 17, but that works better if you limit it between 1 and 20. 37 Works better for larger ranges. When you get beyond 100, you're probably getting to too many numbers for an average person to think about.

[/quote]
In my experience, when you get to larger ranges of numbers, most people will generally stick below the halfway mark, like if you were to say between 1 and 1000, they seemingly usually pick 500 or lower

[Shadowman615]

1st.: Humans are incapable of creating random numbers. They are particularly bad at it, in fact. Ask people for random numbers and you'll find clustering base on culturally significant numbers.

2nd: Because BOTH parties in the comic are human, we can assume that there is a much higher than 1% chance of there being a match. There are a couple of assumptions in that, like that the first party didn't generate his answer through a truly (pseudo) random means.

I think that's how David Blaine does a similar trick where he asks someone to pick a number between 1 and 1000. (It's 333)

[neoliminal]

1st.: Humans are incapable of creating random numbers. They are particularly bad at it, in fact. Ask people for random numbers and you'll find clustering base on culturally significant numbers.

2nd: Because BOTH parties in the comic are human, we can assume that there is a much higher than 1% chance of there being a match. There are a couple of assumptions in that, like that the first party didn't generate his answer through a truly (pseudo) random means.

I think that's how David Blaine does a similar trick where he asks someone to pick a number between 1 and 1000. (It's 333)

I normally pick the smallest value offered. For example if someone asks Pick a number between 1 and 10, (and they actually mean inclusive and integer), I pick 1. Because no one does it. Except me.

Also, in Rock-Paper-Scissors, I always play Rock.

[Sprocket]

GAH! Skippy's card trick from the Brother's Bloom!

[kafkadog]

Also, in Rock-Paper-Scissors, I always play Rock.

Or are you just saying that so you can actually play scissors to my paper?!

[JCCyC]

Ace of Spades is chosen freely about 34% of the time.

[neoliminal]

Also, in Rock-Paper-Scissors, I always play Rock.

Or are you just saying that so you can actually play scissors to my paper?!

I don't know. Let's try... ready? Rock, Paper, Scissors, Shooooot:

Spoiler:

Rock.

What do you know, I did play rock.

[Dr. Bronowski]

Best xkcd strip in a while. Thank you.

[FCN]

The number that I thought of was 20. Or at least that's what I tried to think of - there was a rounding error.

[BrianXVX]

Ask someone to think of the first number that pops in their head between 1-4, don't say it but just think about it. Tell them to picture this number in their head, and guess it's three. Then say let's increase the odds and say think of a number between 1-10 and guess seven.

Apparently, there is like a 80-90% chance that both are correct. It's that much more effective since you guessed right two separate times in a row. Even if you miss one, you won't look bad since chances are you will still get the other right.

It's hard to belive, but I've tried it out three times and gotten both numbers every time. The key is for them to not think about it too much. I just tried a fourth time and got both wrong, but when I asked she sat there and thought for a minute. She admitted those were her first choices, but changed them since she thought there was a trick to it and was trying to mess me up.

[BrianXVX]

Here is another good one that's worked for me both times I've tried it.

Pick a numbrer between 1-10
Multiply it by 9
Add the two digits together and subtract 5.
Think of a letter that corresponds to that number: 1=A, 2=B, 3=C, etc.
Think of the name of a country that starts with that letter.
Take the second letter in the name of that country and think of an animal that starts with that letter.
Now think of a color.

Spoiler:

"There are no Grey Elephants in Denmark"

You can start out any other way so they pick the letter "D" (aka first letter of your middle name, etc) to make it more personalized for more effect.

[Muvlon]

big "get out of my head randall" when i read the alt-text

[VNVKingdom]

Here is another good one that's worked for me both times I've tried it.

Pick a numbrer between 1-10
Multiply it by 9
Add the two digits together and subtract 5.
Think of a letter that corresponds to that number: 1=A, 2=B, 3=C, etc.
Think of the name of a country that starts with that letter.
Take the second letter in the name of that country and think of an animal that starts with that letter.
Now think of a color.

Spoiler:

"There are no Grey Elephants in Denmark"

You can start out any other way so they pick the letter "D" (aka first letter of your middle name, etc) to make it more personalized for more effect.

I ended up with blue eagles. The Denmark part was spot on though.
Stop being so predictable brain or I'll drown you in alcohol!

[Otto]

This works about 70% of the time, in my experience.

Pick a number between 50 and 100.

But, make sure both the digits are even.

Spoiler:

68

Doesn't work every time, but a surprising amount of the time it does.

I've also found it works better if you put a longer pause in between the two instructions, give the person time to think of the number before springing the second condition on them.

[Walter.Horvath]

I got ninja'd and I wasn't even the better of the two.

Chalk it up to me not being a regular in this section

[cheeseheadtotherescue]

The fact that twice, people have posted the fallacy that 2 randomly chosen number between 1-100 have a 1/100 * 1/100 or 1/10000 chance of being correct, gives me an idea.

If you DO play this trick on someone, regardless of if you end up being right or wrong, tell them (not ask them) that you only had a 1 in 10,000 chance of getting it right, and see how often they believe you. As said previously, we've all accepted this logic at some point in our lives before we wised up, how many people still do? (disclaimer: Of course if they call your bluff, you may have a hard time convincing them that you were just testing them.)

[ianf]

Which isn't very fair, because:

Spoiler:

Denmark is the only country that begins with a D

Not true, actually there's also

Spoiler:

Djibouti

Depends on your background what the answer would be, for some of these questions.

Personally, I like the simple "think of a colour" ... "blue" approach rather than the more complex this, then that, the the other type. The second type looks more like a trick. The first (when it works) is more impressive.

[Cynical Idealist]

Here is another good one that's worked for me both times I've tried it.

Pick a numbrer between 1-10
Multiply it by 9
Add the two digits together and subtract 5.
Think of a letter that corresponds to that number: 1=A, 2=B, 3=C, etc.
Think of the name of a country that starts with that letter.
Take the second letter in the name of that country and think of an animal that starts with that letter.
Now think of a color.

Spoiler:

"There are no Grey Elephants in Denmark"

You can start out any other way so they pick the letter "D" (aka first letter of your middle name, etc) to make it more personalized for more effect.

I ended up with blue eagles. The Denmark part was spot on though.
Stop being so predictable brain or I'll drown you in alcohol!

Actually, there are no purple jaguars in Djibouti.

At least, so I am lead to believe. (I usually pick Djibouti for two reasons: one, it isn't expected, and two, its a fun word to say)

And Walter, you're wrong, there is another country that fits (see earlier in my post)

[hthall]

1st.: Humans are incapable of creating random numbers. They are particularly bad at it, in fact. Ask people for random numbers and you'll find clustering base on culturally significant numbers.

True for most people. I remember an anecdote about a university class in which students tried to simulate a random series of coin flips, while a computer program used basic data compression techniques to try to predict their next keystroke. No matter what the students tried, or how much they practiced, in the long run the computer could always predict well over 50% of the "random" keystrokes of any student--with one exception. There was one student who appeared to be generating truly random data. How? He said he was just using his free will.

2nd: Because BOTH parties in the comic are human, we can assume that there is a much higher than 1% chance of there being a match.

More precisely, if we assume that both parties are drawing independently from the same distribution, the probability of a match is the sum of the squares of the 100 separate probabilities in the distribution. This is only "much higher" than 1% if the distribution is very far from equal, but it is true that any inequality at all in the distribution leads to a probability higher than 1%. If the second party knows the distribution the first party is drawing from, he can do even better if his own distribution is 100% concentrated on the mode of the first party's distribution. For example, looking at the breakdown of numbers from 1 to 20 in the "Human random numbers" thread of this forum, two of us both picking numbers from that distribution will agree around 5.85% of the time, as opposed to exactly 5% of the time if our choices were truly random. However, if one person picks from this probability measure and the second person always guesses "12", they will agree 10.58% of the time.

If two different entities create truly random integer in the range of 1 and 100 (inclusive), then the chance that the two numbers are the same is 1%.

In fact, you can say more: as long as AT LEAST ONE of the two entities chooses a truly random integer, the chance that they are the same is exactly 1%, regardless of what strategy the second entity employs.

[catzilla]

Wrenching white potatoes? Eh?

[EthanW]

I practice several mentalist and mathematic trickses, most PUA's know what these are. My favorite one is asking someone to pick a number between 1 and 5, the most common answer is 3, and then between 1 and 10, 7 is the most common answer. blows peoples minds.

[neoliminal]

1st.: Humans are incapable of creating random numbers. They are particularly bad at it, in fact. Ask people for random numbers and you'll find clustering base on culturally significant numbers.

True for most people. I remember an anecdote about a university class in which students tried to simulate a random series of coin flips, while a computer program used basic data compression techniques to try to predict their next keystroke. No matter what the students tried, or how much they practiced, in the long run the computer could always predict well over 50% of the "random" keystrokes of any student--with one exception. There was one student who appeared to be generating truly random data. How? He said he was just using his free will.

If you find an article or anything that puts this past anecdotal I'd love to see it. I don't think free will exists (and btw, this doesn't change my mind even if it's true) but I do find the subject of interest.

As for your other extensions, yes... thank you for expanding on my original work. I'll be including you when I publish.

[Story]

Whenever I'm asked to pick a number from 1 to 10, I always pick the chromatic number of the plane. Noone can prove they guessed right.

[xsk8rat]

I have this schtick i do with two cards: Ace of Diamonds and King of Spades. I explain that i can tell them apart just by the weight.
That gag has a yield of 50% - better, if i can make up some lame excuse about "the wrong ink" or something like that.

xsk8rat

[Lewton]

There's a variation which works 100% of the time but is easier for the mark to figure out how it's done. (And fails completely if they've seen the episode of Columbo which featured it...)

You: "Pick a number from 1 to 4."
Them: "1"
You: "Look under the phone."
They do, and there's a note with the message "I knew you were going to pick 1".

Of course, if they said 2 you would tell them to look under the lamp, if 3 then look under the blotter etc. By increasing the number of pre-prepared hiding places you could easily extend this to 10 choices, but it would get a bit unwieldy after that.

http://www.youtube.com/watch?v=-RXONfpfwxw Penn & Teller doing a (quite a bit) more elaborate version of this trick

[Faranya]

Try this one (works about 40-50% of the time). Say it out loud to friends. No idea if it will work in text form.

Think of a random two digit number... so, anything between 10 and 99.
Now, that's a lot of numbers, so let's narrow it down a little...
Make both digits odd. So like, 1, 3, 5 etc...
And because double digits are silly, make them both different.

So you've got a number now? Two digits, both odd, both different?

53?

Or another one:
Ask someone these calcs (or similar) really quick. No time to think, just answer.
5+1?
half of 12?
3+3?
3x2?
9-3?
Now, think, don't say, a tool in a workshop.
Now, think, don't say, a colour.
Now think, don't say, a vegetable.

Hrm....

But...I was planing broccoli with a black adze...

You sir, are a liar!

[tobias104]

Here is another good one that's worked for me both times I've tried it.

Pick a numbrer between 1-10
Multiply it by 9
Add the two digits together and subtract 5.
Think of a letter that corresponds to that number: 1=A, 2=B, 3=C, etc.
Think of the name of a country that starts with that letter.
Take the second letter in the name of that country and think of an animal that starts with that letter.
Now think of a color.

Spoiler:

"There are no Grey Elephants in Denmark"

You can start out any other way so they pick the letter "D" (aka first letter of your middle name, etc) to make it more personalized for more effect.

That my be, but there are green ostriches in the Dominican Republic. (Which may or may not be a country, but it's what came to mind)