News:

"At the teaparties they only dunked bags into cups of water...because they didn't want to break the law. And that just about sums up America's revolutionary spirit."

Main Menu

An experiment with Babylon and LMNO

Started by LMNO, July 01, 2009, 10:17:23 PM

Previous topic - Next topic

Golden Applesauce

It's not quite a coin flip.  If the game was structured a little differently, with Babylon calling four numbers for each roll, and scoring a 'hit' if one of those four were correct, ('miss' otherwise,)  I think you would agree that the game wouldn't be a coin flip exercise.  The current version is just a special case of the above, where Babylon limits himself to only two groups of four numbers to call.  In fact, you could argue that it works a little against him in that if he mahgiqueally manipulates the dice to roll more 3's and 4's, the control groups will also score more accurately.
Q: How regularly do you hire 8th graders?
A: We have hired a number of FORMER 8th graders.

rubickspoop

What happened to this experiment? I've been lurking and anticipating the results. Now I'm kind of disappointed. Did Babylon puss out?
I'm a celebrity... Get me out of here!

Requia ☣

Babylon and the roller both posted their results/predictions, we're pretty much waiting for the receivers to confirm that Babylon made the predictions first and for someone to tally everything up.
Inflatable dolls are not recognized flotation devices.

Cain

Oh, sorry.

This is what LMNO sent me:

4
4
3
2
2
5
4
5
2
3
4
2
5
3
2
6
2
6
5
5
3
5
6
3
4
2
2
4
4
1
3
4
2
4
2
4
2
5
1
4
6
1
4
6
2
1
6
2
1
1
4
5
1
6
3
3
4
2
1
4
2
5
3
6
5
5
3
3
4
3
4
2
4
4
4
3
2
6
3
1
4
6
3
5
2
1
6
4
1
3
3
4
1
4
1
3
4
5
6
3


Does someone want to do a tally or something?

Roaring Biscuit!


Dalek

Can't wait for the results to come out

Roaring Biscuit!

my calculations aint perfect, as two of LMNO's rolls went awol.  So my tally is out of 98.

BH correctly predicted 66 rolls as L/H, meaning he was correct 67.3% of the time.  whether this is significant is yet to be decided (in my opinion).

He correctly predicted the exact number 16.3% of the time.

So we can test this more thoroughly, I'd like everyone who did control samples to send me the number correct guesses they made (i don't really need the actual results).  Then we calculate a mean+standard deviation and see if BH's result is significantly above the avg.  If a few of you also want send me the number of correctly guessed numbers then  I can do a similar test on that result.

thanks,

edd

LMNO

Morning, everyone.

I'm working up a spreadsheet now.  I've got predictions from a coin flip, Rat, TBS, TI, and Babylon. I've also shown what would happen if you guessed "All High" or "All Low".

On top of that, I've shown the results if we take out all "guaranteed hits" (3&4).

I'll post the spreadsheet when I'm done with it.


I've also come up with a more satisfying way of getting a 2/3 chance, but we'll get to that later.

LMNO

Ok, here are the results:

First, 100 rolls of a 6-sided die:

4
4
3
2
2
5
4
5
2
3
4
2
5
3
2
6
2
6
5
5
3
5
6
3
4
2
2
4
4
1
3
4
2
4
2
4
2
5
1
4
6
1
4
6
2
1
6
2
1
1
4
5
1
6
3
3
4
2
1
4
2
5
3
6
5
5
3
3
4
3
4
2
4
4
4
3
2
6
3
1
4
6
3
5
2
1
6
4
1
3
3
4
1
4
1
3
4
5
6
3


Something to note: 3 and 4 came up 45 times, which will raise everyone's score higher.  So, I've calculated the percentages both with and without the "automatic hits" of 3 and 4.

For example:

If I simply guess "All High", I would be right 69% of the time, but only 43% for the ones that mattered.
If I guess "All Low", I'm right 76%, with a 56% chance of being right if we discount the 3s and 4s.

Onward!

ALL HIGH:     69% / 43%
ALL LOW:     76% / 56%
COIN FLIP:    70% / 45%
RATATOSK:   72% / 49%
TBS:            73% / 51%
THURNEZ:     78% / 60%

And now...

For Babylon's "straight" picks (that is, predicting the actual number): 21% accuracy.
For picking High or Low: 77%
When excluding the Guarantees: 58%



Conclusions: the Hypothesis was that when given a 2/3 chance of being correct, Babylon would guess 25% better than random, which is [25 x .66 = 16.5] [16.5 + 66 = 82.5] 82.5% correct.

Compared to the four other random picks (if you guys want, you can explain how you randomized your picks), it doesn't look like Babylon was significantly better than anyone else (if you notice, Thurnez beat him); in fact, someone who simply chose all low without guessing at all was only a one-guess difference from Babylon.

And, for my prediction (as told to Rat and Nigel): I predicted you'd do better than statistically average, but you wouldn't reach 82% correct.


So, am I psychic?



Anyway, if you want to do this again, I have a better method for 66% using tarot cards.  Let me know.




PS - if you want to check my work (please do - I probably fucked something up), the spreadsheet is here: http://ifile.it/zojq5a3

Cain


LMNO


Cramulus


Bebek Sincap Ratatosk

My numbers were created by taking a random string of data as binary 1's and 0's, then I turned the first 100 bits into Low for 0 and High for 1.

However, as LMNO points out this wasn't a test that really met the 2/3'rds rule, so I'd like to see his next idea :)
- I don't see race. I just see cars going around in a circle.

"Back in my day, crazy meant something. Now everyone is crazy" - Charlie Manson

LMNO

We should only keep going with this if we want to stack randomness up next to Magick.  If someone wants to try magical divination, and really believe in it, we'll try again.  If it's just for shits and giggles, there's no real point.

So far, Babylon is the only one willing to put his money where his mouth is.  Anyone else want to give it a go?


Cain

I'd offer to use the process BH used to get his numbers and repeat the experiment, only my access is decidedly shifty right now.

As an aside, I've been reading Disinfo's Book of Lies again.  There is some quite useful material in there for any budding fiction or creative writer.