And since nobody has brought this up yet, my favourite example of the counter 
intuitive probability puzzle:
http://en.wikipedia.org/wiki/Monty_Hall_problem

Venky, the Second. 


On Saturday, 27 October 2012 at 3:17 PM, Landon Hurley wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
> 
> On 10/27/2012 04:49 AM, Deepa Mohan wrote:
> > I can TELL myself all this:
> > 
> > On Sat, Oct 27, 2012 at 2:15 PM, Landon Hurley <[email protected] 
> > (mailto:[email protected])> wrote:
> > > 
> > > 
> > > 
> > > The probability that you will or will not roll a 1 is .5. The
> > > probability that the roll will result in a 1 is 1 in 6, given the event
> > > that you have rolled a six sided die. You can reduce most events to will
> > > or will not happen. That isn't the true representation of that scenario
> > > though.
> > > 
> > > It can only be .5 if there are two equally probable events. How you are
> > > presenting them is not in terms of the probability of those outcomes,
> > > but in that given two discrete categories (1 or 2,3,4,5,6) the
> > > probability is .5 that you will land on 1.
> > > 
> > > The inverse of rolling a one is important as well. The probability of
> > > rolling a 2 must also be .5, according to your interpretation. But since
> > > all possible probabilities must sum to 1, .5(6)=3 (or .5 added for each
> > > of the six possible events) is a contradiction, not to mention
> > > impossible given that probability is a continuum from 0 to 1.
> > > 
> > > Given your terms:
> > > In terms of a lottery, there are n people who participate. My odds of
> > > winning must be .5, either I do win or I don't. However, what if there
> > > is only one person who participates in the lottery, and that person is
> > > me? Isn't my odds of not winning 0, since I would always win?
> > > 
> > > However in a system where probability is always calculated by the
> > > desired events divided by the total number of possible outcomes, this
> > > contradiction is explained, and empirically valid with your perceived
> > > paradox with conventional frequency based probability. Given one desired
> > > outcome (rolling a 1) and two potential outcomes (rolling or not rolling
> > > a 1) the expression becomes 1 in 2. Charles' example though derives from
> > > rolling a 1 as opposed to rolling a 1,2,3,4,5,6. With the same equation
> > > structure of desired out of total outcomes, you get 1/6. Here is where
> > > your interpretation breaks down though: what is the probability of
> > > rolling a red? Your view would have either red, or not rolling a red,
> > > thus 50-50. However, you can never roll a red. So the probability is
> > > accurately reflected as 1/0.
> > 
> > 
> > 
> > But it ultimately makes as much sense to my intuition as this:
> > 
> > 
> > > iQIcBAEBCgAGBQJQi58mAAoJEDeph/0fVJWsN84P/0GaRnK0CH8SR7IQsWoQcsx8
> > > 4t9kaO59i7l/vgw9PVc4AU7Vkoixj0Q3W/jiw7IYIgejCVWdzvPWJynUhd/U+gDF
> > > 3nMK+iTNOOUjZjEZRv8e6Oki/io2AHfRZRjP/ugNOkOspdo+H78r1fmXOs5yXlqh
> > > i++NP2DAXJ3k+BTc7043PrLvdIOtlrryGNPXG4qs8tvkvtC3v/Wjqq0k0d34RE3T
> > > O5AnymocsoBDm9pYAOuxRveXcphMb1zA0zE3zBQmHW9JDfvNqad5o+/QbLjjpwQw
> > > NWjLkR7PIOA/CUv9XTyVQwLw9LjHr1m+y68ZNMdsgQR4VF1CK0F4qL7QPKLJEIJs
> > > 4GPGlPqg1XUK54PkQqQDSQYSj/rosQqBVUVUxlQPALxIHpfxLWqtG3T67j2Rk8Oo
> > > qObTNoBksJTYu81Ii1VX3Pvt4bjogZgUcH6HJBPc0aIxOnHK4LGQ6p55xuSKLMSy
> > > vfcf6NyGiyoQcziUHMGVdMjdKoy9PPGvRLsov3ezNxvePw/cunMwVnJacBZgf6+/
> > > mU9FLmTpeH18phe+QorheMwA/M9nnS13C/48fGoRFqKt19x+fVOVrtUiubgkAlS7
> > > sx8J9WL8FbT0D5HmrRB6DRavYxQIO4PCsGODBqag22GYBp6vR6GU8/dJe4hz95Yi
> > > PbIz9P4Wml4NillJLvwk
> > 
> 
> 
> 
> 
> Probably isn't intuitive, in fact it's inherently counter intuitive.
> Unless you bother to understand the basic frequency mathematics, there
> isn't really any point in bothering at all in the long run. In the end,
> intuition is what we feel is simplest, and that's Occam's razor. That's
> fine, except that it's a rule of thumb not an absolute. It doesn't prove
> anything. Your argument is that flipping that coin will have a
> probability of .5 resulting in heads. What's the probability of coming
> up with the result of HHHHHHHHHHTHHTH? .5**15, definitely not .5, right?
> And that's because the probability of .5 for coins is a continuous
> string of states, not just independent. Very few things are independent
> in probability, and they are structured on a discrete set of foundations:
> 
> 1. That the event in question happened, or else is included in the
> outcomes. This is where you have an intuitive problem, because you say
> that 1 will either happen or not happen. However, what about 2,3,4,5,6?
> They are all equally NOT 1, but they have their own discrete existences
> outside of ~1, right? They all have to exist as potential outcomes, and
> the probability that one of them occurs has to be 1, given that you
> rolled in the first place. Your interpretation has an inherent
> contradiction, since you can't have 6 equal probabilities of .5!
> Something cannot be 300% likely to happen.
> 
> 
> Obviously, if the event of rolling is even in question, then there are 7
> potential outcomes, NULL,1,2,3,4,5,6.
> 
> Number of desired outcomes divided by number of potential outcomes is
> the primitive of probability. To be honest I don't see a simpler way to
> break it down, maybe someone else on the list has a better method.
> 
> However, I am interested in how you ameliorate the paradox between the
> probability of rolling a one is .5, i.e. half of the N times you role a
> die, you will come up with a 1, and the empirical fact that a you role a
> large N, you come up with a number that approximates 16.66 out of a 100
> occurrences of 1.
> 
> - -- 
> Violence is the last refuge of the incompetent.
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.11 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://www.enigmail.net/
> 
> iQIcBAEBCgAGBQJQi62eAAoJEDeph/0fVJWsUWsQAIueaVPqLw4BVcSSh4oX7k2M
> hzYZHjx7KNAmQJ9JWu1tj/+1VOgEywERFmPMMpEgTyel4pOSR6hIzUeHLBmC7uqy
> 1CRgbIlEcB9a1W5jGUV5qDGxB005f/lRjBgWqJEQkYVrxdhzae5+pxJghJDDbn/Z
> ypZFqJytvmu1OlRchfNDb21672sVQOlTaDuG5/Nyq8mCQKZgFeBIJEhnBQA49Wfa
> 0AxtCTbeSxOoDGWkElpdKyzjbOmwmTGYAGedUMfYQg+aI7MrphjgBbL9QZKDELZr
> AXS3+isQRpgHINA+zKnsTPstnaTHXtwGDbm4OMKopLqEZ9HfFVG1fnxCX0NKYvSN
> BFQXOAWSLoDzl9RY+o0kyK1nb6EJ6/qKp/sy7dA8uJAfzphUsfEkT3lSi4xuyGVU
> fLnOpDwoELg9uOTqMAbi9aaWeGvSpCOw8OdmA1xFjYfMqyM4yQz4IcLKer8kY3+M
> eRbBZ5pkMweBAO5ReFfIT/bsoJenZ29qzk+l7nRUaAdhAiVkF/dkjWcA8nZniHAB
> mQAkOWEiR5IXCwKW+Lwd7fNUjwItqvyO3afxumyvqBWMPUufZpZ/VCUchJ1E9g4h
> cZnh4tztt/FO+g+ATdKYdl9vEg1RZP7piCMqX9LupPPs19MjVwoTJRidc6cGty6e
> KU99+tv7d3bU9fLfCxxv
> =QuaT
> -----END PGP SIGNATURE-----




Reply via email to