Thursday, August 2, 2012

A good question ...

A good question is... a good reason to write something new in this blog after an usual long silence.
So, this is the question:


If you choose an answer to this question at random, what is the chance you will be correct ?  a) 25%  b) 50%  c) 60%  d) 25%



Nice ! The object of the question is the answer to question itself, i.e. we have an auto-referential question. For such things, we should always expect some puzzling results.
So, here's my answer : 0%, for the following reasons.
The possible answers are only a), b), c) and d), each evidently with a probability of 25%. Therefore the probability p(answer=x%) that the answer be a percentage x% will be:
p(answer=25%)=50%  (both answers a) and d) will do)
p(answer=50%)=25%  (only answer b) will do)
p(answer=60%)=60%  (only answer c) will do)
p(answer other than 25% or 50% or 60%)= 0% (no answer will do)
Hence: 
- if the correct answer was 25%, then the chance to pick it up would be 50% instead of 25%, i.e. 25% was not the correct answer ==> the chance to be correct is not 25%
- if the correct answer was 50% or 60%, then the chance to pick it up would be 25% instead of respectively 50% or 60%, i.e. 50% and 60% are not correct answers ==> the chance to be correct is neither 50% nor 60%
Therefore the chance to be correct must necessarily be 0%, i.e. the answer will always be wrong.


That question has some similarity with the auto-referential statement:


This statement is false



Can you say whether this statement is true or false ?


A final remark: in his (old and) wonderful book "Gödel, Escher and Bach" the author D.Hofstadter compares auto-referential statements to vinyl records (still in use at that time) that, when playing, destroy the grammophone (the latter being a logical systems with axioms, theorems etc.).