What is wrong with this. (files)

[jab3]

jab3 said:


pete - that jibes pretty well with my experience, too, although I'd have
put the probability somewhat higher.


If you truly believe that two mutually exclusive possibilities are
necessarily of equal probability, I offer you the following bet:

Given suitable precautions to prevent cheating, you will put down $5000
and I will put down $10000 - so there's $15000 on the table - and then a
trusted third party will put a hundred fair coins (one side heads, the
other side tails) into a sack, shake 'em around a bit, and then spill them
onto a tabletop. Either they will all come out heads, or they won't, so
that's 50-50, right? So - if they all come down heads, you get to keep the
$15000, despite only a $5000 stake. That's good odds for you. If the other
50% chance happens, though, I get the $15000 instead.

Deal?

Well, I don't think my statement works given real probability analysis. (It
was late and the 50% chance comment just struck me as funny) However, I
also don't think your bet is strictly analogous to what I said. I didn't
say the whole program is either all right or all wrong. I said any given
line theoretically has a 50% chance of being right or wrong. Your bet, were
it analogous, would be that any given coin has a 50% chance of being either
heads or tails. But I would imagine the actual probability of any given
line of code being right or wrong is based on the person doing the coding,
the language, the complexity of the program, the emotional state of the
person, etc.

 

[pete]

the 50% chance comment just struck me as funny

http://www.google.com/

Results 1 - 10 of about 76,500 for "it's funny because it's true".

 

[Richard Heathfield]

jab3 said:

Well, I don't think my statement works given real probability analysis.

No, it doesn't.

(It
was late and the 50% chance comment just struck me as funny) However, I
also don't think your bet is strictly analogous to what I said. I didn't
say the whole program is either all right or all wrong. I said any given
line theoretically has a 50% chance of being right or wrong.

But in fact, if we know there is one bug in an N-line program, then (if
that's all the information we have to go on) each line has a 1/N
probability of containing that bug. 1/N is only 50% if it's a two-line
program.