Re: The Fourteen Characteristics of Fascism

On Thursday 15 May 2003 04:47 pm, James Richard Tyrer wrote:
> Andreas Pour wrote:
> > Food for thought: 
> > http://www.secularhumanism.org/library/fi/britt_23_2.htm
>
> First, it greatly annoys me that computer programmers fail at simple
> logic.

You're making a mistake here of assuming that simple logic is the more 
appropriate one to use here, and you illustrate why it's not yourself.


> Socrates' cat is NOT a dog.
>
> For those of you that don't know this:
>
> 	A dog has four paws
>
> 	Socrates' cat has four paws
>
> 	Therefore, Socrates' cat is a dog.

Now, that's all true. 
You can't conclude that Socrates' cat is a dog here; one can't reverse an 
implication.

However, given this sentence:
	Socrates pet has four paws.
The following can also be said:
	The probability of Socrates pet being a cat is a bit below 50% (and the same
	for it being a dog)

By choosing a binary/boolean logical system you are disallowing that type of 
reasoning, which is not absolute by any means, but useful for practical 
purposes. Take for instance the following:

'There is smoke coming out of a room in my house' 

Now, using the boolean reasoning, you would not call the firefighters, since 
you can't really know that it's a fire causing it, since there might be an 
another reason (one could even argue that you wouldn't be able to make a 
conclusion that there is a fire if you see a fire, since you might be 
hallucinating). So realistically, this type of reasoning isn't very good 
outside of math (where it's just perfect) and perhaps philosophy.

Probabilistic reasoning is better in this specific case-- Bayes's rule lets 
one 'reverse' implications to get a probabilistic answer that's often 'good 
enough' to make the sensible decision
It states:
P(something|evidence) = P(evidence|something)*P(something)/P(evidence)

Where P(a) = probability of a, or rather 'how likely we think that a is true', 
P(a|b) = probability of a given b.

In particular, this means that if there is a high probability of symptoms 
given something happens (i.e. P(smoke|fire)), and a low probability of 
symptoms in general (i.e. smoke regardless whether there is fire or not), 
and the event is not too uncommon (fires happen) that one can deduce a 
certainly value such as "I can be 90% certain that there is a fire going on", 
and act accordingly. 

Note that this also doesn't suffer from your example, since given 'Socrate's 
cat is has four paws' we already know that the probability of it being a dog 
is 0

</digression>



Kde-cafe mailing list - [email protected]
http://ofb.biz/lists/listinfo.cgi/kde-cafe

DISCLAIMER: The views expressed on this mailinglist are the personal
opinions of the author and do not represent OfB.biz: Open for Business, KDE or the author's employer.