Bob Lunn
08/15/97 11:10 AM


>> If the estimtated probability that what you will
>> learn from the investigation of the clock radio is about 5% that it is
>> built in a way you appreciate it is 2% likely to cause a fire, which
>> will cause predicted damages of an average 2000 dollars, then the threat
>> has a value of 2 dollars (0.02*0.05*2000), which means you overreacted
>> by 625% (12.5 / 2) on this threat.
>
> I could look at it another way.  By spending 12.50 and saving 2000 in
> damages, I have saved a lot of money.  2000-12.50 = 1987.5.

     You _could_ look at it this way, but you would be wrong.  You are
     saying that I should spend all my salary this week on lottery tickets,
     because I will make a lot of money.

> But the 2000 number is hypothetical.  The fire could cause enough damage
> to just destroy the clock, OR burn down a house and kill a family.  A lot
> of variation in that Harm portion of the equation.  What Harm # would you
> apply in this case?

     The number 2000 is an hypothesised AVERAGE.  It therefore incorporates
     the variation to which you refer, and the estimated threat is valid.
     In any case, the words 'boundary analysis' come to mind.

     You cannot replace quantification and analysis of risk and harm with
     seat-of-the-pants guesswork.

___Bob