RE: Who knows?

[Miles Sabin <MSabin@interx.com>]

Michael Champion wrote,
> As David Megginson said yesterday, the best project managers 
> are 90% right about predictions a month in advance ... which 
> means that they're about 50% right about predictions six months 
> in advance ... which means that even the wisest captains don't 
> have a clue how to "wisely" choose a route for a journey that 
> will take more than a year.

Well, if Davids somewhat mechanistic error model were true then
actually you could do extremely well on binary choices projected
sufficiently far into the future.

The caricature model is 90% accuracy after a month, hence .9^n
after n months. So here's a strategy: For any binary decision you 
have to make roughly six months in advance, flip a coin; for any 
binary choice you have to make more than six months in advance, 
make your best effort to work out the Right Thing to do, then do 
the opposite.

For binary choices one year in advance this gets you a 75% 
expected success rate. For binary choices two years in advance it 
gets you back to your original 90% expected success rate ... and
it only gets better after that ;-)

And the moral of my story? Beware crudely applied stats, 
particularly when you're applying them to people.

Cheers,


Miles

-- 
Miles Sabin                               InterX
Internet Systems Architect                5/6 Glenthorne Mews
+44 (0)20 8817 4030                       London, W6 0LJ, England
msabin@interx.com                         http://www.interx.com/