[FoRK] Human behavior is 93% predictable

Sean Conner sean at conman.org
Tue Mar 2 22:37:41 PST 2010


It was thus said that the Great Ken Ganshirt @ Yahoo once stated:
> --- On Tue, 3/2/10, Jebadiah Moore <jebdm at jebdm.net> wrote:
> 
> > That said, perhaps the actual number for the 10% can be reduced to
> > acceptably low rates.  93% accuracy (which is obviously a longshot) is
> > dodgeable.  If you could build a 99.999% reliable model, that's still a lot
> > of errors given scale, but perhaps it would be prohibitively difficult to
> > actually engineer an error and they would all be random. 
> 
> In practical terms it's likely to also be hugely difficult to get from 93%
> to 99.999%. Once you start trying to get 9s, especially to the right of
> the decimal point, it becomes much more difficult than the arithmetic
> suggests. Just ask any phone company or any business that has genuinely
> embraced six sigma.

  Just to put some figures to this, out of a 365 day year (non-leap year, no
leap seconds), a hosting provider that says they provide

99% uptime	means they can be down for	3d, 15h, 36m per year
99.9% uptime					8h  45m  36s
99.99%   					52m 33s
99.999%						 5m 15s
99.9999% (fabled six-sigma)			31s

  -spc (And to think, people still run Windows servers ... [1])

[1]	Cheap shot, I know ... 





More information about the FoRK mailing list