From: Dan Clemmensen (dgc@cox.rr.com)
Date: Sun Sep 23 2001 - 08:21:21 MDT
Miriam English wrote:
>
> On the subject of the power grid though, power stations are definitely a
> danger point. I have never understood why so much preference was given
> to centralised power sources instead of decentralised ones (e.g. solar
> cells). The more decentralised something is, the safer and more secure
> it becomes.
>
Miriam, centralization is driven by economics. For a great many things,
The economic benefits of centralization outweigh the economic costs.
When the economic environment changes, this cost ratio changes. 911
was an extreme example of a change in the (perceived) costs of
centralization. Manhattan is an example of physical centralization.
Trains and huge airports are much more centralized than the road system.
Huge power plants are much more centralized than small plants or
local generation. Networks with high-speed backbones are more
centralized than grids. City water is more centralized than wells.
In essence, you cannot make a universal qualitative argument in
favor of decentralized systems: you need to look at the economics,
and that means using your calculator or speadsheet.
If you wish to pay for redundancy, you first decide how much you
you want. Basically, you can get redundancy by dividing your
resource into N pieces and then adding an extra piece (N+1 redundancy.)
in general, the cost of your solution is the cost of the extra piece
plus the cost of the additional interconnection and control plus (or
minus) the "efficiency ratio" associated with building more smaller
units rather than fewer big ones.
For office buildings, the equations still work, but tragically we
cannot actually have backups for the workers, since each worker
is a unique human. Even in the economic sense, workers are not
interchangeable. However, we can still cut our failure risks using
the above redundancy model since surviving workers from other offices
can and do fill in as best they can.
I am a Systems Architect, and I've been one for more than 15 years.
My field is data communications switching, so I usually worry about
the number of processing elements in a switch or the redundancy of
a network topology. It is traumatic to apply my discipline to
911, but the analysis nevertheless stands.
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:10:56 MST