Re: Forrester Research foresees death of firewalls

[David LeBlanc <dleblanc () mindspring com>]

Warning - almost totally off-topic - 

At 04:36 PM 6/21/99 +0000, Bennett Todd wrote:
> 1999-06-17-04:09:26 Robert Graham:
> > This reminds me of a lecture I once attended about bridge building. The
> > idea is that bridges are always overbuilt in a very conservative
> > manner. 
 
> > As time goes on, builders gain confidence (because none of the bridges
> > are failing) and take more risks (fewer materials, longer spans, etc.)
> > until POOF a bridge fails, then the go back into conservative mode.

> This analogy in turn reminds me of one of my favourite engineering quotes
> (wish I knew just where it came from): It's easy to build a bridge that
> doesn't fall down; what takes skill is building a bridge that just _barely_
> doesn't fall down.

I have degrees in both Aerospace engineering and a branch of Civil
engineering.  You normally build bridges with a safety factor of 10 over
the load you expect to be the maximum that the structure will handle.  This
is a big load of crap.  The real reason that we do this is because although
we understand the _statics_ of the bridge very well, we have a very poor
understanding of the _dynamics_ of the bridge - that is that we can predict
the stresses quite well if trucks filled with rocks magically appear on the
bridge.  However, if those same trucks are driving over the bridge and
perhaps hit a bump or two, and it is worse if there is wind going under the
bridge, all bets are off.  Since we really don't have the foggiest idea
what we're really doing, and it doesn't usually hurt us all that badly to
overbuild the bridge, we just throw a lot of extra concrete and steel at it
and hope for the best.

Aircraft, however, must be built _efficiently_ - any extra pound of
structure is one less pound of cargo.  Aircraft are normally built to a
safety factor of 1.1.  This is because we spend a lot of time to understand
the dynamics of an aircraft structure (and if you think a airplane wing is
bad, helicopter blades are a real mess - 3 vibrational degrees of freedom),
and a lot of time testing.

You run into this all over engineering - any time you start seeing safety
factors > 2, you can just about bet that the design equations aren't really
a decent 1st order approximation.

Just to give everyone that warm and comfy feeling next time you're on a
bridge...


David LeBlanc
dleblanc () mindspring com