[Phys-l] an all-too-predictable blackout

[John Denker <jsd@av8n.com>]

Sometimes it's not good when predictions come true.

Back in late 1994 or early 1995 I wrote a classified report that
mentioned, among other things, the possibility of a poison-gas
attack against the subway system of a major city.  People thought
it was a stupid report.  A few weeks later the prediction came true.
  http://en.wikipedia.org/wiki/Sarin_gas_attack_on_the_Tokyo_subway

That created a furore.  There were some people who wanted to burn 
me for a witch.  There were some other people who suddenly wanted
to know what else I had to say.

=============

Back in August 2003 there was a large regional power blackout.
  http://en.wikipedia.org/wiki/Northeast_Blackout_of_2003

That was followed by congressional hearings.  There was some
legislation.  Also the industry made elaborate promises about
self-regulation so as to ensure that no such event would ever
happen again.

I was not impressed.  In 2005 I quoted former Secretary of Energy 
Bill Richardson, who said the United States was “a major superpower 
with a third-world electrical grid”.

I went on to make the following prediction (in 2005):

> If I had to guess, I’d say there’s a high probability of another huge
> blackout within the next ten years. I’d estimate a 10% chance per
> year for each of the next several years.

This was included in a discussion of the physics of power grids:
  http://www.av8n.com/physics/power-grid.htm

A little more than half the time-span of that prediction had already
passed, so I was starting to wonder about it.

=========

My opinion about predictions has changed since then.  I now realize
that science should be more about scenarios and less about predictions.
For what it's worth, by way of terminology, in my book a hypothesis 
is not a prediction;  it is just a scenario, and the rule is to
always consider *all* of the plausible hypotheses, i.e. *all* of 
the plausible scenarios.  See e.g.:
  http://www.av8n.com/physics/scientific-methods.htm
especially:
  http://www.av8n.com/physics/scientific-methods.htm#sec-hypothesis
and then there is the infamous poster:
  http://www.av8n.com/physics/scientific-methods.htm#sec-poster

Whatever you call it, treating it as a prediction makes it sound 
like taking sides.  IMHO scientists don't need to take sides, 
and usually shouldn't take sides.  When I say we should consider 
this-or-that scenario, it does not mean I want the scenario to 
come to pass.  It may be that by carefully considering an 
unpleasant scenario we can invent a way to make sure it doesn't 
come to pass.

  If you think this is more engineering than science, that's
  OK.  Whatever.  I've never cared about such distinctions.

Let's be clear:  I'm not happy that my blackout prediction came 
true.  I'm annoyed that the industry failed to consider all the 
plausible scenarios.  I'm annoyed that the regulators weren't
effective.

One thing that made this outcome all-too-predictable was the
simple fact that the industry has never had sufficient incentive 
to fix the problems.  A blackout is treeemendously expensive, 
but the cost is not borne by the utilities;  it is borne by 
each and every customer.  If the utilities we required to post 
a performance bond, indemnifying the customers for the cost
of a blackout, reliability would quickly go way up.  In the
absence of such a feedback mechanism, industry promises about
self-regulation should not be taken seriously.

Similar remarks apply to nuclear power-plant engineering, which
is not the same as power-grid engineering.  Indeed the same
remarks apply to a wide variety of other industries.

On the other side of the same coin, there are some industries 
where the feedback is very strong, resulting in high levels 
of safety.  I'm thinking in particular of the airline industry, 
which has figured out that crashes are very expensive, so they 
go to great lengths to prevent them.  Taking an airliner cross
country is vastly safer than driving cross country.  By and
large airlines welcome regulation, because they are already 
doing far more than is required by regulation, and they
want the /other guy/ to be regulated, because any crash is
bad for the entire industry.