[geeks] More on global warming

[Phil Stracchino]

Patrick Giagnocavo wrote:
> On Dec 21, 2007, at 7:52 PM, Phil Stracchino wrote:
>> However, I venture to posit that merely the fact that TeVeS is unproven,
>> leaving room for our model of gravity to be flawed, does not mean one
>> can blithely step off a cliff and not expect to hit the bottom.
> 
> Actually you have this line of argument completely backwards.
> 
> The model of gravity that we all use in our daily lives, does in  
> fact, work, for most common cases.
> 
> We have a mostly-accurate model of gravity's effects, and are aware  
> of the limitations as things get either very very large  (things as  
> big as the Sun or larger) or very very small.

That was sort of my point.  "Mostly accurate" is "partly inaccurate", or
at least "partly in doubt".  Nevertheless, we don't throw out
predictions based on that model because we don't know whether it's a
complete model.

> It is not at all clear that we understand the impact of CO2 and  
> methane, etc.; or the impact of the Gulf Stream and what would happen  
> if it moved 100 miles east  (on average, it does move around a bit  
> year to year) of where it usually is.
> 
> Let me ask you this:  is it better to spend $100:
> 
> reducing my CO2 emissions
> OR
> reducing methane output (eat less meat, more veggies = less cows grown)
> OR
> reducing my nitrous oxide output (also considered bad for the  
> environment)
> 
> There is no way that you could answer that question, given the data  
> and models we have now.

Correct, we don't know which would be the most beneficial.  We appear to
have, at the very least, strong circumstantial evidence that we should
be trying to achieve at least one, if not all.  So:  Until we know for
sure, which of those can you do with relatively minimal adverse impact,
*just in case*?

>> In general, when a model predicts a probable risk, it's far better to
>> take precautions and develop a plan to deal with it, only to find  
>> it was
>> a false alarm, than to dismiss the problem as an artifact of the model,
>> and then ten years later be told "Hey, guess what, we've now confirmed
>> our projections, but it's too late to do anything now."
> 
> The models are themselves unreliable, that is the issue being discussed.

The problem is that the fact that they are unreliable does not mean they
are wrong, nor that they can safely be ignored.  There are two possible
hypotheses and courses of action, each with its own error cost.

Hypothesis 1:
Assume they are correct.  Act accordingly (try to reduce our greenhouse
gas emissions as much as we can) unless and until they are proven wrong.

Cost of error:
We spend a bunch of money, maybe alter some industries, for nothing.  On
the other hand, maybe we end up with better and cleaner tech out of it,
and perhaps have a substantial renewable-energy infrastructure already
in place by the time fossil fuel reserves finally crash.

Hypothesis 2:
Assume they are false alarms.  Act accordingly (disregard them) unless
and until they are proven correct.

Cost of error:
By the time we have irrefutable proof, we may have passed a tipping
point and massive global climate change may be irreversable.  That may
in fact *be* our proof: irrefutable proof may come in the form of, for
example, flooding of low-lying regions and the loss of many coastal
cities, or in the loss of the Greenland ice cap leading to shutdown of
the Atlantic thermohaline circulation and causing ice-age conditions in
northern Europe.


The principle of minimum cost of error suggests that the first is the
more prudent course of action.  Until we *know*, dismissing the warning
because the models are unreliable, when the possible error cost is so
large, is imprudent at best.  This is especially true because we do not
know if we will *ever* have a complete, or wholly accurate, climate model.

Many of the other models of various processes that we use to predict
risks are also incomplete.  Nevertheless, when a model predicts a
possible risk, we don't ignore it because it may be a false alarm.  We
do our best to confirm or refute the prediction, and we develop
contingency plans to fix them, and in cases where the potential error
cost is high -- for instance, "this system on this type of airliner may
have a failure mode that could cause it to crash" -- we will typically
take precautions such as grounding that type of airliner until we have
determined that they are *not* at risk, or until we have a preventive
measure in place.




-- 
  Phil Stracchino, CDK#2         ICBM: 43.5607, -71.355
  Renaissance Man, Unix ronin, Perl hacker, Free Stater
  alaric at caerllewys.net            alaric at metrocast.net
          It's not the years, it's the mileage.