Total Pageviews

Sunday, July 12, 2020

The Decline of Science

Physicist Richard Feynman said 
“It is whether or not the theory 
gives predictions that agree 
with experiment. 

It is not a question of whether 
a theory is philosophically delightful, 
or easy to understand, or
 perfectly reasonable from 
the point of view of common sense.”
"QED: The Strange Theory of Light and Matter"
(Princeton: Princeton University Press, 1985), 

Scientists are grappling 
with the problem 
of model uncertainty.

Modern scientific knowledge 
depends upon mathematically 
expressed theories that 
could be validated by prediction 
and observation (measurements). 



A valid scientific theory requires:
 (1)
 A mathematical model 
expressing the theory.

 (2) 
Precise relationships specified 
between the theory and 
measurements of corresponding 
physical events. 

(3) 
Validating data: 
Predictions derived from 
he theory compared with 
measurements of physical events.

 (4) 
A statistical analysis 
that supports the theory
( the predictions match 
the physical measurements )


The meaning 
of a scientific theory 
depends on the 
connection between 
the mathematics 
and experience.

That connection occurs 
via the process 
of validation.

Science requires 
that we be able to make 
accurate predictions 
of future events, based 
on mathematical models.



Thousands of daily 
decisions are made 
by politicians, managers, 
and bureaucrats that are 
directly or indirectly 
related to science. 

A leader must understand
the validation process: 
What predictions 
are derived 
from the theory, 
and to what extent 
have those predictions 
agreed with observations?

Good decision-making 
requires a fundamental 
basic knowledge of science.



The climate is 
always changing. 

The scientific question 
is whether climate change 
can be predicted. 

The predictive skill 
of a model is usually 
measured by comparing 
the predicted outcome 
with the observed one. 

But climate projections
for 50 or 100 years 
can't be validated directly 
through observed changes, 
without waiting for 
many decades. 

So any confidence 
in climate models 
must be coming 
from faith in the
medels, not 
validation .

Repeated wrong 
climate predictions,
however, are always
anti-science.

We've got that !

Data mining may look 
like real science, but
what is missing is the 
connection between 
the mathematics and
verifiable accurate
predictions.

We don't have that !