Pursuing Prediction

George Rebane

The latest global catastrophes – natural and manmade – again make us re-examine our ability to predict future events and what we are doing to improve those skills.  A current discussion of this is going on at Edge – The Third Culture.  Edge is one of those exclusive foundation think tanks for intellectuals from diverse backgrounds who are invited to come together and “arrive at the edge of the world’s knowledge, seek out the most complex and sophisticated minds, put them in a room together, and have them ask each other the questions they are asking themselves.”  There are some impressive names in their list of contributors.

So a group of these folks were asked to consider prediction, and submit short essays summarizing their thoughts on the topic.  As a lifelong worker in this vineyard, I was interested to see who would be saying what about prediction.  My first amazement came when looking at the contributing ‘whos’.  The mix of people was eclectic and politically correct.  It included fine minds practicing physical oceanography, magazine editing, cultural anthropology, psychology, media analysis, science writing, sociology, chaos theory, and, yes, robotics.

One worthy, with science credentials, even lamented that –

That inability to predict may come from some deficiency in our knowledge, or it may be the result of a great complexity inherent in the phenomenon (for example, we may not have high-enough-resolution data to represent it, or the process may have a chaotic component that keeps us from determining exactly when it will occur). We are then left only with probabilities.

To a systems scientist (of whom there were none) the remarkable thing about such a statement is the last sentence that seems to assert that there are fields of human knowledge in which we are NOT ‘left with probabilities’.  The fact of the matter is that we are always and only left with probabilities that are the invariable meta-data which comes with every piece of knowledge that we may ever claim.  It is the ascription of these probabilities that guides our use of the claimed pieces of knowledge.  And as most intelligent people know, these probabilities range from one (expressing certitude) to near zero (expressing the rare or a notch above the impossible).

When we describe the knowledge of experienced and/or repeatable events, there is a mature science which allows us to assess and assign a probability to such knowledge.  When we wish to describe the likelihood of a future event (i.e. predict) we have long had means to compute its probability, which then is called a (measure of) belief as held by sentient creatures.  The primary means of doing that is the business of a rich and much used field called Bayesian estimation that correctly combines existing knowledge with new observations to update what is known.  The entire collection of the cited essays contains neither mention of the Reverend Bayes, nor the considerable science to which his theorem has given birth since the 18th century.

We even have the lesser known contribution of J. Richard Gott who discovered a simple yet powerful means of estimating the probability of a process terminating in the next interval of time given knowledge only of how long the process has been going on.  (I have expanded his theory to arbitrary future intervals and generalized the terminating event to consist of more than one branch.)  And these essays on prediction also contained no mention of Gott.

The conclusion that I drew (actually reinforced) from these essays and the self-prestigious Edge is that their work products are too often spotty and gratuitously auto-aggrandizing.  Their well assembled pronouncements, especially on matters of science and technology, leave the layman in a quandary on the credibility of the offered assessments and opinions.  And much of that comes from their desire to include in their mix of ‘experts’ people who have low or no bona fides about the topic at hand.  Today, the reading public is constantly running into the cited  pronouncements of such foundations, institutions, and even academies.  Caveat emptor!

(BTW, you may search RR for more on ‘Bayes’ and ‘Gott’)

Comments

[Russ Steele]

George you wrote:

Their well assembled pronouncements, especially on matters of science and technology, leave the layman in a quandary on the credibility of the offered assessments and opinions.

ooo

Today, the reading public is constantly running into the cited  pronouncements of such foundations, institutions, and even academies.

I am reminded of the whole global warming issue, with computer models being used predict the temperatures in 2050 and 2100, when weather models cannot predict the weather reliably more than five days. Yet, we have "foundations, institutions, and even academies" telling us we are doomed if we do not control the emissions of CO2.

Watts Up With That has an interesting post on the communications of climate risk, but it should apply to all science that attempts of predict the future.

A major challenge facing climate scientists is explaining to non-specialists the risks and uncertainties surrounding potential” climate change, says a new Perspectives piece published today in the science journal Nature Climate Change.

The article attempts to identify communications strategies needed to improve layman understanding of climate science.

“Few citizens or political leaders understand the underlying science well enough to evaluate climate-related proposals and controversies,” the authors write, at first appearing to support the idea of specialized knowledge–that only climate scientists can understand climate research.
But, author Baruch Fischhoff quickly dispels the notion. “The goal of science communication should be to help people understand the state of the science,” he says, “relevant to the decisions that they face in their private and public lives.”

Fischhoff, a social and decision scientist at Carnegie Mellon University in Pittsburgh and Nick Pidgeon, an environmental psychologist at Cardiff University in the United Kingdom wrote the article together, titled, “The role of social and decision sciences in communicating uncertain climate risks.”

As Anthony points out, it may not be the message that is the problem in this case, but the science. Interesting read.

[George Rebane]

Exactly so Russ. And in the case of global warming, as you may recall, my contribution to a "reasoned dialogue" was (is still?) rejected due to the firmly held belief that ALL science (i.e. the realworld) is simple enough to be explained simply. Well, it turns out that it isn't, as was found out by those who kept presenting thousands of counter arguments in the form of simple one-dimensional temperature charts, etc.

For that reason, it has become more and more clear that science and systems must be understood in their closest native formats (which includes the holistic consideration of their multi-dimensionality), or they are not understood at all.

When it comes to prediction models (global warming or catastrophes), no one will make progress unless they understand the meaning and relationship of the nested venn diagrams I presented. And also, no one will be able to provide useful predictions of a tsunami unless they understand the works of Bayes, Gott, and their intellectual descendants.

[D. King]

"The conclusion that I drew (actually reinforced) from these essays and the self-prestigious Edge is that their work products are too often spotty and gratuitously auto-aggrandizing. Their well assembled pronouncements, especially on matters of science and technology, leave the layman in a quandary on the credibility of the offered assessments and opinions.”

This seems to be pervasive.

Like this.

NASA finds arsenic-based life

http://www.youtube.com/watch?v=z1NHQKCyryI&feature=related

NASA finds arsenic-based life…debunked.

http://www.helium.com/items/2040928-nasa-study-of-arsenic-based-life-may-have-fatal-flaws

http://hotair.com/archives/2010/12/07/scientists-nasas-alleged-discovery-of-arsenic-based-life-is-crap/