Scientists Don’t Really Know it All

As a scientist, I’m gratified by our current high favor in the court of public opinion.  I must admit, though, sometimes our omniscience is overrated.  In the first place, scientists tend to have narrow areas of specialization.  As a result, they are not remarkably superior to other mortals in seeing the big picture.  In the second, we simply lack the knowledge and/or adequate data in some areas to justify positive opinions one way or the other.  Finally, scientists are human beings, subject to human needs.  It is not out of the question that their research results may occasionally be influenced by such mundane considerations as the desire to eat.

To illustrate the potential liabilities of narrow specialization, let us consider the issue of nuclear power, with which I have some passing familiarity.  If it’s a question of solving the neutron transport equation for a particular core design, a scientist is definitely the guy you want to talk to.  However, if it’s a question of deciding whether the nation should prefer nuclear power to the various competing sources of energy, it ain’t necessarily so.  To address such overriding issues, one must be well informed not just in a narrow technical area, but also in a host of environmental, political, economic, other matters of relevance.  I have seen anti-nuclear advocates from non-governmental organizations (NGOs) defeat nuclear engineers hands down in debates over the merits of nuclear power, because they were better informed on such matters.  That doesn’t necessarily imply the anti-nukers were right.  Rather, it illustrates the fact that narrow expertise is not adequate for deciding every issue.  I’ve known scientists who were brilliant within their own technical bailiwick, but shockingly ignorant if they ventured outside it.

Then there’s the matter of technical uncertainty.  Here, one might cite global warming as exhibit A.  It happens that my personal opinion on the matter is that carbon dioxide and other greenhouse gases do inhibit the re-radiation of solar energy back into space, and, as a result, we are likely to see significant increases in global temperature and sea levels over the next century.  However, I have a problem with those who claim they know with certainty exactly what the effects of global warming on our climate will be and how long it will take before it’s “too late” to do anything about it.  They can heap scientific opinion on scientific opinion ad nauseum.  It doesn’t matter.  Given the current state of the art, we cannot predict with certainly what will happen one way or another. 

In order to accurately predict the future behavior of a system, it is necessary to have means of accurately measuring all the data relevant to the response of that system.  In the case of climate modeling, the necessary data, ideally from many billions of data points, is inadequate.  The data we do have is subject to significant measurement uncertainties, or “noise.”  Furthermore, we’re not even sure what data we need, assuming it were even available, to accurately solve the problem.  Finally, even if the necessary data were forthcoming, no perfect mathematical models would be available to use it.  With the biggest and fastest computers that exist now or in the foreseeable future, only dominant or critical climate effects could be modeled.  Such models are prone to leave out “minor” effects that may actually turn out to have a critical effect on the accuracy of the outcome.  Even the effects that are included must be modeled with approximations that are never perfect. 

Climate modeling today is not and cannot be based on any deterministic model.  Significant uncertainty is built in to the current ensemble and Monte Carlo forecasting models.  Scientists know they can’t even be sure they have accurate knowledge of the starting conditions to plug into their models.  As a result, they often just come up with an “ensemble” of plausible ones, and run them all through the model.  Then they use interpolation and approximation methods based on all the outcomes to decide which one is “best.”  In other words, while it would certainly behoove us to take what effective steps we can to avoid potentially harmful climate changes, we have no way of knowing “for sure” what those climate changes will be.  Our mathematical models are even less capable of predicting exactly what the impact will be of the steps we might take to limit greenhouse gas emissions.  It is inadvisable to mandate extremely expensive but highly visible measures to limit global warming if they are unlikely to have any significant impact on the problem one way or another.

A third weakness of “scientific expertise” is the human tendency of scientists to tell customers what they want to hear.  There is intense competition for research grants and awards.  There is also a wide and probably accurate perception among scientists that the sponsors of the limited available research funds are more interested in positive and striking findings than in null results, and are, therefore, more likely to reward those who produce positive results with more funding.  I leave the effects this might have on the result, for example, of studies of global climate change to the imagination of the reader.

We scientists can be proud of our contributions to the welfare of society.  However, we have our limitations, and we need to keep them in mind.  Do not even the lawyers the same?

Leave a Reply