On the varieties of prediction

In some of the comments below I engaged in a discussion about the power of prediction, the necessity of skepticism, and so on.  In the format of a weblog the full overgrown shape of one’s thoughts can be somewhat muddled.  For example, I evinced a skepticism of predictions of the future from rational a priori assumptions, and therefore a particular prejudice toward custom & tradition.  But across the set of species of predictions obviously various degrees of skepticism are warranted.  After all, I think it would be ridiculous to be skeptical of an astrophysicist who made a prediction as to the arc of celestial orbits.  The record on these things is rather different in terms of precision & accuracy from predictions of, as a contrast, macroeconomic performance.  It stands to reason that the cudgel of skepticism should be selectively applied to different classes of rational systems and the projections thereof.  On the great chain of predictive being, I would rank it like so:

Formal mathematics – High precision

Physics – Great precision

Chemistry – Good precision

Biology – Modest precision

Economics – Fair precision

Sociology – No precision

To be fair, these are broad-brush generalizations (some economists claim are extremely precise in their formal systems because of the problems with testing their models easily when compared to experimental physicists). But, my point is that whereas I believe that experience tells us that one should put great store in the predictions of physicists and the engineering sciences which derive from physical precepts, much more skepticism is warranted when it comes to the wisdom of economists. This is not to say economics is not a worthy or important field, but the nature of the domain of study is by its nature less tractable than particles or galaxies.  Formal methods clarify in the social sciences, but the simplifying assumptions naturally result in a large margin of error from expectations generated.  Many of the biological sciences span the great middle lands between the precise power of physics and the conditional and caveat girded inferences generated by economic data sets.  On the one hand biophysical work can attain a relatively high level of deterministic precision when it comes to molecular pathways, but it concedes a great deal in generality.  Evolutionary work couched in the language of statistics & probability is far more general, spanning an enormous expanse of space & time, but the predictions are characterized by enormous error bars because of the unaccounted for variables and the random acts of historical contingency.

The human sciences have problems which statistical biological sciences do not, the special nature of humanity and the difficulties of controlled experiments.  During the fall the debate over whether the financial bailout was interfering with corrective market processes was peppered with assertions to the effect that we should not, can not, risk a test of the economic theory!  Of course in the natural sciences testing theories is the ultimate check on both flights of fancy and excessive skepticism, but the falsification process in economics can have very grave pubic policy implications.

From physics to economics you obviously are spanning a great range of disciplines, but the important point is that physics is much simpler than economics.  The problem of complexity is what generally warrants greater skepticism of formal models, projections and predictions in particular social domains.  Social systems may exhibit many interaction effects which mask general dynamics, or, the interaction effects may be so great that general inferences may be trivial in their utility.

And it is not simply in human societies where complexity reigns supreme.  Consider ecological and climatic systems. Here the modern Left has discovered the importance of contingent systems which may be shifted out of equilibrium by modest exogenous shocks.  In contrast, the American Right tends to be sanguine about the threats posed by environmental degradation due to non-linear dynamics on the horizon.  Rather, the American Right focuses on the fragility of social and cultural systems, while assuming the robustness of ecological systems.  The American Left on the other hand assumes a thin model of human culture and society which can absorb shocks and changes in norms and values, as the systems reequilibrate with ease.  Libertarians and a certain species of European conservative are consistent by these lights, the former in their optimism about the possibility of relatively easy change, and the latter in their pessimissm.

I will not offer any judgment as to the accuracy of the perceptions by the Rights and Lefts on the issues above. I only point to these as evidence that the same general cognitive toolkit is universal across ideologies, but the skeptical inclination is often not taken uniformly.  The same variations can be seen when it comes to the trust of rational systems or empirical data.  More on the Right than the Left may accept the compelling logic of comparative advantage, while more on the Left than the Right accept the logic of evolution via natural selection.  More on the Right than the Left may accept evidence on sex differences from psychology and neuroscience, while more on the Left than the Right may accept evidence that sexual orientation has a strong biological component.

As human beings we are not consistent, and our worldviews are filtered through our ideological presuppositions.  But it is important to note the context and processes through which our conclusions and inferences are generated.

This entry was posted in culture and tagged . Bookmark the permalink.

13 Responses to On the varieties of prediction

  1. Donna B. says:

    I agree completely. This is the type of post I expected to see here rather than all the jabber about religion and relative intelligence. Those topics are well covered elsewhere.

    Since I have never been religious, my conservatism must have derived from something else and I was looking forward to an intelligent discussion of conservative ideals and ideas, not religion and IQs.

  2. David Hume says:

    Those topics are well covered elsewhere.


  3. Robert says:

    I would add that the treatment of randomness is important to assessing a discipline, and that the common error in that treatment is to assume that complex things which we do not understand are random. They often only look random, however; it’s just that we don’t have the ability to articulate why a particular decision was made or a particular event occurred. The squishy subjects are rife with this error. “I don’t know why” and “it happened for no reason at all” aren’t the same thing, even if we can legitimately treat them as being the same at some levels of analysis.

  4. Roger says:

    Great post! Just a minor point, I think it would be more accurate to say “low” precision instead of “no” precision in sociology.

  5. David Hume says:

    Roger :

    Great post! Just a minor point, I think it would be more accurate to say “low” precision instead of “no” precision in sociology.


  6. Donna B. says:

    David Hume :
    Those topics are well covered elsewhere.

    The obvious answer is found by searching ScienceBlogs for “IQ and religion” and it is, of course, Gene Expression. What I do not understand yet, is a)why recycle, even if the information is relevant and b)why the information is relative to conservatism.

    Is religion (specifically evangelical) not simply relative to one component (a loud one, granted) of the conservative tent, rather than all? How is this more relevant than comparing the IQs of strict atheist libertarians with staunch atheist progressives?

    What I was hoping for here was a discussion of conservatism sans religion, not a discussion of how religion does not help conservatism. If I have misunderstood the tone of the previous writings (by several contributors), I apologize.

  7. Donna B. says:

    sheesh… relevant, not relative. I’d try to pass that off as a typo, but no one would buy it.

  8. lol says:

    Precision in genomics is on the order of 10e-4 for 1e6 datapoints per measurement event.

  9. InquilineKea says:

    What would demographics fall under? Granted, I see a lot of sociology as being partly demographic in nature, and the demographic aspects of sociology, at least, do have a good deal of precision.

  10. halifax says:

    Political skepticism of a conservative type has generally been informed by one of two traditions, an epistemological one and a religious one. The second group, the neo-Augustinians, have believed that human beings are inherently flawed in a moral sense and thus grand political projects cannot be trusted and government is understood to be a limited activity primarily concerned with maintaining order. Notable conservatives of this sort (other than Augustine) would most obviously include people like Martin Luther, but there are others like Herbert Butterfield, George Kennan, and Reinhold Niebuhr who are a bit more contemporary. The other group, including statesmen like the first Halifax and Lord Salisbury and theorists like Hume and Michael Oakeshott would claim that there are limits to the capacity of human beings to comprehend the world, and thus utopian projects (based as they are on hubristic calculations of human knowledge) are not only implausible but highly dangerous.

    I don’t think that your ranking of appropriate degrees of skepticism really has much to do with conservative politics, however. In fact, the one object of faith that is most common among those skeptical of religion is a quasi-religious faith in the metaphysical validity of modern natural science which betrays a lack of familiarity with both the history and philosophy of science. Philosophical skepticism roams free and allows itself to question all current pieties, not just those held by Bozos in polyester suits giving ‘come to Jesus’ sermons on the local access channel (or in the oval office with W).

  11. A-Bax says:

    “In fact, the one object of faith that is most common among those skeptical of religion is a quasi-religious faith in the metaphysical validity of modern natural science which betrays a lack of familiarity with both the history and philosophy of science”.

    Hasn’t Quine more or less addressed this very concern with his supposition that a thoroughgoing empiricism is open to the possibility that experience itself could (possibly) render the scientific method untenable?

    Once this difficult point is conceded, then the reliance that we place on science is not a “quasi-religious faith”, but rather a measured trust – one that welcomes philsopophical skepticism insofar as such skepticism helps sharpen the questions asked and answered. (As opposed to the kind of skepticism which Wittgenstein derided, somthing along the lines of “kicking up dust and then complaining that one can’t see”).

    My point is basically that even if we concede a kind of “point-of-departure” faith within science, it is on a different order than the faith employed in religious systems (which is a kind of clinging to a set of propositions, no matter what the evidence), and equivocation between the two is just squid-ink.

  12. Pingback: Secular Right » Who is scientistically inclined?

  13. Caledonian says:

    “In fact, the one object of faith that is most common among those skeptical of religion is a quasi-religious faith in the metaphysical validity of modern natural science”

    I do not consider metaphysics to be any more respectable an intellectual pursuit than phrenology or alchemy. Considerably less, in fact, since both of those fields made sufficiently clear claims that we could easily see that they were invalid, whereas metaphysicians surround themselves with ambiguity and vagueness.

    So whether modern natural science is valid according to metaphysics is utterly irrelevant. The question I am concerned with is whether metaphysics is valid according to modern natural science – as it is not, I reject it.

Comments are closed.