Roger wrote:

You exclude certain practices from science because they are too vague.
I do not see the basis for this. By the standards of space-telescopes,
the measurements of, say Copernicus or Tycho may have been imprecise,
but does that make them any less scientists? No. You say inflation is
imprecise. But (a) I could say "The government sets its goal as n%
inflation measured by the all items retail price index ... as precise
as you care to make it. (b) There is imprecision and "negotiation" in
science, as much of the sociology of scientific knowledge literature
makes clear. By requiring that science involve the application of
tried and tested theories you are confining science to what might be
called Kuhnian normal science.

The precision standards of science are determined by the technology--if someone today used the measurements of Copernicus or Tycho, would you call him a scientist? I would not.   But say that government is specific about its inflation goal--to achieve a certain level of inflation as defined by some measure, can we then provide a scientific answer. I believe not. You might do it by the way Abba Lerner and I suggested--creating property rights in prices, and establishing a market in changes in prices, thereby eliminating inflation. You might do it by price control; you might do it by limiting aggregate demand, or a variety of other ways. Each has advantages and disadvantages that go far beyond economic issues or economic theory, which is what I meant by the issues being vague. In my view,  economic science cannot answer what method we should use to control inflation. 

Studying various methods theoretically  make for interesting research questions for economists, but that research won't provide a policy answer. My problem is that some economists have argued that economic science provides an answer to how to control inflation--control the money supply to control inflation for example, and have built a view that doing so will work into their models, and then they present the models to policy makers as providing the scientific economic answer, and that other answers are unscientific. That stops fruitful discussion about a whole range of issues. I am not at all trying to limit science to the application of tried and true theories--I am simply trying to stop people from using science as a tool to shut up other views. When dealing with policy, it is best to see it as all art, so that a broader discussion can occur. It also removes the discussion of inflation from issues of policy and will allow scientific researchers much more flexibility in exploring inflation theories. Early on in my career, when I tried to develop a pure theory of inflation that relied on the difference in the average number of products consumed vs. the average number of products with lumpy costs of processing information. Few economists were interested in it because at the time, everyone "knew" that inflation was caused by an increase in the money supply. So I can hardly think that I can be accused of trying to confine science to Kuhnian normal science. My goal in limiting science is exactly the opposite--to free it to be far more imaginative than it is. 
For example, I strongly advocate taking a complex systems approach to thinking of macro questions. That approach hasn't had much traction because people are worried about policy implications, not simply understanding the macroeconomy and how the damn thing works. Even brilliant researchers such as Mike Woodford, got sidetracked into explorations of inflation targeting within DSGE models rather than developing his early work on sunspot theories and complexity.     


Roger wrote: 

In our culture, to call something science is, whether we like it or
not, to express approval, and to say that something is not science is
to denigrate it (I should perhaps add some qualifications about what
the science is being used for, but the point would still stand). Maybe
it should not be like that, but that is the way things are. Thus by
privileging anything that is not precise the effect is to reinforce
the profession's fondness for abstract theory (mathematics is the only
thing that can be completely precise) and, in practice, even if this
is not intended, to denigrate the messier practices of much that
should, in my view, be considered real science because it is trying to
understand what is going on in the world. The goal may be precision
(no quarrel with that) but that is a different issue. Surely
scientists who tackle important problems are scientists as much as
those who leave aside difficult and important problems for easier ones
that they can tackle more rigorously.

This gets at the heart of the difference between us.  Science is seen as good, and non science is bad. So everyone wants what they do to be considered science. Roger is arguing that we should succumb to that tendency--and call what policy makers do science. I believe that doing so has negative consequences for both pure science and applied policy. For pure science--it limits the range of questions it asks, and pulls too many scientific researchers into pseudoscience where they create models and then use the models for policy advice, with the ideology hidden in the assumptions of the model. For applied policy, it takes people away from the discussion of the assumptions and gives far too little importance to historical understanding, understanding institutions, and a broad knowledge of how models relate to reality--and gets them doing too much modeling for the sake of convincing others, not for the sake of shedding light on policy.  

I really don't care whether people are called scientists or not.  But I do care when what I consider an inappropriate methodology is used, when ideology gets masked as science, and when somehow one thinks that a pure scientist has something relevant to say about policy simply by the fact that they are a pure scientist, and generalists who have a sense of the institutions and history are not turned to for policy advice because they don't do science.  That is what I think the current state of affairs is in economics, and I think the Classical distinction which kept pure science out of policy, and made clear that its goal was not policy was useful.  As I said in an earlier post, it that goal could be achieved by some other way--say using Pigou's distinction between realistic science (applied) and pure science, I would be happy with that. But, given what happened with Pigou's realistic science welfare economics, I don't see it as likely that it will. I've toyed with defining an engineering or applied policy (political economy--Robbins' solution in his Ely lecture to this problem) classification to distinguish the two, but my suggestions haven't been taken up by the profession, which considers all of economics science. It is that "scientific" mindset that rules out the study of the history of thought or even economic history as economics--it's not science.  Other sciences don't teach history of thought--why should we. If you accept that economics is a science, then you can't answer them that you need history of thought and economic history to do good policy analysis.    

Having spoken with Roger about these issues, I think I can say that we are in large part in agreement about goals--our views about what constitutes good economics are similar. Where we differ is in the best way we see to go about achieving those goals. He favors going along with our culture's admiration of science, and for economists who do policy work to call themselves scientists. In principle, I have no problem with them doing so, but in practice I do, because that (1) leads people to use the mantle of science to rhetorically win policy arguments and to shut up debate, (2) it reduces the emphasis on knowledge such as history of thought and economic institutions that I think is central to designing good policy) and (3) it reduces the amount of really creative work in pure science.  

Dave