Comment by matt4077 | Hacker News Reader

Comment by matt4077 | Hacker News Reader

Parent Full thread

By "rigor" you mean "math". But economics itself is the best proof that math can be used to cargo-cult your opinions to fame and fortune.

While there are some attempts with evidence-based policy, these sciences just aren't really at the point where math is useful (yet).

Math mostly requires reducibility: the ability to split it into small parts and look at these individually. That approach worked incredibly well in physics, and with the Manhattan project at the latest, it proved that it was mightier than any human endeavour before it.

But while, theoretically, everything could be reduced to physics, that isn't always the level of abstraction that works best. If you're depressed, a physicist isn't going to help you. Medicine may – but a psychologist, even though the discipline is scientifically flawed - is still an important tool in the set available to us.

Sociology is psychology on a population-scale. In that regard, it may be almost impossible to reduce because of nonlinear effects, chaos etc. That severely limits applying results from prior experiment X,Y,Z to situation B – because what you usually do is split situation B into components (B1=X,B2=Y,B3=Z,...), each of which is quite generic and has been studied in detail. So you'd have to run your own experiment on B, and the questions of sociology would require you to run an A/B experiment, with both A and B being, for example, New York.

I remember coming across this field of study some time ago, along with its psychological counterpart. I'm not trying to argue for or against your point but I thought it might me interesting to us all none the less:

https://en.m.wikipedia.org/wiki/Mathematical_sociology

> theoretically, everything could be reduced to physics

Is there a thing as free will, and if so, is it explainable in terms of physics? And with consciousness (if you believe there is such a thing)? Is morality and ethics reducible to physics as well? There's a lot of presuppositions to the statement that everything is reducible to some purely mechanical process.

I have no doubt that many serious people believe as you have stated, but it's not a self evidently true claim.

You seem to be claiming that the results of sociology cannot be predictive, only descriptive.

If so, that makes sociology completely useless for policy. A field which cannot predict what will happen next time can't actually inform a policy maker as to whether their policy is useful.

Math mostly requires reducibility: the ability to split it into small parts and look at these individually.

This is simply false. If this is your criticism of math in economics, it applies equally well to math in geophysics, oceanography, climate science, etc.

My fav quote on that: "Trying to predict the future is like trying to drive down a country road at night with no lights while looking out the back window." (Peter Drucker)

The better critique is that math requires homogenous data points. In sociology, every policy implementation is different in a million ways. A welfare program in Sweden is different than one in Chicago. The charter school program in Boston are different than those in Detroit. So doing statistical studies on a question like "do charter schools work" that treats all these implementations as identical, is often not very useful for actually learning something about policy.

The better critique is that math requires homogenous data points.

This critique is similarly wrong. There's a field called "machine learning" - you might have heard of it - which often uses inhomogeneous data points.

So doing statistical studies on a question like "do charter schools work" that treats all these implementations as identical, is often not very useful for actually learning something about policy.

So instead build a model that understands the important factors which make charter schools work/fail. Then use it.

If you can't do this, don't pretend you have anything to say about whether I should implement charter schools.

Climate science – as one example – uses the thermodynamic properties of CO_2 as a major building block that can be studied extensively on your desk.

And I'm not claiming that sociology cannot predict. I'm just saying that they cannot measure the small (CO_2) and extrapolate to the large to make predictions. Their predictions therefore need other methods.

> A field which cannot predict what will happen next time can't actually inform a policy maker as to whether their policy is useful.

That charge could reasonably be made against economics. From 2008: 'During a briefing by academics at the London School of Economics on the turmoil on the international markets the Queen asked: "Why did nobody notice it?"' [0]

[0] http://www.telegraph.co.uk/news/uknews/theroyalfamily/338635...

Economics makes good predictions most of the time. It's a known flaw in the model that the movement of stock prices is fractally distributed. The model assumes normal distribution instead, since it's mathematically much more tractable. This is usually fine, but causes the model to underpredict rare events like the 2008 crash. I don't think this invalidates economics though. That's like saying physics is invalid because the Newtonian model doesn't 100% accurately reflect reality.

> It's a known flaw in the model that the movement of stock prices is fractally distributed.

There's plenty more to the economy, and to economics, and to the crisis than predicting stock prices. Reducing the question to stock prices is avoiding the question, IMO.

> That's like saying physics is invalid because the Newtonian model doesn't 100% accurately reflect reality

Not really. Newtonian physics is inaccurate in ways that are hard to measure at everyday human scale. The 2008 crisis was pretty noticeable, all over the world. I think an apt physics analogy would be something like getting a nuclear explosion out of starting a fire (technically, out of buying a house), something that's been done successfully for millennia without major incident.

> There's plenty more to the economy, and to economics, and to the crisis than predicting stock prices. Reducing the question to stock prices is avoiding the question, IMO.

You're absolutely correct. The point I was trying to make is that the economic model we use drastically under-predicts "rare" events like the 2008 crash. I wasn't trying to make a statement about the "cause" of the crash, so much as state that the model under-predicts such events.

> Not really. Newtonian physics is inaccurate in ways that are hard to measure at everyday human scale. The 2008 crisis was pretty noticeable, all over the world. I think an apt physics analogy would be something like getting a nuclear explosion out of starting a fire (technically, out of buying a house), something that's been done successfully for millennia without major incident.

All theoretical models are based on arbitrary axioms/assumptions. The degree to which they predict reality successfully is largely predicated by the degree to which the underlying axioms correctly reflect reality. I don't think this is a binary thing. I don't think that the field of economics is "invalidated" because it "failed" to predict the 2008 crash. I might argue that it's "less" correct or reflective of reality than Newtonian mechanics is, but again, I don't think the degree to which our predictive systems represent reality is binary. I don't think that they're either right or wrong, only more or less successful.

Some people did, but they were not listened to.

Yes. And that's one reason why I don't buy arguments about economic 'rigor'. There's a narrative that goes along with the rigor, and the narrative isn't always right.

I don't understand your grievance with the usage of 'rigor'. I think there's a difference in occasionally failing to predict and (almost?) always failing to predict. When the field that consistently fails is often dependent upon autoethnographies and the like, it's hard not to criticize it for lack of rigor. Economics isn't perfect, but economists aren't basing their predictions on how they felt when they interviewed someone either. It seems to me that rigor is a perfectly valid distinguishing characteristic between the fields.

1) How much sense do equilibrium models make when money, and, especially, debt, which functions as money, can be created at the whim of certain participants (central banks, banks, credit-issuing companies) in the 'equilibrium'? It's ironic to me that the logic and analysis of scarcity is associated with the least scarce, most easily created material (money/debt) in the universe. At least with physics, matter can neither be created not destroyed. Feynman: 'There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers.'

2) If you put a Keynesian economist, a neoclassical economist, and an Austrian economist (just to pick three flavors) in a room together and ask them to sort out what they agree and disagree on, the assumptions they'll make are more about narrative than about rigor... which would suggest that narrative is more important than the rigor folks give it credit for.

To elaborate:

http://www.twill.info/wp-content/uploads/2013/03/SOME-ECONOM...

Dean Baker's paper from 2002:

http://cepr.net/documents/publications/housing_2002_08.pdf

>If so, that makes sociology completely useless for policy.

Only in as much as one believes that descriptive results are "completely useless" for policy.

For me they are just as important, if not more, than predictive results. Predictive models only tells us what will happen -- not what is going on, if it is any good, ways of thinking about it, and whether we should ask for more or less of it.

I'm a policy maker. What do I do with a non-predictive result? Please give a concrete example.

"Blacks are humans just like we whites are, not some inferior race that's OK to be exploited".

I'm a policy maker. I put an end to slavery.

No prediction involved.

"X prevents women from getting into IT".

I'm a policy maker. I take steps to help women get into IT.

No prediction involved.

Conditions in prisons are bad because so and so.

I'm a policy maker. Let's change the conditions.

No prediction involved.

Sometimes the prediction is obvious and implied (removing X will allow more women to get into IT), other times it's not needed at all (when the change is about doing what's right, like with slavery, not about doing something that we predict will turn things in some specific way).

Your first example is based on moral philosophy, not sociology.

Your other two examples - as you note - are based on an explicit prediction. Now the question arises - are sociological predictions any good?

You can't simultaneously claim sociology is descriptive, and therefore can't be judged on it's predictions, and then turn around and use it's predictions to describe it as useful.

>Your first example is based on moral philosophy, not sociology.

Sociology is inherently tied with "moral philosophy" and normative ideas about society, and has always been, and it's something sociologists, at least here in Europe, have been perfectly open about (e.g. Foucault Bourdieu, Pierre Bourdieu, Henri Lefebvre and tons of others) It's not, for example, just a positivistic gathering of statistics and a supposedly "neutral" construction of prediction models.

>Your other two examples - as you note - are based on an explicit prediction

No, they are based on observation of existing conditions. No explicit prediction involved, as noted. At best there's an implicit prediction that's an obvious conclusion from the observations (e.g. "if we reduce macho culture, more women will get into IT").

People may be able to make accurate predictions after reading sociology for its qualitative findings, without it being possible to make these predictions formally using our current mathematics.

Reply on news.ycombinator.com