[jstrom]

Isn't that the logical approach to teaching it though? Start with the base case, analyze how it functions, then start looking it how it changes when you remove an assumption.

Similar to how physics starts with perfectly elastic point-masses in vacuum on a friction-less plane.

[jerf]

I used to think of the world of Physics 101 as PhysicsLand. It is a horrible place. There is no air. There are a huge number of infinitely sharp, yet infinitely rigid objects lie in wait to slice you to bits; even cubes can't be trusted with their infinitely sharp edges. Feathers and bowling balls are constantly falling on you from above, and there's no terminal velocity to slow them down. You can't even stand up to escape, unless you've been there long enough that friction has finally gotten installed. But beware, because shortly after that, the infinitely sharp, infinitely rigid things start whirling around.

It's a horrible place... and as far as most people make it in physics. Econ 101 may only be able to teach very simple models before most people also wander away to learn something else, but they are still useful models, and I still think people come away from discussions of how irrational humans are and how simplified economic models are and all these other things and think that the basics of Supply and Demand have therefore been disproved and I can go off and socially engineer without having to worry about them... but they haven't. The edges of the Law of Supply and Demand may fractal off into ever-more complicated corner cases as you approach their edges, most of which happen in the real world, but the core idea is still valid and you're still naked in the face of the real world's complexity if you think they aren't relevant, just as, for all the immense simplification, Physics 101 is still relevant to the real world even if it's the only physics you ever take.

[dragonwriter]

> Isn't that the logical approach to teaching it though?

If you assume what used to be called “the standard social science model” centered on rational choice theory is, if not a actually right, a reasonable approximation for common conditions akin to Newtonian mechanics, sure.

OTOH, if, as seems to be increasingly common (across the social sciences), you see it more akin to Aristotelean mechanics that coincidentally looks like some real phenomenon but gets the mechanism wrong for the general outline of how things behave, then, no.

[js8]

The problem is Physics 101 is a good approximation to reality, and you know where. The free markets are not a good approximation of normal markets, because the strategies of the actors are completely different.

The problem is with game theory - the limit of optimal strategies for some games is not always the same as the optimal strategy for the limit game. This breaks the ability to approximate.

So, for example, you cannot make conclusion from a game with infinite number of actors ("free market") to a game with finite number of actors.

[JumpCrisscross]

> The free markets are not a good approximation of normal markets, because the strategies of the actors are completely different

You can use freshman economics to predict the average oil price in a given year, from tables of quantities supplied and demanded. Where one finds deviation, e.g. when OPEC was founded, meaningful new information arrived.

Most markets don't follow freshman economics which is why there is lots of interest in developing better models. But we don't start physics with CFD.

[js8]

> You can use freshman economics to predict the average oil price in a given year, from tables of quantities supplied and demanded.

Not sure if I completely understand what you want to do, but if I do, this is not drawing supply/demand curves, this is just predicting the prices based on history of supply and demand. The supply/demand curves (that is, the model) is what I am criticizing.

[icebraining]

You can't predict based on history without having a model that tells you how to extrapolate that history.

[js8]

You could have a statistical model. You record supply, demand and price over time period and then you can predict price by matching it to supply and demand. No knowledge of supply/demand curves is needed.

But I am not clear if this is what parent wants to do.

[JumpCrisscross]

> You record supply, demand and price over time period and then you can predict price by matching it to supply and demand

That's what the damn curves are! Even calling them curves is misleading. Freshman economics looks at linear systems. You take data, draw a regression and then predict a price.

Supply and demand isn't voodoo. Early economics courses are inaccurate because they start with linear models a general population of freshmen without strong mathematics training can grasp.

[JamesBarney]

It is if you plan on eventually teaching the exceptions.

But if 95% of the students aren't going to study long enough to make it to the exceptions, and 95% of the real world are "exceptional" markets than maybe we should rethink how we teach it.

[scythe]

If you assume competitive markets are "simpler" than uncompetitive markets this is certainly true. However it is questionable to me that this would actually be the case.

Rather the advantage of the traditional approach to economics isn't that competitive markets are simpler but that the axiomatic framework which is built on them can be generalized in a mathematically rigorous way to describe most kinds of uncompetitive markets. In order to provide real intellectual (rather than political) competition a different way of teaching economics would need to start with axioms of uncompetitive markets and deform them to also describe competitive markets. That doesn't seem to be happening in this book, however.

[icebraining]

If you assume competitive markets are "simpler" than uncompetitive markets this is certainly true. However it is questionable to me that this would actually be the case.

I think it's true by the Anna Karenina principle; there are many ways for a market to be uncompetitive.

[quadrangle]

This is a broader question of pedagogy. There's first-principles approaches to teaching and then there's example-based, case-study and other approaches.

I happen to have sympathies with first-principles, but I've learned over time as a teacher that it can be quite flawed. I've grown to embrace statistical learning: show lots of complex real-world cases until the aggregate makes the statistically-significant patterns common to all the cases clear to the students. That's often actually superior to the abstract fiction of first-principles.

[arethuza]

"physics starts with perfectly elastic point-masses in vacuum on a friction-less plane"

It's been a while but I'm pretty sure our high school physics classes started with weights accelerating (or not) down an inclined plane - i.e. we actually started with actual trolleys pulling paper tape through ticker timers...

[tw1010]

Sorry to go off topic, but isn't this a case where "reasonable" would be a better choice of word than "logical". There's nothing particularly logical about teaching a thing one way over another.