Hacker News Reader

Comment by aetherson | Hacker News Reader

Let's be clear: It's wrong in all scenarios.

It turns that it produces results that are close enough to reality to be good enough in lots of scenarios. (And, similarly to how you can ignore relativity in lots of scenarios, you can also ignore lots of other complexifying bits of physics some of the time -- plenty of mechanical equations will give you "good enough" results without taking into account air resistance, for example).

Yes, much of economics is like that. Economics is studying an inherently more complicated problem than simple billiard-ball mechanics, but the idea that economics has nothing useful to say about the real world is an agenda sold by people who don't like the implications of orthodox economics.

View on HN

>Let's be clear: It's wrong in all scenarios.

You are attempting to measure the objective truth inherent in some model. Be careful with this (or at least be explicit about it).

Newtonian mechanics is pretty accurate to an astonishing degree[0]. To call it "wrong in all scenarios" is saying that because it is wrong in one scale, it is wrong in another. At which point every physics theory we have is "wrong in all scenarios". The standard model can't predict or talk about planck level physics, and thus, it is "wrong in all scenarios". GR can't talk about quantum gravity and thus is "wrong in all scenarios". Hell, the Schrödinger equation is "wrong in all scenarios".

If you believe in a fundamental truth inherent in some equation or model of some system, that is fine. But I think that is far from an accepted point of view.

[0]http://www.npl.washington.edu/eotwash/sites/www.npl.washingt...

To put your point in other words, Newtonian mechanics isn't wrong. It's a system of measurement that is only useful at certain scales.

Imagine a yardstick that is only marked at full inches. You can use it to measure things in the scale of a couple of yards, a few feet, and many inches. You can't use it to measure anything smaller than an inch because it's not marked for that scale. You can't use it to measure anything more than a couple of yards because that's unwieldy. That doesn't mean that the yard stick is "wrong".

This is a flawed analogy, because the yardstick is accurate at the scales for which it is practical, whereas Newtonian mechanics is inaccurate at all scales, just less so in its sweet spot.

Newtonian mechanics is more accurate at yardstick scales than any real-world yardstick is.

Of course. Nobody is arguing that Newtonian mechanics is not a phenomenal approximation at scales that are very useful for humans, but that doesn't make the analogy any less flawed.

Then I don't understand what you mean. You said the yardstick is accurate at some scales whereas Newtonian mechanics is always inaccurate, which to me implies that the yardstick is more accurate at some scales.

This is a fair point.

The yardstick analogy was set up by the parent to be 'because the yardstick doesn't have marks less than an inch'

Then yardstick is always an approximation that is useful within a particular domain, just as newtonian mechanics are always an approximation that is useful within a particular domain. So far we are in agreement.

Newtonian mechanics always produces an incorrect result, however when the error is small enough to be neglected, because our measurements are noisy or we have no requirement for greater precision, then we can say that they are accurate for our purposes. This is pretty much the definition of an approximation.

It also must be pointed out that in order to know whether our application falls within the domain of values for which Newtonian mechanics are accurate enough, we must also understand something about relativity and quantum mechanics.

Newtonian mechanics alone can't tell you anything about when it is grossly inaccurate, and when it gives you a value that is indistinguishable from experiment. You must understand its limits in order to use in in the general case. It is therefore not 'perfectly accurate', but merely a good approximation based on limited data.

Economics is even more complex than physics in this regard, because in physics there is often a "right" answer. In economics, there may or may not be a right answer depending on the subject, but it's impossible to know if you've achieved it in most cases anyway.

And what investors try to do would be akin to asking a physicist to examine two cars before a Nascar race and tell you who was going to win the race.

You seem to have completely missed my (and Asimov's) point about how there are degrees of wrongness.

Newtonian mechanics may not be 100% accurate but it's extremely accurate for a wide range of real-world, useful scenarios, to the extent that any discrepancy with reality is often beyond what can be measured. For example, the Pioneer Anomaly was a discrepancy of a few thousand kilometers in distance, over a total distance of more than a billion kilometers. This was still enough to prompt a lot of investigation, and the problem was finally solved by properly accounting for photon pressure from thermal emissions from the spacecraft itself. This is a discrepancy on the order of parts per million that could not only be detected but was considered extremely significant.

I get the impression that economists, on the other hand, are ecstatic if their theories are within even 1% of reality. That's a whole different class of wrongness.

In reality, you have limits to the precision of your measurements.

In 99.9% of use cases, Newtonian physics are completely accurate, not a decimal out of place.

Market explanations of 'efficiency'? Way out of whack.

Newtonian mechanics are not completely accurate at any time. However as you point out, the error in the model is often dwarfed by the error in the measurement.

If it gives the exact correct answer, who are you to say it's not accurate?

It's not the exact correct answer. It's an approximate answer.

In certain kinds of abstract thought experiments, it might not match.

If you're doing orbital calculations, it's often wrong.

In the vast majority of real world calculations, it's not approximate. It gives the same answer as relativistic calculations.

Oh, I have an analogy. It's like using 64 bit integers instead of bigints. Any time you input real world numbers, they never go above ten million. Both models give you the correct answer, no error.

No model ever captures the entire nuance of reality. Even a notion of speed is misleading when your objects contain heat. So don't peek inside the model and judge it based on how the pieces interact inside. Judge it based on the results. And Newtonian physics typically give you ideal perfect results.

---

Or maybe I should put it a different way:

Newtonian physics are an approximation for the question "In a counterfactual thought experiment where my measurement had infinite precision in trailing zeroes, what would the answer be?"

Newtonian physics are exact for the question "What is the answer based on my measurement?"

The drawback in this reasoning is that with Newtonian mechanics alone you cannot know whether your measurements are in the domain for which it is accurate or not. For that you need relativity and quantum mechanics. Newtonian mechanics alone is seriously wrong, but if you know (because you understand relartivity and quantum mechanics) that you are in a domain for which it is sufficiently accurate, then you can use it to produce accurate results.

Reply on news.ycombinator.com