Beware “mathiness”: Algebra and data to reinforce ideological preconceptions
johnkay.com
johnkay.com
- Mathiness is truthiness, but in the way that putting down any mathematical symbols lends an air of intelligence and truth to an argument, for many people, especially those who don't or can't take the time to understand the argument logic.
- Then he explains how economics data are not as easy to construct with universal agreement as with, say, physics (temperature). Agreement on practical action is hard when contrary frames and interpretations appear valid.
- He then mentions how this might be a problem for defining capital and other constructs for Piketty, but without any useful argument beyond this assertion.
- He then contrasts Feynman's integrity, trying to disprove his own work to make it better, to economist George Stigler's rhetorical style of conviction, ignoring contrary arguments, and playing the polemicist.
- He then mentions Isaiah Berlin's distinction between foxes who know little about something , and hedgehogs who know one big thing.
All these are interesting frames by which to compare and contrast various things. Yet his analysis, after bringing in all these ideas, is just to say that economics needs both careful analysis and effective rhetoric. Well, duh, but how does this tie into all the great setups he's made so far?
And mathy people are good at neither rhetoric nor polemic? But surely, if you're afraid of these people, they would be a mathy person who is using rhetoric to undermine the real philosophy and logic that should - pace Plato - be informing the argument.
It's a post full of story and setup, but as yet, signifying nothing.
Does it qualify as an actual prediction? Broken clocks are correct twice a day but...
It is quite another thing to replace assumptions by handwaving, in which case the point of math and logic is lost, and the argument becomes non-falsifiable. This is what the article calls "mathiness" and argues against.
Every careful person equipped with a reliable thermometer will make the same reading of temperature. There are alternative scales, Fahrenheit and Celsius, but both record the same thing...
to
Economics is genuinely harder. National income is a more complicated concept than temperature, and there are plausible alternative sets of rules for calculating it. Serious minded statisticians have spent many years discussing these issues, and there is now a UN-sponsored standardised system of national accounts.
But it is easy to write a mathematical symbol without giving thought to what observable fact in the real world corresponds to that symbol, or whether there is such an observable fact at all.
But isn't that exactly how we settled on the truth of temperature? Years of debate about what the right constructs for defining temperature mathematically are?
No. The concept of temperature is based on the bulk interactions of matter due to the motion/collisions of atoms and molecules relative to an inertial reference frame. A frame that usually has to exist outside of the system for temperature to have practical meaning so it is a concept that by definition has an absolute zero, even if we haven't observed it yet.
In this case, absolute zero really means zero motion relative to the instrument measuring the sample so it is possible for a Type 1+ civilization to build a planet sized (or bigger, don't know how the math works out but it probably has something to do with constricting atoms to a precision within a planck length) laser or other electromagnetic trap to cool some nontrivial number of atoms to absolute zero like those used for Bose-Einsten condensate experiments.
It may be possible, but we seem to be a very long way away from being able to test such a hypothesis.
Let me turn this around. What do you think the hypothesis is, and how would you know if you've succeeded? Do not use the term "absolute zero" because you think that is a hypothesis, and not a definition.
Your position is absurd. It's equivalent of saying, that distance is an unproven hypothesis, (which literally doesn't make any sense, because that's not what the word means), or perhaps more charitably a suspicious concept because we've never measured anything with zero distance between two items, because the closest two particle can get is planck length.
I'm not trying to be mean, but what you've been saying is pure gibberish.
You really need to understand the order of dependence of explanations.
Temperature scale is a good model that maps the territory quite well. There is nothing to suggest it will suddenly stop working at 0K, if it works very well even as we get closer and closer to absolute zero.
Contrast with most economic models which as maps correspond to territory in a similar way the map of China corresponds to the Middle Earth - there are trees and not trees, you have to turn it around to make the sea be on the proper side, and... that's about how much it matches. Be wary of zooming too much in, lest you get eaten by an Orc.
In fact, there is a good argument that 0k is unachievable [0].
[0] http://wordpress.mrreid.org/2014/07/03/why-cant-we-get-to-ab...
People acting as groups exchanging capital is just as real as atoms acting as groups exchanging energy.
Even the concept of "temperature" is not obvious. Give someone a hot piece of bread and a hot piece of metal and the metal seems hotter even if they're at the same temperature.
What we can do with our temperature scale is build a model that says every time we convert n liters of oil into heat and direct it at x liters of water we get a change of temperature of y degrees. Reliably, and with small margin of error.
We can't do this with GPD, and if we declare that a 10% increase in education spending will lead to a 5% increase in GDP over 15 years, we don't have a testable scientific model, we have a guess based on our interpretation of extremely messy data.
Much later, we figured out what it actually meant in terms of atoms.
Assuming unstated conditions, like normal air pressure, is just as well an entry point for error if you want to reliably describe the assumed state of things.
http://www.newyorker.com/magazine/2010/01/11/after-the-blowu...
I'm no economist, but I am a mathematician, and I'd guess the problem isn't that people are trying to use math to knowingly give false credence to plainly ideological ideas, but rather that the mathematical skills of those who listen are too weak to isolate the problems in the arguments. I would hope economists are better than that, but I know politicians are not.
http://paulromer.net/stigler-conviction-vs-feynman-integrity...
I don't have access to Stigler's 1955 paper. It would be interesting to know what he was talking about but putting him in opposition to Feynman seems pretty harsh. It makes it sound like Stigler is a proponent of idealogical thinking and that's how he practiced economics.
People give a lot of authority to mathematics, because mathematics is immensely powerful to explain the natural world. This unfortunately leaves the opening for charlatans to do fake math and lend their garbage research a completely unwarranted air of legitimacy.
Yes, you see, there are ideologically-driven economists like Piketty, who have a political stance. Then there are are the fair, neutral, unbiased economists who disagree with Piketty, who are only motivated by the search for the truth.
There is no such thing as truth in Economics. There's not even a defined goal. Creating a mathematical model of how 7 billion people behave in a global economy is impossible.
You're always going to have to simplify your model enough for it to be useful, as the expense of something. That compromise is your bias. Economists always make assumptions about something, and those assumptions are always driven by some sort of world view.
Not to mention, the other part of economics is deciding what policies or actions should be taken. Do you optimize for economic growth? For happiness? To reduce inequality? To eliminate poverty? Maximum employment?
Other sciences describe things which can be replicated, or things which are (more or less) concrete. Economics can't be a concrete, truthful science because people are react to their environment, which is ever changing and will never be the same at 2 points in history. And everything affects human behaviour, from the weather to things like religion which aren't exactly concrete...
And if that's not difficult enough, you have to assign time-values to all of these. E.g. reducing inequality today might lower growth in 20 years.
It's not called the dismal science for nothing.
The mathiness comes through the fact that the simulation is nothing like real life (e.g.people care about far more than their immediate neighbors) and yet the claim is that tweaking the parameters of this model gives some intuition about the quantitative real life effect.
It’s the same in economics, but economics is a field that is both complex (informationally) and complicated (incidentally), so our models aren’t very good. Often they’re only falsifiable in theory, because a practical assessment would be too costly.
What I have in my mind when I said mathematical models provide ad hoc suport is this. Today, it is consensus in the philosphy of sciences that there are two kinds of confirmation of a theory: confirming instance vs positive instance. Mathematical models generate "positive instances". A non-adhoc hypothesis generates confirming instances, a special type of positive instances. For more, pp. 61-62, Larry Laudan's "Science and Relativism: some key controversies in the philosophy of science.
Lets look at a hypothesis that says "All swans are white". One keeps on doing field work, saying that 1001th swan is white, 1002th swan is white, etc. All these field reports are "positive" instances. Such field reports don't add much.