May 1, 2003
From the desk of Jane Galt:I've gotten some interesting mail and comments on yesterday's post. A lot of them had to do with predictions.
Many of them made the very true point that economics doesn't make rigorous predictions like physics. Which is true, but then physics doesn't either in the way that they're implying. When you throw a baseball, a physicist could theoretically predict where it will go, provided he could take accurate measurements of all the variables -- knew exactly what the wind was going to do, could trace all the micro-movements you will perform with your arm as you hurl it, could get an instant measurement of the force with which you threw, and otherwise could get an instant handle on all the variables affecting the path of the ball. In a dynamic system, he can't, so the best he can do is give you an approximate heading. The fact that he can't instantly tell you exactly what is going to happen in every physical instance doesn't mean that the insights of physics and chemistry can't give us general principles that enable us to build a pretty good bridge. Similarly, no economist can tell you exactly what the economy is going to look like in six months (although if you average their predictions, they apprently do a better job than you'd think at setting a rough range of performance). That doesn't mean that nothing they say is of any possible interest, to be used only when you need window dressing for your pet political causes. There are a lot of areas where economics can, and does, give us guidelines for building prosperous systems.
Then there were the people from the other side, the humanities. I'm afraid I've spent far too much time already in my life arguing about whether knowlege is socially constructed or not, and I don't want to start up that debate again, so I'll just say this. Science is the attempt to understand a phenomenal world that I'm going to posit actually exists, and actually has immutable properties. If you don't think this is so, I invite you to step out the nearest fourth floor window and then send me a report on the social construction of gravity.
Others made more interesting points. There were a surprising number of political scientists arguing that they didn't want predictive models; the important thing is a good explanatory model. Said one "I'd rather have a really good explanatory model than a mediocre predictive model." There are a few problems with this. The first is that predictions are how we test whether we have a good explanatory model. I can construct an explanatory model for any set of facts, given enough time. Starting out knowing exactly what I have to explain, I can build a beautiful story for why it is so. But the only way I can know if this model actually works -- if my explanation of the War of 1812 is valid for anything besides19th century former British colonies -- is to make predictions about the next war and see if they hold up.
Think of Greek Astronomy. It worked pretty well as an explanatory model for several thousand years. All the planets revolved around the earth on a series of interlocking rotating cycles. The whole thing was beautiful, fit all the known facts, and was easy to understand to boot. The only problem is, it was wrong. And how did they know it was wrong? Because over time, it failed to accurately predict the movement of the stars. Now, I could say "I have a great explanatory model for star movements between 1200 and 300 B.C.", which is true, but that doesn't make it any less wrong.
Which brings me to my second point: what is the use of a science that can only tell us what's already happened? That's not science, but art. Which I love. I don't want my novels to provide me a mechanistic model of human behavior; I want them to make me feel what it would be like to be someone else. But while art can describe and illuminate the human condition, it does a very poor job of helping us change it, which is why political novels are so uniformly dreadful.
We already have a humanities discipline that tells us what happened: history. Building models of why it happened doesn't seem to me to be particularly useful, or even different from what history already does, unless I can hope that someday, it's also going to tell me how to avoid the bad parts, and replicate the good parts, before it happens again. I don't think political science has to be at this point right now, for after all, the history of all the sciences is mucking around with a lot of things that don't work before you find some that do. But I find it disturbing that I received a number of emails arguing that political scientists shouldn't try to build predictive models -- that the only really important thing was talking to each other about what had already occurred. Or at least that's what they seemed to be arguing. If that's really what you want to do, you ought to strip the word "science" out of the discipline's name.
There was one reader who emailed me something about the scientific method that I think is important, which is that if hypotheses have to be falsifiable, results have to be reproducible. Which is to say, in other words, that your model has to be broadly predictive. If the only person who can get the results you get is you, that weakens your theory to useless in the eyes of other people. I had an interesting conversation last night comparing this with the humanities. In some disciplines, when the only person who can produce your results is you, this enhances rather than destroys your reputation. More often, you see incidents like the Bellesiles case, where no one even tried to reproduce his results for an extraordinary period of time. In the sciences, if you produce extraordinary results, you see a rush by other teams to try to reproduce it in the hopes either of getting to be the guy who takes down Icarus, or of getting a little reflected glory for yourself. (And of course, the hopes of finding out whether or not some extraordinary result is true.) Bellesiles, who said something that flatly contravened the common understanding of a fairly major thread in American history, should have been mobbed by people trying to reproduce his results -- and their fellow historians should have been cheering them on. In reality, anyone who challenged this finding was dismissed as a gun-nut driven by vainglory and vendetta.
Posted by Jane Galt at May 1, 2003 10:16 AM | TrackBack | Technorati inbound linksExplanatory theories can be useful, particularly because there is not often a concensus about why things happened. Take history, for instance. Many historians were convinced that slavery was instituted merely because of the evil of the whites, and that is was a poor system all around. Fogel and Engerman's book Time on the Cross put paid to this hypothesis, by using statistics to prove that slavery in the antebellum South was efficient. It may not have been the conclusion some wantd to hear, but it was no less important for it.
Even though it's fun to take down Icarus, there's one more reason why reproducible results are important: if you're doing something useful and you get it to work, other people will want to do it too to see if it will work on what they're doing...
For example, many Nobel prizes go to scientists who pioneer techniques that are really, really useful and robust (I can do it, even if I stink at it!) and finally, have a good explanatory model as to why they work (and other attempts haven't.)
If you don't think this is so, I invite you to step out the nearest fourth floor window and then send me a report on the social construction of gravity.
You don't understand. It's irrelevant whether an individual believes in gravity or not. What's relevant is whether the consensus believes in it. If you don't believe, the force of the consensus is still going to splatter you on the sidewalk.
[/tongueincheek]
A number of people believe that the science of economics is going to end up emerging (sorry for the pun) from complex systems theory. The general idea is that complex phenomona behave according to a set of principles yet to be characterized in any sort of formal way, and that these phenomona are not tractable in the current models (the same way quantum phenomona were not tractable under Newtonian analysis). The same type of problem is thought to underly schools from evolutionary and developmental biology to spin glasses and turbulent flow.
So, my thought is that the science of economics is awaiting a new analytical tool (or, really, a model which itself provides tools), and until then will not be as effective as we might like it to be.
-BF
NICE point, David.
In the humanities, it IS consensus that will get you splattered on the sidewalk. It doesn't matter what the truth is regarding, say, racism. Whoever disagrees the majority gets splattered.
"Building models of why it happened doesn't seem to me to be particularly useful, or even different from what history already does, unless I can hope that someday, it's also going to tell me how to avoid the bad parts, and replicate the good parts, before it happens again"
Why should history "tell" you anything? Why can't you learn from whatever shape/size/form that it was? Human nature being what it is ( I do not really need a behavioral sceintist to tell me what my nature is - if I have a head and can use it ) - how difficult is it to contextualise what happened alongwith what is happening to arrive at the nature of the process that one is dealing with/surrounded by/is part of ?
Specifics can go jump out the window - one needs just a sense of pattern to figure out what is what......
or maybe i need some ivy league education to do that?
hmmmm........
i see!!
heheh!!
"what is the use of a science that can tell us only what has happened?"
did i read that correctly? it is for "me" to put it to use - i kind of figured that that is where "imagination" is called in to play.....alongwith "reason".
or is it that the "need" for comfort implies that one desires ( unspoken as yet ) to be rid of the responsibility to "imagine" "reasonably"
....
I'm sorry, T.Rex, but I have absolutely no idea what your point is.
> I'm sorry, T.Rex, but I have absolutely no idea
> what your point is.
The point I got was that he needs an Ivy League education, or at least a community college writing class.
[Sorry - I couldn't resist. I haven't read much on this site yet, and realize that T.Rex might be a well informed and respected contributer. That one post was pretty difficult to follow, though...]
-BF
hmmm.....
let me do it again - i guess my comments and that which prompted them stem from a rather disturbing cognition that most of us do not really really know what we are talking about.
that takes into account even "knowledge" or concepts thereof.
to amplify further - since i see scope for further loss of meaning.....we do not really trust that which we do KNOW - already.
somehow we seem to lack imagination or the ability to use it effectively - and therefore are condemned to re-experience history ( of errors translating into colossal consequences and deaths and suffering and all that jazz )in a "modern" "scientific" "rational" ecology.
my point was to merely wave a flag at the statements in dear galt's post which i italicised - they open up a pandora's box of interpretations - none of which are favourable ( though i admit - none of them might be what galt intended ).
i guess i just threw in what came to my contrarian mind - and allow me to assert that this does not necessarily deduct from their validity - that is how i might have spoken if i were to be responding to them in a conversation.
"the devil is in the details" - if we do know so well whatever we do - where do the errors come in from? those that cause human catastrophes and misreable consequences?
is it not reasonable to surmise that it is because the subtleties of impact and interpretation - e.g. little ones such as those of just those two statements... are often missed by those who "know"/"speak"/"do".
whose job is to "ensure" that the essence of the meaning ( and intent ) is conveyed to the recipients of such knowledge/words/actions?
i'd suggest that it is easier to do anything than to ensure that the meaning of it is surely conveyed across.
case in point - just these very comments of mine.
lil flag - but big wave........
hope it registers
:D
( if it does not - one can try again )
what is the core issue actually? the debate is fine and fit for a nice evening over drinks - but when it comes into action on one's feet.....?????
i guess i am commenting at the moment - because this post somehow gave me a funny sense of dissonance - it "seems" rational and balanced and critical enough.....and yet seems to skim the surface.
regards
This reminds me of an interesting role playing game, Mage: The Ascension. The "metaphysics" of this world is that that the natural laws are what everybody believes those laws to be. So, if enough people can be persuaded to really believe that, say, gravity is optional for a person who says "Wingardium Leviosa," then it will be true.
heheh! you're pretty accurate there - tom!
contrarian me - lets see.
paragraph one : - general principles.
general in terms of encompassing the whole of physics ( for example ) or for effective utilisation by the general populace?
what exactly is the point that is being conveyed?
paragraph two : - science is an attempt.....
right.
in which case it is laced and interlaced with error.
in which case it needs to be treated with a healthy does of skepticism.
yet it is the religion of the day.
???
paragraph three ; - "The first is that predictions are how we test whether we have a good explanatory model"
is it so?
i do not really see what good explanatory models have to do with predictive power - it is such a fabulous chaos out there and the context alters in richness from instant to instant and the variables that are to be considered alter radically - how on earth can any kind of theoretical explanatory model be good enough when stressed under the pressures of reality as it IS.
( note - this brings one to register that a counter argument is in the offing - in terms of what galt had to say that there is a phenomenal world out there with immutable properties.....maybe i shall take that on separately :D )
paragraph four : -
is that the only reason that it was wrong? or was it ( also ) that there was additional info that was dug up that contradicted what was earlier held as the gospel truth? or was it evolutionary?
lil bits of info kept adding to the original hypothesis that caused it to evolve to a form that proves the earlier form wrong?
paragraph five : -
does art do a poor job in changing the human condition? or do we? ( if we are really really so interested - then why don't we start with the one nature that we do know? our own individual selves?
but there we enter into a tradeoff with convenience and comfort ( of any kind ) and thereby enter the realms of approximations and therefore errors - why does this fact not figure into these debates and discussions?)
( this is the crux of my earlier comments actually )
who is art for? who is it "supposed" to affect and therefore influence?
what causes it to fail? is it that art is inadequate? or is it us?
paragraph six :-
"But I find it disturbing that I received a number of emails arguing that political scientists shouldn't try to build predictive models -- that the only really important thing was talking to each other about what had already occurred. Or at least that's what they seemed to be arguing"
I'd see this as a challenge for us to expand the rather narrow definition of science to incorporate talking to each other about what had already happened.
why not?
why is this not "scientific"?
( though this contention of mine could be dismissed easily on the face of it - it is rooted in the idea of neural networks and communications and all that jazz of complexity and influence - pretty scientific those things - or are they not?)
parting shot: -
"If the only person who can get the results you get is you, that weakens your theory to useless in the eyes of other people"
it just means that this one person needs to be placed in the place of dear dubya!!!
regards
i meant to draw an "analogy" here by referring to the parallels in what political scientists are asking for or recommending and that which seems to be necessary in the development of AI
as regards the "parting shot" - just plain and silly nonsense
:D
Best be careful about repealing the law of gravity, lest you get flung (along with all the atmosphere, and everything/one else) at about 1000 miles per hour.
worth a looksee -
http://examinedlifejournal.com/articles/template.php?shorttitle=lesauxresponse&authorid=14
Hi -
I've been lurking here occasionally, finally a reason to post.
I've been working as an economic forecaster for the last 15 years or so, first as a freelancer for the machine tool industry in the US, then in Basel, Switzerland for BAK and for the last six years for Feri Research GmbH in Frankfurt.
I currently run a 114-sector model of the Austrian economy, covering NACE rev 1 A to O for those for whom this means something. For those who don't know, it means we cover the entire supply side of the economy from agriculture to consumer services, but exclude government and private contributions to the creation of GDP.
Forecasters exist for two reasons: first, they are the ones who get blamed when decision makers make the wrong decision. Second, they generally do a better job at it than lay people, and anyone without formal training in econometrics is a lay person.
That said, my forecasts are input into credit risk assessment programs. My forecasts give a sound appraisal of which sectors of the Austrian economy are going to do better than others, and which ones are going to be real dogs.
We employ around 15 economists. There are a number of PhDs, and they have their place. But the really effective people, the ones churning out the work, are not PhDs, but rather, like me, MAs or MSs who decided they wanted to work in the real world rather than in academia. The biggest hurdle when someone starts working for us is for them to adjust from academic interests to real-world needs. We don't place much value on superb statistical analyses, since we know how terrible the statistics are. We place value on sectoral behavior that makes sense, not on academically nice developments that are not usable in the real world.
Just my two cents: if your forecaster knows what significance r2 has in reality, it's worth talking to him. If he knows how to do Durbin-Watson off the top of his head, then be suspicious of him, since he probably loves the statistics more than is healthy. If he can explain in great detail why the model behaves the way it does to a credit risk manager, who is not a trained economist and whose eyes glaze over after 5 minutes, then you've got the right forecaster.
John
hi t rex,
"I'd see this as a challenge for us to expand the rather narrow definition of science to incorporate talking to each other about what had already happened.
why not?
why is this not "scientific"?"
it could be. some guy called marx came up with a "science of history." though many cry that it was proven wrong with the fall of the ussr, that experiment is still actually being played out, and we will get to see whether the man was right or not in the next 50 odd years. i ain't saying he is right, but i ain't saying he is wrong either. i'm taking a wait-see attitude on that one. but it is still a predictive model. and it also happened to be an explanatory one as well... marx was a philosopher as well, and pretty big on history, economics, etc. a real polymath.
"is that the only reason that it was wrong? or was it ( also ) that there was additional info that was dug up that contradicted what was earlier held as the gospel truth? or was it evolutionary?"
feyerabend's work, against method, is a dangerous read. i thought he made an interesting claim--the old model still explained better than copernicus' new-fangled model. but copernicus won out because his was aesthetically more pleasing to those who followed than the ad-hoc ptolemain model, and one that admitted of more imaginative possibilities. scientists have artistic tendencies, even if they don't want to admit them--elegance and simplicity for example--and these are judgements of style (i think of paul erdos here who always looked for a more elegant mathematical proof).
and as for the parting shot--of course it is just "plain and silly nonsense."
I think the point I was trying to make is that a model that predicts how 95% of citizens will behave - which would be a damn good model in the social sciences, including economics - can still be wrong in the aggregate (e.g. 2000). People are damn hard to quantify, which is why we run around making all sorts of simplifying assumptions like rational choice and utility = $.
That being said, if your R^2 is .1 or you can't outpredict the modal category in a limited dependent variable, chances are your explanation will suck as much as your prediction will.
Cas, By any chance, are you a fan of Bellesiles? Cause, they did the same thing, manufacture their information to "prove" their point. Even more embarrasing to Marx, it wasn't uncovered by a critic, but one of his disciples, Bernstein.
There is nothing "wait and see" about a fraud.
cas : hi
i agree @ elegance and simplicity.
but that's how it all ties in - don't you see?
art and science....and all else.
i somehow do see this as funny - we make these categories and distinctions and then align ourselves firmly along them and then argue from within those confines.
polymaths - that could be an interesting subject to teach. (:D)
seriously though - if one is looking for a grand and unified theory of everything - then these categories of ours need to be merged somehow.
and then there is the added consideration - if these categories were made - and they persisted - why so?
what was the original purpose?
would it not be that it was meaningful towards the relative ease of "understanding" since the depth/degree/dimension of each field was getting too expansive for any single man to examine easily?( irresepective of how this purpose/meaning has got lost/bastardised over time )
however - whatever came of it ....if meaningful - has to be hauled back and interpreted in the light of all other categories ( we do not do much of that these days - or do we?).
regards
hi nick m.,
thanks for the thoughts.
"By any chance, are you a fan of Bellesiles? Cause, they did the same thing, manufacture their information to "prove" their point. Even more embarrasing to Marx, it wasn't uncovered by a critic, but one of his disciples, Bernstein.
There is nothing "wait and see" about a fraud."
i am not really familiar with bellesiles, apart from a cursory glance at a couple of critiques. i find it interesting that you sweepingly put him and marx into the same category.
on bernstein, thank you, for i was unfamiliar with that debate. i found this, http://www.marx2mao.org/Other/BMSI68.html, a very interesting read, and i liked this:
" There is no doubt that Bernstein expressly rejected all this. The best proof, if proof were needed, is his concern to demonstrate the possibility of the 'self-regulation' of capitalism. Cartels, credit, the improved system of communications, the rise of the working class, insofar as they act to eliminate or at least mitigate the internal contradictions of the capitalist economy, hindering their development and aggravation, ensure for the system the possibility of unlimited survival. In other words, for Marx's basic conception according to which the advent of socialism has its preconditions and objective roots within the process of capitalist production itself, Bernstein substituted a socialism based upon an ethical ideal, the goal of a civilized humanity free to choose its own future in conformity with the highest principles of morality and justice. As Rosa Luxemburg acidly commented: 'What we are offered here is an exposition of the socialist programme based upon "pure reason". We have here, in simpler language, an idealist exposition of socialism. The objective necessity of socialism, as the result of the material development of society, falls to the ground."
i don't think marx proved his point. i don't think he was really given a chance to. i just think that we are at a "world historical" moment when we get a reasonable chance to see whether marx is right or not. i am agnostic on this. if i read the bernstein article rightly, he was critiquing marxism's move to look at the inevitable natural law that would bring about the downfall of capitalism, rather than marx's historical tendency for there to be such a breakdown. i am interested in marx, rather than marxism, so here is the argument as i understand it:
marx's central mechanism is the "law of the tendency of the rate of profit to fall to zero." at that point, business owners have to squeeze workers to make money, or go under, leading to more u/e, more immiseration, etc in a viscious downward spiral. eventually the masses rise up against the concentration of power, etc, etc.
before you roll your eyes and call me all kinds of names, hear out the rest (then call me all kinds of names, though i would prefer it if you didn't!). the law is a tendency, not an ironclad surety. three things worked against it:
1. cannibalizing firms to get their capital cheaply, so that you could use it to make items more cheaply, and thus, at given prices make a profit;
2. technological innovation that lowered costs, and kept firms ahead of falling prices
3. new markets and/or cheaper materials helped boost demand and/or lowered costs.
as long as you had any or all of those, you could stave off the evil day when profitability fell to zero.
the key is number three. the soviet union, under this thinking was an aberration. marx would have thought so. the ussr delayed the integration of world markets into the capitalist system for close on a century. that impediment is gone, and it is possible (if china doesn't descend into civil war and/or anarchy) that the last vestiges of non-integrated markets will be gone in the next 10-20 years. so, the dream, espoused by many folks, of a world-wide capitalism would become a reality. then, since it is world-wide, we get a good chance to see whether marx was right or not. it still may not happen, since we still have the wonder of technological innovation. and that has sustained us, and nano-technology, or something along the lines of stephanson's diamond age nano-forge scenario might be on the cards and might settle the issue once and for all. also, we might continue to see that wonderful self-regulation bernstein talked of.
i adopt a wait and see policy on this, that is all.
hi t rex,
thanks for the thoughts. i am still digesting what you have said--there are a lot of lines of flight and dense textures in what you say, and it takes me quite a bit of time to think through what you are saying.
but a quick thought re:
"i agree @ elegance and simplicity.
but that's how it all ties in - don't you see?
art and science....and all else."
out of curiosity, do you think of nietzsche's take on the relationship of art and science in say, the geneaology of morals, 3rd essay is germane to this discussion?
Jane, you wrote:
"Think of Greek Astronomy. It worked pretty well as an explanatory model for several thousand years. All the planets revolved around the earth on a series of interlocking rotating cycles. The whole thing was beautiful, fit all the known facts, and was easy to understand to boot. The only problem is, it was wrong. And how did they know it was wrong? Because over time, it failed to accurately predict the movement of the stars."
Um, are you making this up? Please don't pull stuff out of your ass in a post about scientific rigor. It's unseemly.
As it happens, Ptolemaic astronomy (which I've studied in great detail) was perfectly descriptive and predictive right up until it was replaced by the heliocentric model. It wasn't until later, with the invention of Galileo’s telescope, that it was possible to observe something that was incompatible with Ptolemy.
Someone, somewhere else, criticized you in this context for seeming to be someone whose limited exposure to the philosophy of science was in reading Popper. Predictive ability is a terribly important component of a "valid" scientific theory, but it certainly isn't the be-all and end-all that you seem to think it is. Because, first of all, you might ask yourself how you define predictive success.
There's a whole bunch of things that Newtonian physics can't explain — for example, the precession of the perihelion of Mercury. It it wrong in this regard? Yes. But, you know, our current model of the solar system is not built upon General Relativity. For that matter, it's not really built solidly on Newtonian gravity, either — a simple Newtonian four-body problem is intractable.
As someone whose education was delving deeply into the development of science (and not a narrative gloss, but actually recapitulating some of the work), it seems to me that your view of reality and how science describes it is naive. You're not horribly wrong, you're just naive. Of course gravity is not “socially constructed” as your unfortunately no-doubt real-life examples claim. But your intuition about how some theories are “true” or “false” deserves some intense examination along with quite a bit of supplemental reading.
Well, according to Adam Smith's History of Astronomy, that isn't true -- they had to keep adjusting the system, making it ever more complicated, because it kept getting out of step with the real world. That also jibes with what I know -- that there was a thriving business in adjusting the Ptolemaic system all through teh Middle Ages, until Copernicus came along and decided to see what came up if you junked the model.
No, the Ptolemaic model remained accurate, it just was not sufficiently precise as improved observations demanded. Elaborations were made to more closely fit observations.
These errors did not invalidate the theory — and almost every physical theory, save perhaps GR, has been similarly adjusted to account for better data.
(As an aside, I'd like to ask: what economic model, ever, has had the same predictive accuracy that Ptolemaic astronomy had prior to the use of the telescope? I'm asking, but I suspect the answer is "none".)
So I think you're quite wrong in your claim that Ptolemaic astronomy wasn't sufficiently predictive. It was.
Also, Copernican astronomy was no more predictive than Ptolemaic astronomy. The shift to a heliocentric did not occur in order to find a more accurate model — because Copernicus's wasn't. Rather, the shift occured because a) it radically simplified the math and increased elegance (at a cost); and, b) what was previously unthinkable (the Earth in motion, not the center of the cosmos) had become thinkable. Copernicus still needed little arbitrary epicycles and such to square (heh) his model with observation, and he achieved no better accuracy than the Ptolemaic model.
Furthermore, the heliocentric model was known to the Greeks — it goes back at least to Aristarchus — and it was known to Ptolemy, as well. In an aside in Almagest, Ptolemy says something about the simplicity of the heliocentric model. He then goes on to argue that what the model demands is completely unacceptable.
I don't disagree with you about the great importance of predictive ability. It is important. But you're overstating its importance in your failure to realize that predictive ability is nebulous and real science never throws out a theory simply because it has failed in a prediction. It will throw out a theory if repeated observations are contrary to the theory's fundamentals. There is a subtle interplay between, for example, predictive ability, elegance, ease-of-use, and cultural acceptability.
Ptolemaic astronomy is everyone's whipping boy for bad science. But that's only because most people are quite ignorant of both Ptolemaic astronomy and the messy reality of historical and contempory science.
One thing I learned from my education was that most oft-ridiculed (or smugly derided) historically "wrong" scientific theories were just not that bad. More to the point, I don't have a lot of reason to believe that contemporary theories are as good as is popularly supposed, either. Is the scientific rigor of the social sciences woefully inadequate when compared to that of the hard sciences? You'll get no argument from me about that.
But the idealization of science and the hard sciences in some ways makes it that more difficult for the social sciences to improve their rigor since it's unrealistically measuring them according to a standard that would find all fields badly wanting.
I think Jane's comments regarding the stars were a bit misguided. However, the problem with the Ptolemaic model wasn't that it was wrong, it was that no one understood the underlying mechanics. Someone just came up with a model that appeared to fit the data, but there was no understanding of why the planets behaved the way they did.
Copernicus was better, but not much. Copernicus proposed a different, simpler model. But he had no understanding of the underlying mechanics, either, and of course his model didn't fit observed behavior, either. Then Kepler came along. With the help of Tycho Brahe's precise measurements, Kepler concluded that planetary orbits were ellipses. Unfortunately, although Kepler's contributions were large, they didn't lead to understanding. They just led to an ability to predict quite a bit better.
Newton put substance into the predictive model by determining a law of gravitational attraction that fit observed behavior and demanded conic-section orbits. Newton's contribution allowed prediction of hyperbolic and parabolic trajectories as well as elliptic ones. It wasn't until Einstein, though, that a mechanical model was developed that fit all the observed data.
Lacking understanding of economic models, I'm going to resist any temptation to draw parallels.
I highly recommend Timothy Ferris' Coming of Age In The Milky Way to those who are interested in our evolving worldview. It's a fascinating read.
Keith makes a good point about predictive ability. The Ptolemaic models didn't predict precisely, they had relatively good long-term predictive ability.
That said, I'm going to shut up now. I think I just crossed the line between what I know and spectulation.
I'll go back to the book when I get the time, but the cycles had to be adjusted three or four times, according to Smith, because they were measurably off.
I'm not dissing the Greeks. Their model worked very well, given what they knew. Which is precisely the point -- you can have a beautiful model that explains all the known facts, and it can still be completely wrong. I know Copernicus got it from the Greeks, and it's not an argument about superiority -- just testing. What copernicus did was not give us a perfect model, but give us one that was perfectible.
Jane, you wrote:
"...but the cycles had to be adjusted three or four times, according to Smith, because they were measurably off."
Yes, I'm not disputing that. I'm disputing the significance you're attributing to it. My point is that I don't know any powerfully predictive physical theories (modern or otherwise) that did not have to be similarly modified because observation did not agree with them. There's a big difference between finding a theory is basically correct but in need of tuning, and finding that an observation is directly and fundamentally contrary to it. You're giving the impression that in the case of Ptolemaic astronomy this is what happened, but that's not correct.
"What copernicus did was not give us a perfect model, but give us one that was perfectible."
But he didn't, as I'm about to say.
And:
"...model that explains all the known facts, and it can still be completely wrong."
This "completely wrong" thing disturbs me. I feel like you're asserting the possibility of some complete description of a physical system that is "completely correct" and that seems to me to be both philosophically and practically naive. (Someone explained to me your moniker today and so the former doesn't surprise me. I'll set that issue aside, then.)
Looking again at the motions of the planets, as a practical matter we've never had, nor are we likely to have, a description that is universally correct. As I've said, a simple Newtonian description of this system is beyond us; it is and will likely forever remain incalculable. Thus, even in the case of Newtonian physics you cannot assert that this description of this system is "completely correct" — not just because we don't know that it is, but because, apparently, we can't know that it is. And the physics that we do believe correctly describes this system, General Relativity, is even more limited in the possibility that it be used to describe the entire system.
Now, I don't deny the existence of reality. But since I am very skeptical, as least as a matter of practice, of the ability of even the simplest and most limited physical systems to be described "totally" and inerrantly, I'm equally skeptical of the idea that a reasonably descriptive and predictive theory could nevertheless be "completely wrong". It seems to me such a theory clearly is not completely wrong. If you say otherwise, you are implicitly judging it by a standard of "completely right", which is, I assert, unobtainable. You're being very idealistic, in the Platonic sense.
This matters greatly in the context of science and this discussion because science's successes have occurred precisely when we stopped being Platonists, and became pragmatists. When we stopped making the Perfect be the enemy of the Good, we started making progress.
David Perron above says that
"Newton put substance into the predictive model by determining a law of gravitational attraction that fit observed behavior and demanded conic-section orbits."
...and it seems to me that he's at least implying that Newton added something qualitative to an understanding that heretofore lacked any real comprehension of what was "really" going on. But this is exactly backwards. Newton's gravitation was entirely a formalism. It did not signify anything more than its mathematical self and its descriptive utility. He did not concern himself with the mechanism of gravity, of what it "really" is. He just described it.
This was only the most dramatic step in a long journey of increasing formalism. The Greeks didn't make much headway with their science because they were caught up in a search for Truth — for the real, qualitative comprehension of the cosmos.
The key to modern science's triumph is just how much it disregards measuring something against an independent standard of "truth". Instead, it concerns itself with "more or less descriptive", "more or less useful", "more or less predictive". Almost all scientists are still psychologically Platonists, chasing that cosmic revelatory truth is why a lot of them are in business. (Although many are just in the business because they like the work.) But they don't conduct their professional lives according to this.
Jane says:
If the only person who can get the results you get is you, that weakens your theory to useless in the eyes of other people.
True enough. But that does not render such a theory useless in all cases. Predictive models usable by only a single person can be of enormous value in some circumstances. If someone has a model for predicting the stock market and the model works then it may well work only if no one else tries to test and use the model. The accuracy and therefore the value of the predictive model in this case only exists if it is used by a single person. The effect of multiple people acting on its predictions will undermine the accuracy of the predictions.
That's true, and it's one of the most interesting problems in financial theory, IMHO. But the difference is that someone theoretically could repeat the process, probably without degrading the predictive ability of your model (unless the market is awfully illiquid). If I write a brilliant paper on the sociology of my family, it may be brilliant, but it's going to be hard for anyone to reproduce being my parents' eldest daughter.
Keith:
I'm well aware of what Newton did and what it represented. You have something against formalisms? I'm thinking that an understanding the mechanism by which gravity works was an extremely valuable distinction. And one that hasn't really been sharpened much in the last three centuries or so, other than some relativistic extensions. Gravity is nearly as much an unknown now as it was then.
"mechanism by which gravity works" should be struck and replaced with what I really meant to say, which is something having to do with inverse-square laws and conic sections.
"mechanism by which gravity works" should be struck and replaced with what I really meant to say, which is something having to do with inverse-square laws and conic sections.
Comments are Closed.