Friday, May 07, 2010

A Sciencimilitudinous Three-Fer

1. Naughty Economics
You know, there are lots of things I wish I wrote. Happens all the time. More rarely, I encounter stuff that I like to pretend some more sophisiticated, educated, practiced, knowledgable, motivated, and skilled version of myself could pull off--not just what they said, but I'd like to have said it that way too. Readers of this-here blog, and God bless both of you for sticking around this long, might have noticed my sometimes-uneasy fascination when it comes to contrasting things like macroeconomics and demographics against engineering and the physical sciences. Maybe the fact that I'm such a slapdash practitioner of the latter is why those former things appeal in the first place, but I also think, more generally, that quantification and understanding the relationships between variables are important skills for understanding aspects of the world beyond basic science, and I believe that there are patterns in those dynamics that either actually repeat, or that humans are especially adept at recognizing, and that these different schools of understanding should inform each other better than they appear to. And I love nothing better than torturing a scientific metaphor as well, there's always that.

The field of economics adopts phsyics math concepts rather consciously at times, and I've frequently thought I should be able to get some traction there, at least get it up to my shitty physics comprehension. But there's something that's always been niggling at me about the whole damn thing, and in all those posts (and as usual) I sidle up to a point repeatedly, and try to wrestle the language into representing what the hell it is exactly that's bugging me. And so I say that economics just doesn't support the precision it claims. I say it's a behavior model which in many cases fails to model behavior, one that tries to be a phenomenological representation, but that thinking about it that way will only get you so far. Also, I distrust financiers who purport to give me advice, when they own suits that are more expensive than my car. Fuck those guys.

Anyway, I found this paper via a commenter on Balloon Juice, who referenced another blog. You should read the article, it's awesome. I very much enjoyed the discussions of deductive reasoning vs. empiricism (both good things), and I also liked the just-enough-epistemology-to-be-dangerous, including the part acknowledging some continuity between natural sciences and the humanities, so that's why all my friends are engineers and lawyers.* Deirdre McCloskey details the sins of economics, including the virtues that are mistaken for sins (should economists apologize for the urge to quantify? for trying to understand via deductive reasoning? absolutely not!), serious sins, and ultimately the two dark, secret sins:

Economics in its most prestigious and academically published versions engages in two activities, qualitative theorems and statistical significance, which look like theorizing and observing, and have (apparently) the same tough math and tough statistics that actual theorizing and actual observing would have. But neither of them is what it claims to be.
Okay, the article is not perfect; there's always something to complain about. I don't how her Libertarianism has survived, why assumptions about fairness are supposed to be an ennobling (and weakly defended) virtue instead of a perhaps venial sin of naïvete (even if a motivation to fairness is all well and good). But she gives plenty of tools to address what's wrong with this outlook. It's a bit more implicit in her piece, but I might further emphasize a general failure to embrace induction--observing laws based on the data—maybe it's buried in her sin of institutional ignorance (not spending time in the places where economics actually happens), the first serious one not unique to the field, or maybe it's a part of the second secret sin.

[I don't know about her anthropology either. I have never suspected a tendency to exclude impersonal variables in favor of humanist ones, and my limited reading has certainly suggested the opposite. If Marvin Harris and E. O. Wilson (who, okay, is not really an anthropologist) are atypical, then I'm going to have to revise my opinion of the field! The poster at Language Log didn't think her closing comparison with linguistics was apt, either. Oh, and Euler's formula is damn useful in addition to being a particularly elegant expression of pure mathematical theory. Dissention noted.]

The second sin is a discussion of what we choose to measure and say is important, which I think is common everywhere, but when you're dealing with human variables, as you do in ecomics, what matters is inherently subjective. McCloskey tells us that the field has deviated unforgivably from this, and has, almost as a rule, confused statistical significance with either an essential truth, or worse, an essential importance. I appreciate the first argument more: that economic theory is based on pure deduction, heavily dependent on its description of starting conditions and its governing assumptions, both of which can be suspect, and which does not result in a quantitative, measurable result. That's an artifact of my naïve background, of course; my last ten years has been more about exploring the humanism.

None of which means that such an endeavor is not useful and important, just that we should understand what it is.

*that and the life of relative, though not impressive, privilege

2. Inherent Complexity
Dear Deirdre:
"I have to keep saying “pure” because of course it is entirely possible—indeed commonplace—for novelists, say, to take a scientific view of their subjects...Likewise scientists use elements of pure narration...or elements of pure make scientific arguments. I do not want to get entangled in the apparently hopeless task of solving what is known as the Demarcation Problem, discerning a line between science and other activities. It is doubtful such a line exists. The efforts of many intelligent philosophers of science appear to have gotten exactly nowhere in solving it. I am merely suggesting that a science like many other human practices...should be about the world, which means it should attend to the world.
It's not quite where she's going with the quote, but I'm taking a chance on using it as a segue. It's not easy to dispassionately observe the world even in the pure sciences; the fact that we're people is hard to escape. A recent Archdruid Report post was about complexity (I have no intention of latching onto the guy like a remora nibbling up post ideas--it's more that I made a long comment, the only reply to which leaked over here and wrung another long comment out of me. He's been getting the brain working.). He points out some contemporary versions of complexity, and relates it to Joseph Tainter's theories of societal collapse, in which the diminishing returns of bifurcated attempts at problem-solving doom the endeavor at some critical point. It sounds like a reasonable description of how humans behave when they run into problems, and sometimes diversifying the approach isn't going to be enough to overcome a deeper problem (simplification may be—can we branch from an earlier point?). I can't quite get myself to describe collapse is an inevitable consequence of complexity, however. And I think we need to be clear about what complexity actually means.

As I understand it (knowledge! I read a paper yesterday), the usual measure of complexity, pulled from the computational world, is defined as the length of the minimum necessary description of something. As a useful description of the natural universe, of course, this is fucking nuts. Anything can be "minimally described" by nothing more than "that thing." When I describe a pin, it's simple as can be, but I tend to neglect the innumerable quantum mechanical angels dancing around on its head. Nature gets more complex the deeper you look at it. Wheels within wheels, dude, and turtles all the way down. For that matter, some things that look complex are overestimated. A wave looks more complex than a gradient...until you do the appropriate transformation on it. I mean, "complexity" is so obviously and closely tied with our ability to describe something, that working on that definition as a natural phenomenon is a mighty philosophical head-scratcher. You can't have complexity without invoking the ability of humans to represent a system, right?

[And as far as the complexity of finance goes, it doesn't seem to follow Occam's suggestion of a minimum necessary description of trade and investment. That's something beyond what an inherent complexity might be: I think intellectual positions of power within society (whether financiers, academics, or priests) have a temptation to add complication in order to perpetuate their exalted and insulated role. (But don't worry, when I use jargon it's only because it's necessary.) Unnecessary complexity is another word for bullshit.]

And yet....

Complexity really seems like it could be a material property, a thermodynamic variable. We already have entropy, which--and this still blows my mind--has a statistics basis that is related to the information content of something. Maybe call entropy a measure of meaninglessness, or of equivalence. Complexity, if it can be cornered, has got to be something related. And statistical thermo also has some of the same issues that a description of complexity would, namely, how do we account for the all of the microstates (not to mention all the underlying nano-, pico-, femto-, and myrmecological attostates, and so on, all the way down)? The dodge is that any coarse-grained microstate is a superposition of a ridiculous number of quantum mechanical (atto?)states, which end up being equivalent enough for government funding.

I'm going past my own observations now. I'm not the first to head down this path, unsurprisingly, and now I'm specifically looking at some descriptions from a paper by paper by Rojdestvenski et al. (It was the first credible-looking source when I Googled the topic yesterday, and I mean, really, if I knew you could publish irresponsible speculations like this, I'd have quit this thankless paying job long ago. Maybe they were reading my blog? Nah, the dates don't work out.) Unfortunately, I can't link a pdf, but trust me that it introduces itself via a ride on some of my old bloggy hobby horses (shameless self-linking follows), such as admitting an early problem with completeness in describing the information content of something real, the peculiar situation of looking at coded things when we are the expressions of coded things, and entropy vs. evolution. It was a little validating, I admit.

They make a point that complexity, like the usual thermodynamic variables, is better thought of in terms of differential changes anyway (whether any third-law equivalent applies to complexity, who knows). The macro relations hold, in other words, and if complexity is a valid thermodynamic quantity, then this is how it must behave. If you think about things like internal energy, then the microstates aren't generally counted well there either. It's not that unusual, but it still feels weak on what complexity is.

I am a good deal less comfortable with their evasion of that question. They say that DNA, which has an inherent pattern, is independent of our observation, and is thus especially well-described by some (hypothetical) thermodynamic complexity. But doesn't everything have some level of organization? Surely any such material property must apply to non-living systems as well. It reminds me uncomfortably of the crank arguments that say the second law of thermodynamics proves creationism, and for that matter "complexity" is also a buzzword of intelligent design, so who knows what sort of search results I'll get on this one. Rojdestvenski has the opposite agenda, I suppose, and unlike creationists (or irresponsible economists), he (or she) is at least trying to refine a theory that describes evidence. The reason the authors wish to keep complexity in the realm of living things is because they are interested in forming a physics-based description of evolutionary processes, and they draw up some entertaining little cellular automata simuations to describe how evolutionary adaptions to deal with that energy balance might affect the development of populations. But still, even if evolution is unique to living things, I don't believe that complexity is.

I don't know enough about Tainter's ideas to know if they can work with evolutionary theory or not. For some extreme perturbations, the more complex individual (or species) will fail, even while complexity of the whole ecosystem appears to make life itself more robust, just giving us a bigger toolbox of accessible traits and adaptations to adjust to the big change. Evolution appears to increase complexity, but there also should be something to explain that the populations of complex organisms ain't got nothing on the simpler ones--even at this late date the bacteria are outnumbering us, possibly even pound for pound. The authors observe that in their highly limited simulation, highly variable conditions favor simpler organisms. That seems reasonable enough.

And it might also not be a bad shot at imagining how societies end. If the perturbations are big enough, then a more complex system will be less robust, and if I were clearer on what Tainter says, this would be a far better post. I'll go back to this eventually. Maybe in another couple years.

3. Pissed Epistemology
Okay, this is getting long, but since I've cited the thread already, here's a good example of something that's bothered me for a long time. The poster had a great start:
Mental constructions of gardens are much simpler to create than real ones, because putting a theoretical person on a plot of rhetorical soil is just the beginning...
What follows is a description of examples of the challenge of understanding a small agricultural ecosystem, much of which is ignored by industrial-scale farming, which, I'll bet you a lap in the nearest shit lagoon, is at our long-term peril. Gardening is a complex effort, even in the human-limited language of the comment. But clearly, gardening is technology. Hawk1eye starts advocating a better, deeper level of understanding, and the conclusion, to "chuck the idea that the best way to proceed is by working the industrial teeter-totter of 'empirical observations' leading to 'theoretical formulation'" just doesn't follow from those kinds of arguments. I mean seriously, if we want to entertain the idea of intuitively listening to the land instead of gaining an understanding and keeping track of what works and what's real, then let's start shitting in the wild, eating random berries, and take our best chance at surviving the first cold night. What, all of a sudden we have to observe and be careful? Well, no shit! So how should we know how much to observe, and how are you going to justify how much? If there's one bottom-line lesson I learned from nearly a decade of engineering edjumication is that intuition is something that's earned.

Here's my broader complaint: I hate it when people try to invoke reason to dismiss reason. People arguing something like intelligent design will often try to corner you in a rational-esque Encyclopedia Brown kind of gotcha moment. "Hey, it couldn't be that complex without thwarting thermodynamics, so therefore God made it six thousand years ago. Huh? Am I right?" Faith isn't necessarily such a bad thing, but when it's willfully pitted against science (which is really just an organized attempt to be descriptive), sometimes I wish they'd go full-on mystic and be done with it.

It's a similar phenomenon as demanding authority to regulate authority. Maybe it works when it's different authority with a different constituency, or if you're reasoning around different observations or assumptions, but you know, if that's what you're saying, then go ahead and say that. For a guy who doesn't know much philosophy, I realize that using rational argument to probe the nature of rationality (to uncover paradoxes and such) is valid enough, but that's a slightly different animal than using reason to prove non-reason. If arguments in general are not valid, then neither are the arguments making that case. (And maybe a contradiction like that can be derived from some advanced thinking--those knowledge paradoxes again—but those don’t really deny the thinking that got them there, and I don't think what we're talking about incorporates quite that much thought--on the practical human scale, it's well past the why-bother point.)

Describing our epistomologies and ontologies while using those same things is a formidible challenge, but making the effort to do so is both reasonable and important. Like Deirdre said, the asking of the why is no sin.


Cindy said...

Okay. I like your Book Reviews and your Miscellany. But this is taking blogging to a whole new level.

I'm not usually one to spend much time on serious stuff on my laptop. Well, not LONG serious stuff. I can do short, quick insights. But usually I'm in a hurry.

This post, however, has inspired (forced?) me to print the damn thing out in regular sized type, get my highlighter and go through it a few more times.

I'm using brain cells that thought they graduated and retired already.

"intuition is something that's earned" - that's the anchor, that's the lens. I can torture metaphors with the best of them. But this is a well-stated and much needed bit of wisdom.

So, you have altered the course of my otherwise lazy Mother's Day. In addition to the put-your-feet-up-and-relax part of the day I'm going to study.

I'll let you know what I come up with.

Hope you and yours enjoy the day. :)

Keifus said...

I'm really sorry Cindy. I had no intention of ruining anyone's weekend with this stuff.

Schmutzie said...

Whenever you make one of those remarks about "both" of your readers, I feel like I should say..."Present."

Keifus said...

I suppose there's a fine line between self-deprecation and self-flagellation. I've never been really sure what exactly I want from this blog. And there's a handful--we both know our little community.

Cindy: or "learned," possible I meant to write that, but the meaning's the same. The guy I called out replied over at that other place, and was decent enough to clarify. More mystic than I realized, and it didn't seem appropriate to offer up a rationalist statement of princples over there, after one comment already moderated away. (And no reason to doubt he's a fine gardener, but even if you abdicate any honest analysis of hows and whys, it's still a matter of "I observe this, so therefore I think that should happen." Drives me nuts.)