Models: Gotchas

Science is about making models. But models involve inclusion and exclusion. In the process, a particular model may leave out something vitally important. Then you’re in trouble.

One interesting example of this comes from geology. A hundred years ago, geological models left out all effects from living organisms. Geologists felt that the influence of life-forms was simply too small to have any noticeable impact. The biosphere was restricted to a very small slice of the planet, from a short distance up into the atmosphere to a short distance down into the crust. Life just didn’t make much difference to deep geological processes.

Oops.

We now know that life can have huge impacts on the planet. You’re probably thinking about human-made climate change, but that’s small potatoes compared to the Great Oxygenation event. 2.4 billion years ago, the rise of photosynthesizing algae completely changed the composition of Earth’s atmosphere, taking it from about 3% oxygen to our current 21%. Amongst many other effects, this oxygenation basically rusted all the iron exposed on Earth’s surface. We can tell all this from banded iron formations formed around that time. Before the event, there was plenty of raw iron in surface rocks. Afterward, you could only find iron oxides.

That’s just one example of the potential problems with models. Another example is the financial crisis that started around 2007. Economic models of the time simply ignored the possibility that banks and bank-like institutions (like hedge funds) might universally act like ass-hats: taking crazy risks and using dodgy investment vehicles to squeeze money out of the public, on the assumption that if everything blew up, world governments would bail them out.

Oops again.

It’s easy to say, “I’d be smarter than that,” but one of the basic principles of Buddhism is that we aren’t. The Buddhist claim is that we construct deluded models of ourselves. We say, “I’m this type of person,” or, “I always do this,” when the truth is that we change from moment to moment. We’re different around our parents than we are with our friends; we’re different at work than we are at home; we’re different when it’s sunny than when it’s raining. We can be furious one minute, then laughing the next. We may have general tendencies, but even those tendencies change with time and circumstance.

The Buddhist word for this is anatta: no permanent self. Whatever you think you are, you aren’t like that all the time. Any self-image you have is incomplete, and often dead wrong.

Ideally, you should give up trying to characterize your self and thinking of your self as a single unified thing. Instead, just try to be aware of what you are from moment to moment. Such awareness takes a ton of practice; it’s the reason that Buddhists meditate.

Eventually, you’ll recognize that you really don’t stay the same, not even over short periods of time. But that’s okay. Nothing stays the same. Be kind to yourself and others, and don’t try to grasp at any particular identity. It won’t work and it’ll just make you miserable.

[Picture of banded iron formation at Dales Gorge by Graeme Churchard from Bristol, UK, Uploaded by PDTillman) [CC BY 2.0 (https://creativecommons.org/licenses/by/2.0)%5D, via Wikimedia Commons]

Models: Multiplicity

In a previous post, I talked about science being all about making models. You observe a lot of phenomena, then you try to make a model that represents your observations. By creating a model, you make a generalization that (you hope) will apply to things you haven’t seen as well as the things you have.

But there’s a huge caveat that applies here: sometimes different models can be used to represent the same phenomena.

Most famously, light can be modeled as a wave or a particle. (Light is also modeled as a ray in Geometrical Optics.) It’s important to stress that these are models. We’re sometimes sloppy and say that light is a wave or a particle, but that’s going too far. Light is light. Waves and particles are models that help us predict how light will behave, and although they’re excellent models, they’re abstractions. We can’t say they’re real.

Another famous example of models are the different ways to represent the solar system, specifically the Copernican and Ptolemaic models. It’s well known that the Ptolemaic system used to fit observational data better than the Copernican model did, at least to begin with. Ptolemy’s system of multiple spheres had so many fudge factors that it could be adjusted to match reality pretty closely, whereas Copernicus had problems because he tried to use circular orbits instead of ellipses. But in the long run, the Copernican model was modified to become more accurate, and it “won” because it was much much simpler than Ptolemy’s spheres.

As another example, think of maps. Maps are models: abstractions of actual landscapes. We have road maps, topographical maps, numerous kinds of geological maps, and much more. Each can be based on the same terrain; the difference depends on what you choose to include and exclude.

Let me emphasize exclusion. The whole point of a map is that you leave things out for the sake of simplicity. Maps only show a tiny subset of what’s actually on the ground. They may also exaggerate the size of some geographical objects so they’re easier to see; a road map, for example, shows roads much wider than they would be if they were actually drawn to scale. We might say that maps are deliberately wrong—they deliberately hide some things and distort others in order to make certain information more comprehensible.

The same is true of economic models. The actual economy is hopelessly complex; it consists of a huge number of transactions between people, companies, governments, and other organizations. No model could possibly capture so much complexity. As a result, economic models make enormous simplifications—they ignore almost everything that actually happens.

We all know how that can lead to problems. Different economic models arise from ignoring different things, and what you ignore may be precisely what bites you in the ass during a financial crisis.

But my favorite example of multiplicity in models is what we see in role-playing games. Every RPG contains a system for representing characters: often a list of numbers and abilities aimed at modeling human beings (or human-like entities). Different games use different models…and while some game systems are moderately similar to one another, others are wildly divergent.

Even more interestingly, slight differences in models can lead to substantially different gaming experiences. The Call of Cthulhu character model, for example, is pretty close to a lot of other models, except for a single number: a ranking of your sanity. That SAN rating takes on an overwhelming importance as you play the game. Sanity considerations can affect every action taken by individuals and by entire groups. It gives the game a much different ambiance from games that might otherwise be similar.

My point is that models are chosen, and often by selectively omitting or exaggerating details. Models often impose and reinforce a view of what is and isn’t important. This has consequences…and in the next installment of this series, I’ll take a look at what those might be.

[Picture of Cthulhu by Alexander Liptak. Image used with permission under Creative Commons repository. Attribution 3.0 Unported licence.]

Sharing: June 6, 2018

Things I’ve enjoyed recently.

Article: There Are No Laws of Physics. There’s Only the Landscape
A good introduction to the concept of “the landscape” in modern physics, and why it has physicists both excited and disappointed.
Book: Gothicka by Victoria Nelson
A survey of recent developments in Gothic fiction. To a first approximation, Gothic fiction used to be synonymous with supernatural horror, but in recent years, not so much. Classic monsters like vampires and werewolves are more likely to be heroes than villains these days, as in the entire genre of urban fantasy. Why did this happen and what does it mean? I don’t agree with Nelson on numerous points, and she gets a few specifics wrong (especially when it comes to comic books), but there’s lots of food for thought.
Computer Game: The Witcher 3: The Wild Hunt (I recommend the Game of the Year edition containing all the DLC)
It took me a while to get around to The Witcher 3, partly because I got super-annoyed with technical glitches in The Witcher 1 and partly because the creators said ill-informed things when taken to task for the game’s lack of diversity. But as Anita Sarkeesian often says in Feminist Frequency videos, it’s possible to both enjoy a game and be aware of its problematic aspects.
The Witcher 3 is too male-gaze-y and lacks people of color, yet it’s the most inventive computer role-playing game I’ve ever played. It has many great story arcs, long and short, great game-play, and plenty of surprises. Over and over, I found myself encountering things I’d never seen in any other game…and even though it’s now several years old, nothing since has ever come close to its level of variety and story-telling. After more than 200 hours of play, I’ve started it again from the beginning and am still enjoying it a lot.

Models: Why They’re Good

The Buddhists in Love article that I linked to yesterday has got me thinking about models. So allow me to pontificate a bit.

During my first term at university, I came to the realization that science is about creating models. This idea struck me during Economics 101. It was a strange class—unlike most Econ 101 classes I’ve ever heard about. The professor had written a book in which he tried to distill the low-level principles of microeconomics into very simple definitions and axioms about preference: an Economics version of Russell and Whitehead’s Principia Mathematica. Ultimately, he hoped to derive all of microeconomics from these elementary propositions, just as Russell and Whitehead derived arithmetic and set theory from symbolic logic.

I don’t think the professor ever succeeded. If he had, he would have become famous, at least in Economics circles. And frankly most of the class was baffled. What did these weird little formulas about transitivity of preference have to do with running a business or managing inflation?

I was baffled myself, until I realized that he was trying to make an abstract model of thought processes that we usually take for granted. He wanted to state explicitly the principles underlying how a person makes choices. He invented a symbolic notation for preference, indifference, etc., with the hope that once he wrote down the obvious in an abstract form, he could start manipulating the symbols and discover ideas no one had ever noticed.

This kind of process happens all the time in pure mathematics, dating back to Euclid or before. It’s also what Newton brought to physics in the other Principia Mathematica: first, you use math to model physical processes, then you play with the math to learn new things and to see how different phenomena are secretly related.

In other words, you use math as a model for real world things. Typically, you start with very simple models (for example, ones that ignore factors like friction and air resistance), then you make the models more sophisticated so that they can deal with more complex phenomena.

But scientific models don’t have to be purely mathematical. Biology, for example, often makes use of the kind of models you see in the Wikipedia entry for Mallard Ducks. The entry contains such information as a mallard’s average size, how many eggs a female lays each year, usual habitat, and so on. Such a description constitutes a model: what a typical mallard is like. It’s an abstraction, based on observing a lot of mallards. It isn’t true for every mallard ever, but it gives you a good mental picture that’s reliable most of the time.

Other sciences use other types of models. Social sciences often use statistics and graphs. Some sciences use case studies; for example, an observer goes to live with a group of people for a while, then writes down a description of what their lives are like. This description is another type of model: an abstraction from real life.

My point is that collecting specific data may be part of scientific activity, but what science actually aims toward is production of a model, a summary, an abstraction: getting beyond individual specifics to derive something with wider applicability.

Often this is a good thing. We all know what good things science has given us. But there’s a downside too, and I’ll talk about that in the next post.

(Picture of mallards realized by Richard Bartz by using a Canon EF 70-300mm f/4-5.6 IS USM Lens [CC BY-SA 2.5 (https://creativecommons.org/licenses/by-sa/2.5)%5D, from Wikimedia Commons”)