The (un)Wisdom of Crowds

Does the Wisdom of Crowds work for elections? Should we think that the result of the British Brexit vote, because it was a free vote put to the people, was not only democracy in action, but also a wise method for a nation to make such a decision?

I've touched on this matter in my latest Bloomberg View column, drawing on a new study on group decision making by researchers from the Santa Fe Institute and the Max Planck Institute for Human development in Berlin. This study asks under what conditions we should expect larger crowds to make better decisions, and finds that, in general, this is only the case when the problems being faced are relatively easy -- so that any any individual has greater than 50% chance of getting the right answer. When problems are diffficult, the wisdom of crowds tends to fail, and small groups make better decisions.

Most importantly, they find that if the problems faced by a group come in an unpredictable mixture of easy and hard -- the more realistic case -- then the best decisions are made by groups of fairly small size, ranging from 10 to 40 or so. This insight doesn't apply directly to the UK referendum, which didn't necessarily have a right or wrong answer, but I think it does back up the view that a referendum is an extremely crude means for a nation to decide such as complex matter as whether to stay in the EU or not. As the researchers point out, many democratic decision making bodies around the world -- from juries to town councils to parliamentary committees -- make decisions with a fairly small number of people, usally from 5 to 40. There may be good wisdom in this.

And maybe it suggests that the best path forward for the UK is for parliament to weigh the decision to leave the EU using all its resources, and not being constrained in advance by the referendum result. Whether they ultimately decide to leave or to stay, that also would be democracy in action.

The Bloomberg thing is here.

Improve technology -- and still use more stuff overall??

A while back I had a brief argument with Paul Krugman and some other economists over economic growth and the future of the planet. It's common knowledge, of course, that human use of materials has grown over time -- globally, we now use more steel, plastic, glass, oil, water, etc. than ever before. We use more energy than ever before, and our agriculture puts more phosphorous and nitrogen into the oceans than ever before, and these trends toward more usage of physical stuff of every kind continue. All of this puts pressure on a variety of planetary processes, which we rely on, and threaten at our own peril.

As economies grow, I argued, they inevitably use more physical stuff, even if advanced economies do shift increasingly to providing services. Hence, economic growth (at least GDP growth as we presently know it) will need to be limited if we're to avoid over-burdening the planet, and to preserve its ability to support us in an acceptable environment.

The economists countered that my argument reflected a physical scientists' typical mis-understanding of economic growth. It needn't involve more "physical stuff," they countered, but could use less physical stuff over time as our technology makes us more efficient, and still generate ever more economic value. This is the idea of "de-coupling" between economic growth and physical materials use. I agreed with that point, in principle, but pointed out that -- for all economists' faith -- this hasn't happened so far, and we have no evidence that it will happen or even can happen. Economic growth is still closely associated with increasing consumption of physical materials and energy. Why should be think it won't be in future?

That was then, and we dropped the argument (although I just noticed that Tim Worstall at Forbes took one more swipe at me, and deeply mis-understood my point). But some new research offers an update on the story. It comes from some engineers who have looked at the best data they could find on technologies  and technological development over the past half century, and asked if these technologies -- which have generally improved at an exponential rate on many measures, including efficiency -- have led to a decreased use of materials and energy. Their paper is here, and the short answer they give is "no." I've written a Bloomberg View piece on the paper, and I'll just quote a short section:

Two engineers, Christopher Magee of the Massachusetts Institute of Technology and Tessaleno Devezas of the University of Beira Interior in Portugal, looked at two sets of data covering 116 different technologies existing between 1940 and 2010, ranging from the chemical industry and electronics to metals, wood and energy. Almost every technology over this period shows exponential improvement (though at different rates) in prices, performance and efficiency of energy and material use. Over 20 years up to 2009, for example, the price of photovoltaics consistently dropped about 10 percent per year.

The improvements weren’t enough, though, to outpace the combination of population growth, economic expansion and the rebound effect. As a result, overall material use tended to increase: Those photovoltaics, for example, consumed about 13 percent more materials each year.

To be sure, the data are far from perfect. Information on many of the 116 technologies exists over intervals of only one or two decades. Still, the fact that none of the data fit the usual story of decoupling suggests that the concept is at the very least highly questionable. The only six exceptions were technologies for producing substances such as asbestos, mercury and thallium -- all toxic materials that were ultimately controlled by policy intervention and legal restriction.

The results don’t imply that humans won’t ever achieve decoupling. They simply suggest that the historical record so far isn’t encouraging, and that there’s no reason to expect it to happen on its own.

I don't think this is the end of any argument; just more information to consider.

One final comment on the Forbes Worstall post, which I've just now seen. He suggests that I was arguing, here, that economic growth will eventually have to end because we will face "hard limits" to available energy. I'm not sure where he got that; it's not anywhere in my article. I don't think we're ever going to run out of energy, at least not for a very long time. I've even gone to some lengths to examine how much energy is available from solar sources (it's immense).

My argument was never that we will run out of energy, but that we will be forced, by deteriorating environmental conditions, to reduce and restrict how much energy we use. The energy we use always ends up in the environment, changing the environment, modifying its chemical makeup, the nature of its flows, and even its temperature. There's no getting around it; this is thermodynamics. And the effects, over time, are not small. Of course, somewhere along the way, even if we do restrict our energy use to some safe level, we might be able to eek out a bit more growth and extra GDP by improving energy efficiency, but that will come to an end too -- there are limits to efficiency. Once we've reached maximum efficiency in our technologies, we'll be limited in how much we can do.

Worstall suggests we might have another "13 millenia of exponential growth" before running into any problems, but this is a vast misunderstanding. See physicist Tom Murphy's famous post in which he estimates how long it will take continued exponential growth in energy usage -- along the trajectory we've seen the past few centuries -- to make the oceans boil due to waste heat. It's not 13 millenia, and not even close. It's 400 years. 

My point in all of this, of course, is not to predict with precision the situation we will face in this future point. I don't know any more about it than the economists do. The point is that, according to the current data, the rosy picture economists often paint about near term de-coupling look mostly like wishful thinking.

Economic growth -- vastly slower than we thought (maybe)

You'd think that by now we'd have a pretty good empirical understanding of how economies grow, i.e. what the normal pattern of growth is through time. We've been studying economies for a couple of centuries, and have had reasonably good numbers of the (crude) measure of GDP for half a century. In economics -- and among the financial media generally -- it's almost beyond question that the normal pattern of economic growth is exponential growth. Indeed, almost no one ever supposes it might be different; we only debate how fast or slow recent exponential growth has been.

But we might be wrong, especially for mature economies. That's  the conclusion of some recent research by a team of European economists and statisticians who looked at the data on 18 mature economies from 1960 onwards. They find that the best fit to the data isn't exponential at all, but linear, suggesting that if growth was ever exponential (in young economies), it isn't  like that any more.

I wrote a piece in Bloomberg about this a couple weeks ago. I wasn't aware of this line of work, but apparently a handful of (mostly) German economists have been pointing to this evidence for nearly twenty years.

It would hardly be surprising, of course, if human economies -- like individual people themselves, cells, bacterial colonies, trees and anything else alive -- turn out to have natural stages of growth, with fast growth eventually slowing toward something more gradual and, eventually, stopping altogether (which wouldn't imply the end of change, just some kind of balance). Current ideas in economics might need considerable re-thinking, of course.

But that wouldn't  surprise me either.

Shifting view

Hey, I've changed the title of this blog!

Why? Because I'm going to shift its perspective a little. As all of you will know, I've (mostly) stopped blogging here in the past 4-6 months. Reason? Because much of what interests me now has no immediate link to "finance," and so "physics of finance" doesn't seem quite right. I'd like to eliminate this psychological barrier (for myself). 

So, expect more posts, but maybe on different topics.

Cheers

Mark