Mathematical Hubris, or Simply an Author Who Doesn’t Understand Statistics?

My brother, who works in urban planning, called and told me I should read the article “X And The City” in the current issue of Smithsonian Magazine.

I did, and was quite disappointed. Here are my thoughts:


Hi, <brother>,
Thanks for pointing out to me the article on urban math in the current, May 2013 Smithsonian.
I was fully expecting to be quite enlightened and entertained, as I am by most Smithsonian articles, but I have to report that I was quite disappointed by it, and thought that the author was being naive. A lot of the conclusions seemed to me to be contrived or invalid.
I felt strongly enough to write this letter to compose my thoughts carefully.
The author does write, correctly, that “Cities are particular: You would never mistake a favela in Rio for downtown LA” and that many large cities will be surrounded by what they call ‘slums’ and others call shantytowns — developments put up informally and outside of any bureaucratic or official network of laws or public services of any sort at all, and which often exhibit a lot of negative behaviors and outcomes for their residents as a result.
Some of those bad effects are lack of public schools, no safe drinking water, no urban sanitation system, no safe and corruption-free police system, no public health facilities of any sort, no safe and reliable and dependable transportation system, no reliable electrical or postal delivery system, no zoning or building safety regulations that mean anything, and no real defense of the private property or land-ownership rights for those who have only small amounts of them. (Defense of private property only goes to the very rich and powerful. Matthew, you know.)
As the article correctly notes, we have very little dependable data on most of those shanty-town dwellers: neither how many people nor how much they earn or spend in the underground economy. A recent article I read indicates that cash payments are quite common in the US as well; so much of what is written about GDP per capita, anywhere in the world, is guesswork at best.
You know the saying about computer systems: Garbage IN, Garbage OUT.
Maybe I missed something, but I don’t see anything in this article that would allow any individual or group to use any of this data to do anything that would concretely help anyone in any significant manner.
One specific quibble concerns building heights. “…the equation H=134+0.5G where H is the height of the tallest building in meters and G is the Gross Regional Product in billions of dollars”… but previously, the author says the relationship isn’t strong. Well, how strong is it? It’s not clear at all. This page shows very different metrics, and rather different conclusions.  When I look up the so-called ‘Zipf Law’ I find that a number of people think it’s a trivial and unimportant correlation that one will find in almost any distribution of random-sized objects.
I did amuse myself by making this log-log graph with standings of population sizes of cities in the US.
power law US cities population
This reminds me very much of graphs I used to see at the Naval Research Lab’s gamma-ray astronomy section, where they would have the logs of the energy of various gamma=rays that hit the Compton gamma-ray telescope on the left hand side, and the count of how many such photons on the bottom axis. Only the graph of the line of best fit went up to the right, not down to the right.
And those gamma rays were just about random, coming from anywhere in the universe. Weaker ones were much more common than strong ones.
A couple of minutes of work on a table from Wikipedia giving the masses of the largest solar system bodies and I get this graph:
power law largest 60 solar system bodies
Also apparently works for usage of words in ordinary language. Some are used very very often (like “the”, “is”, “of” and so on) and others hardly ever (“disestablishment”, “cornucopia”, “prolix”). You get a power law distribution. Not so special with cities, then.
I like the part where one of the people interviewed said “it’s just a coincidence” about correlation angle of sunrise and street numbers….
Only 50-60 years ago, New York City had the world’s tallest buildings, and had the record for a long time. Until a few city promoters/corporate idiots in Dubai and Kuala Lumpur and Hong Kong decided they would blow a few billions of dollars in building essentially useless tallest-in-the-world status symbols.
That being said, the data on heartbeats of and life expectancies of mammals do apparently fit a nice logarithmic line. That’s real data that anyone can measure — but now that I think about it, animal life expectancies very much depend on conditions, and critters living in zoos or labs are quite different from those in the wild… so I wonder how good even this data is… And are those resting heartbeats, or what?
In another case discussed in the article on making a decision whether to drive or take the subway to a Yankees or Mets game in NYC, or just to go home and watch it on TV if the traffic is bad enough as measured by Twitter or GPS on cell phones — I am skeptical, though I know that our smartphones have traffic-reading capabilities that do a fair, but not perfect job of showing you why you are stuck in traffic.  Seems to me that the decision on whether to go to a major-league sporting event is only partly based on traffic, and a lot on other value judgements that are not even considered.
Anybody who lets the folks in charge of technology decide for you how to spend an afternoon and evening needs to think again.
Published in: on April 28, 2013 at 9:11 pm  Leave a Comment  
Tags: ,

The URI to TrackBack this entry is:

RSS feed for comments on this post.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: