We are what we are studying

I find Andrew Gelman inscrutable when he talks about multilevel models and annoying when he talks about data visualization, but he’s always interesting.  Today, in a post on the anthropologist Marshall Sahlins,  he captured my aspirational worldview on matters political:

“What struck me is the relevance of this ‘anthropological’  mode of thinking to political science, where we must have understanding and sympathy for a wide spectrum of political opinions ranging from opposition to interracial marriage (supported by 46% of respondents in a recent poll of Mississippi Republican voters) to support for the nationalization of the means of production (still a popular position in many European countries, or so I’ve heard). As a political scientist studying public opinion, I have certain tools and academic experiences. But I am fundamentally the same kind of object as the people I am studying. It’s an obvious point but still worth remembering.”

As much as we divide ourselves into tribes, we’re all pretty much the same. Step back, take a breath, and remember that our variations are small and our commonality is big. There is no ‘other.’ The other is you.

Scripted interactions turn people into automatons

Scripts are a great way to learn. We get scripts from our parents and our peers. We repeat, internalize and refine. Kids with Asperger’s couldn’t function without scripts. But if you’re going to tell employees to read verbatim from a script, you might as well automate the whole process. Because rather than pay a person to use a computer, you’re paying him to be a computer.

Says David Heinemeier Hansson at 37Signals:

Most corporate customer service departments seem to have been reduced to call scripts of apologies with no power whatsoever to actually address the problems they encounter. That’s the conclusion I’m left with after dealing with three business bureaucracies this year: Comcast, Verizon, and American Airlines.

PhoneRobotThe only reason to talk to a  person is that a person can solve problems that a machine can’t. Generalized scripts are great: These are your options, this is what I can do, here are some alternatives. Then the general script gets translated into human-speak: open, immediate, smart, honest, and informal. It starts as a script and a person makes it human. And even if the problem can’t be solved to the customer’s satisfaction, most customers – not all, but most – will walk away from the call feeling better for having touched someone.


A New Yorker’s Rosetta Stone to Boston neighborhoods

When I moved to Boston a few months ago, I didn’t know anything about where to live. My friend Sam Knox over at CFO  offered to put Boston into terms that I could understand. Here is his guide.

Boston New York
Beacon Hill Murray Hill
South End West Village
Back Bay Upper East Side
Kenmore Square The upscale Village near NYU
Dorchester Bronx
Mattapan Harlem
Brighton Woodside Queens
Alston Woodside, Queens
South Boston Bay Ridge, Brooklyn. Archie and Edith would be right at home.  Now gentrified.
North End Little Italy.  Touristy Italian food.  No parking.  NO street crime.
Brookline Village, Upper West Side, Chelsea.  Coolidge Corner is wonderful.
Quincy Queens. Was Irish and FOB labor for shipyard.  Now many Chinese and Brazilians.  All reasonable. some very nice.


Where did I end up living? Well, I had to commute to Lexington out beyond 128 every day, so it wasn’t like I could live in Dorchester. I ended up in Arlington. Sam’s take:

Boston New York
Arlington A dull, cheap bedroom community. At least it’s not Belmont.
Belmont The upscale Arlington (now featured in Charles Murray’s “Coming Apart” – and not kindly)
Other suburbs Stay inside 128. Don’t ever go outside it.



Work for the machines? Not if I can help it

This article in Wired describes what happens when the QS (quantified self) movement meets workflow automation.

I’m about to launch a survey on marketing automation and this idea is very relevant. As in, if you’re not working for the machines now, you will be soon.

Emissary from SkyNetIn a way it’s more insidious than SkyNet – at least you can blow up SkyNet, or you could if it weren’t distributed.

We were talking about this last night at a professional meeting, though not exactly in these terms. You go to a website. A box pops up and asks if you found what you’re looking for. If you say no, it tries to get you to chat with customer service. If you say yes, it sends you to Amazon or Yelp and tries to get you to leave a positive review.

As a customer, you end up feeling manipulated, violated and turned off.  Just because you can automate something doesn’t mean you should. Especially when you’re talking to people with choices, people you don’t want to alienate, like customers.

If it’s employees that you’re manipulating, you’ve got more power. But that still doesn’t mean it’s a good idea. The effects may not be visible in the short term. Employees will grit their teeth and go through the motions. But when the revolution comes, you know where the guns will be aimed.

The power law of art schools

It’s amazing the number of prominent artists who taught at or passed through the Art Students League on West 57th Street. It’s not a particularly prestigious school. Anyone can take classes. My father did in the 1940s, when the Upper West Side was an Irish slum. But just look at this list: everyone from Frederic Remington, Winslow Homer and Norman Rockwell to Georgia O’Keefe, Louise Nevelson, Isamu Noguchi and Maurice Sendak. (I didn’t compile this, just counted it. The source is Wikipedia.)

The Art Students League has a big edge because it has been around since the 1880s. Lots of time to accumulate artists, especially early in their careers. It’s part of the New York art cluster (hello, Richard Florida). And anyone can take or teach a class. But what strikes me is how much the U.S. art world has not been centered on New York. You’ve got to go down to No. 8 to find another New York school – Parsons – and the next one, NYU’s well-endowed Tisch, doesn’t come up until No. 13.

Are Wikipedia entries the best way to gauge an artist’s prominence? Sure, as long as you don’t confuse prominence (which you can measure) with quality (which you can’t). The best thing about Wikipedia entries is that you can count them.

The Economist as the center of the magazine world

A few years ago – in 2006, to be exact – I wrote a scraper to crawl Amazon.com’s affinity links for The Economist. Think of affinity links as the basis for Amazon.com’s recommendation engine. They’re the links at the bottom of each page with headings like “People who subscribe to The Economist also subscribe to…” These links give you a recommendation: If you like The Economist, you’re also likely to be interested in, say, Foreign Affairs or The New Yorker rather than Guns & Ammo or Mother Earth News.

I wrote the spider in Perl (though since then I’ve moved on to Python, executing my scrapers on the great ScraperWiki site). Once I had the data, I put it into Pajek – a wonderful network visualization program out of Slovenia’s University of Ljubljana – and gave the resulting diagram to an artist over at The Economist.
People who read The Economist...
The board of directors over at The Economist loved this diagram because it showed their magazine as a bridge among high-end specialist publications. (Just avert your eyes from Wired, which has a similar claim.) It’s exactly what a sophisticated general interest newsweekly should be.

But much more came out of this exercise than a flattering diagram for The Economist. How is Martha Stewart Living connected to Soldier of Fortune? You’ll have to talk to me to find out. Or maybe dig a little through the older posts of this blog.

The party of the rich? They’re all rich.

The other day on a Dutch blog I saw an offhand statement about the Republican candidates being a bunch of plutocrats. It made me curious about who the rich presidents – and presidential candidates – really were.

An annual income of about $380,000 puts you in the 1% nationally. In Washington, where most of these guys live, it’s more: about $520,000 per year, says the New York Times.

Of course it’s dangerous to conflate net worth and income. But the net worth data is easier to get. A site called 24/7 Wall Street has gathered the data and adjusted every president’s net worth to its equivalent in 2010 dollars. (Because a number of early presidents made and lost fortunes, net worth is measured at its peak.) And a few Google searches yield the same information for recent presidential candidates like McCain, Kerry and Gore.

The result? Look at all the blue in the chart. Leaving aside the Virginia land barons like Washington and Jefferson, the Democrats have an edge in raw wealth. The richest Republican aside from Romney was Teddy Roosevelt, who wasn’t exactly a Tea Partier. Back then, the Democrats were corrupt and the Republicans were the reformers. The next richest Republican? Herbert Hoover, with about $70 million.

The 15 presidents not on the chart had a net worth of close to zero. Lincoln, Grant, Coolidge, Truman, Taft – all frugal civil servants.

In general, though, forget about the parties. The presidency is a rich man’s game – no matter what party you’re from.

Why is this so funny?

I love the 2×2 matrix: a scatter plot with a four-square grid imposed over it. But as a visual metaphor, the matrix is overused. It’s also laden with jargon. Each square gets its own catchy phrase, like “cash cows” or “problem children.”

Instead of revenue growth vs market share, try pineapples vs seedless grapes. I’ll take easy and tasty. From xkcd, courtesy the heroic and hilarious Barry Ritholtz.

The 2×2 matrix: Popularized by Bible salesman turned management consultant Bruce Henderson, founder of Boston Consulting Group.

Returns vs size of US university endowments

There’s a lot of talk in the world of university endowments about David Swenson’s “Yale Model.” It worked well from 1999 to 2009, yielding annual growth of almost 12%. In 2010 the value of Yale’s endowment fell off a cliff. After that horrific year, Yale dropped into the bottom half of its peers (as the scatter chart below shows).

Swenson’s insights were twofold. One, expand the definition of asset classes and diversify broadly across them. So it’s not just a mix of large cap, small cap and bonds; instead, spread your bets across the Wild West of real estate, private equity and hedge funds. Two, long-term investors shouldn’t embrace liquidity. They should avoid it. On average, the more illiquid the asset class, the higher the return. Nobody gets rich buying T-bills.

What went wrong in 2010? Couple of things. When markets go south, everyone runs for the exit at once. A balanced portfolio won’t help you when the correlation across asset classes approaches one. Second, as more money flows into alternative investments, it becomes harder for them to outperform. Hedge funds fell 9% last year when the S&P 500 was flat. The managers still do well – “Where are the customers’ yachts?” as the saying goes – but for many, their days are numbered.

Any system that outperforms the market will eventually be arbitraged back to the mean. That’s the ultimate problem with the Yale Model. In fact, that’s the problem with modern portfolio theory. Fortunately, it’s a very long-term problem. A friend of mine is busy scouting out companies in Cambodia for private equity investments. Want to diversify? Maybe it’s time to look in Cambodia.