This explains a lot about the drug war in Mexico

on Mar 16, 2010
This is beyond parody, the Mexican Army spent 10 million dollars on plastic boxes attached to antennas —modern versions of divining rods— and used them to find drugs and weapons. Supposedly you could program them to find all sorts of things, with special cards, which invariably turned out to be nothing but pieces of paper. No wonder Ciudad Juarez turned into the most violent city in the world [update: there's a better version of this chart in the next post]after the army took control of it (chart made with ggplot).

As of April 20, 2009, the army had purchased 521 of the GT 200 detectors for just over $20,000 apiece, for a total cost of more than $10 million, according to Mexican government documents. Police agencies across Mexico have made additional purchases, records show.

“We’ve had success with it,” Capt. Jesús Héctor Larios Salazar, an officer with the Mexican Army’s antidrug unit in Culiacán, said recently. “It works with molecules. It functions with the energy of the body.”
 It's got molecules... There's more:
In Culiacán, a city in Sinaloa State where Mexican drug traffickers have a strong presence, the military showed off the GT 200 in December. Canvassing a residential neighborhood, soldiers walked up and down the street with a GT 200 waiting for the antenna to point toward a suspicious residence. There were no discoveries.

But the soldier trained to operate the detector walked by one of the army’s armored vehicles and the antenna swung quickly toward the high-caliber machine gun sticking out the top. He took several steps back and walked by again. The antenna pointed again toward the gun.

“See?” he said.

But in November, at a checkpoint on the highway leading from Mexico City to Monterrey, the same device pointed at a Volkswagen containing a man, a woman and a child. Soldiers surrounded the vehicle and a search was conducted for illegal drugs. But all they found was a bottle of Tylenol — evidence, the soldier operating the device said, of how sensitive the GT 200 was.
Since they were pretty expensive, I wonder if money from the Merida Initiative was used to buy this things.

Cluster analysis of what the world eats

on Mar 9, 2010
Keeping with the theme of the post below, I used a clustering algorithm to group the different countries according to what they eat. I simply played around with the number of clusters until I got something I thought resembled reality, so don't interpret this as an in-depth analysis.
  • Group 1: Lot's of coffee and offal. They don't like fish and fruits, but eat their veggies. Similar to Group 2.
  • Group 2: Lot's of calories per day, all kinds of meat, alcohol, milk, butter, potatoes and sugar. Vegetable oils are used for cooking. No beans here.
  • Group 3: Fish and rice (which coincidentally is what I had for dinner yesterday). They use coconut oil and don't like milk or cheese.
  • Group 4: They like bovine meat, fruits and sugar. They also use palm oil, and don't like veggies.
  • Group 5: Not many calories per day, lots of starchy roots, beans and pulses.
  • Group 6: Lot's of carbs (wheat and cereals). They eat their veggies and use olive oil and soybean oil.
The code is here and the data (taken from a database put together by the blog Canibais e Reis) is here.

Weird dietary habits in the US

on Mar 8, 2010
Using this database of food consumption data the blog Canibais e Reis kindly put together, I calculated all values for which the US was at least 2 standard deviations from the world average.

Here are the outliers in standard deviations from the world mean: 2.5 4.5 2.1 2.1 2.1 

Lots of sugar and Omega 6 fatty acids. Not very healthy.

In case you were wondering, the only weird dietary habit in Mexico is (ahem) a high consumption of beans, 3 standard deviations above the world average. Since I suspected maize was also an outlier but it wasn't included in the database, I downloaded the 2005 data from the FAO website and indeed, Mexico is also an outlier in maize consumption. I bet if soft drink consumption were tracked, Mexico would also be an outlier.
The code is here and the data is here