Google and Culturomics: What 5 million books tell us about how we eat

Google has been digitizing books for years now, and a team of researchers just published some amazing findings based on the corpus so far -- over 5 million books.  That's 4% of all books.  Ever.  Printed.  You can read about the paper here.  The full paper is easy to read, and incredible (free with registration).  Google also released this tool to view the frequency of any n-gram since 1720.  (An n-gram is a set of characters separated by a space.  1-gram = technology, 1942, R2D2; 2-gram = yellow fever, John Wayne.  And so on.)   I wish they had tools like this when I was writing history papers back in college.

So what can we learn about health, and how we used to eat?  (Click on each title for a larger image.)

The Decline (and Rise) of Cooking with Animal Fats

tallow, lard, margarine, vegetable oil

  • The good guys: Tallow and lard see a steep decline in the mid-20th century.  Good news -- there is a noticeable up-tick after 2000.
  • The bad guys: Margarine is a war product.  The first spike is during WWI, and it spikes again during WWI.  And then there's a spike in the 50s.  Vegetable oil continues a slow and steady rise.

Vegans Are Taking Over Vegetarianism

vegan, vegetarian

 

Coffee Overtakes Tea as Stimulant of Choice

coffee, tea

 

Bacon Destroys Cupcakes

bacon, cupcakes

  • You can see the fat phobia decades from the 50s to now.  Thankfully, we appear to be coming to our senses (and are writing more books about bacon).

 

So what other stories can you find in the data?  Leave any good ones in the comments.

Updated Fri 6pm EST:

Comments

Now that is a cool tool.

Submitted by dbadbadger on Sat, 12/18/2010 - 09:58.

Now that is a cool tool. Close call between bacon and vegetarian but bacon still rules!!!

http://ngrams.googlelabs.com/

Submitted by Justin @paleonotes on Fri, 12/17/2010 - 19:18.

http://ngrams.googlelabs.com/graph?content=beef,chicken,pork,turkey,lamb...  beef, chicken, pork, turkey, lamb

saturated fat,

Submitted by Paul on Fri, 12/17/2010 - 18:53.

saturated fat, monounsaturated fat, polyunsaturated fat, saturated fat, trans fat produces some interesting trends, also "fat, carbohydrate, protein"

 Whoa!Try 'whole grains.'

Submitted by Richard Nikoley on Fri, 12/17/2010 - 17:37.

 Whoa!Try 'whole grains.'

Check out whole grains +

Submitted by John on Fri, 12/17/2010 - 18:54.

Check out whole grains + complex carbohydrates!

http://ngrams.googlelabs.com/graph?content=whole+grains,+complex+carbohy...

"Sprint" v. "jog" showed

Submitted by M Miller on Fri, 12/17/2010 - 17:30.

"Sprint" v. "jog" showed recent spikes in each, with sprinting always lagging a bit behind. "Sprinting" v. "jogging" shows a recent (sort of) plateau for jogging and a recent steady rise for sprinting, but with jogging always much higher.

Try adding butter to the

Submitted by My Paleo Life - Brock on Fri, 12/17/2010 - 17:12.

Try adding butter to the "tallow,lard,margarine,vegetable oil" search and, Wow, everything else is insignificant...Butter rules!!

 FUN!!!  I put in Mutton,

Submitted by Victoria on Fri, 12/17/2010 - 17:29.

 FUN!!!  I put in Mutton, Lamb, Beef, Pork, and Chicken...  There's a spike for the last three around WWI and WWII, and chicken gains around the time the lipid hypothesis becomes popular while all the other meats make no gains.  From 2000, however, meats have been on a steady climb, with the exception of chicken that appears to be starting a falling trend.