“Data is ten times more powerful than algorithms.” – Peter Norvig1

The CDC says that today more than ⅓ of people are obese2. United Health predicts that 52% of Americans will be either diabetic or prediabetic by 20203. The interventions used to combat this crisis aren’t working. Meaningful large-scale data about how people eat in the real world is hard to come by: a ten thousand person study is often prohibitively expensive to run and the data is often collected using faulty after-the-fact questionnaires. Yet, it is that very data that we need to enable us to combat the crisis.

Today, we are happy to release the results of analyzing real-world eating of hundreds of thousands of people. And, to please your pixels, we’ve done it in infographic form. The data gives a never-before seen look into how people really eat.

The data was obtained from over 7.68 million food ratings of half-a-million foods by Eatery users from over 50 countries over a span of 5 months. As far as we know, this kind of data has never been available at this scale before. Did you know that San Francisco eats 4 times the amount of brussels sprouts as the rest of the US? Or that picking any specific diet (it doesn’t matter which) will, on average, improve your eating by roughly 20%? Or that poor eating is transmitted like a virus, with a transmission rate of 34.5% among friends?

We’ve been able to glean insight into how people think they eat, how they actually eat, where people eat, what they eat, when they eat, and with whom they eat. Each infographic tells a story regarding the effects of the who, what, when, where, and how on healthy eating. Click below to read them!


A quick note on the veracity of the data. We often get asked if crowd-sourced data can be trusted. We had a gut feel that the answer was “probably yes”.

Famously, one of the most accurate ways to guess the number of jellybeans in a jar is to average the guesses of everyone in the room4. The crowd-sourced method beats much more advanced algorithms. To test our hunch that the same applied in nutrition, we looked at the aggregate Eatery scores for all meals eaten in a city versus the published obesity level in that city5. It turns out there’s a strong correlation. Eatery data can accurately predict obesity levels of cities in the United States. That is, Eatery data strongly correlates with the healthiness of its users.

Furthermore, findings from the Eatery aligns with current scientific research. For example, the influence rate of food choices by friends matches closely with the obesity transmission rates6 described by Christakis and Fowler. Breakfast eating findings are also in line with research conducted on the effects of breakfast eating—that people who eat breakfast tend to eat smaller portions7,8,9,10,11 and healthier food throughout the day12,13. Additionally, as expected, controversial foods, such as coffee, diet soda, orange juice, and bacon are flagged with higher standard deviations from user ratings on the Eatery.


If you are affiliated with a University and would like to use our anonymized data for research, please contact sylvia [at] massivehealth [dot] com.


  1. Norvig, P. “The Unreasonable Effectiveness of Data.” Internet: http://www.youtube.com/watch?v=yvDCzhbjYWs, Oct 11 2011 [Apr 19 2012].
  2. “Nutrition, Physical Activity and Obesity.” Centers for Disease Control and Prevention. Internet:  http://www.cdc.gov/Features/ObesityAndKids/, Oct 17, 2011 [Apr 18, 2012].
  3. “The United States of Diabetes: New Report Shows Half the Country Could Have Diabetes or Prediabetes at a cost of $3.35 Trillion by 2020.” UnitedHealth Group.  Internet: http://www.unitedhealthgroup.com/newsroom/news.aspx?id=36df663f-f24d-443f-9250-9dfdc97cedc5, Nov 23, 2012 [Apr 18, 2012].
  4. Sunstein, CR.  Group Judgements: Deliberation, Statistical Means, and Information Markets.  U Chicago Law & Economics, Olin Working Paper No 219; U Chicago Public Law Working Paper No. 72. Aug 2004.
  5. Christakis NA, Fowler JH. The Spread of Obesity in a Large Social Network over 32 Years. N Engl J Med. 2007; 357-370-9.
  6. Clark CA, Gardiner J, McBurney MI, Anderson S, Weatherspoon LJ, Henry DN, Hord NG. Effects of breakfast meal composition on second meal metabolic responses in adults with type 2 diabetes mellitus. Eur J Clin Nutr. 2006;60:1122–9.
  7. Liljeberg HG, Akerberg AK, Bjorck IM. Effect of the glycemic index and content of indigestible carbohydrates of cereal-based breakfast meals on glucose tolerance at lunch in healthy subjects. Am J Clin Nutr. 1999;69:647–55.
  8. Nestler JE, Barlascini CO, Clore JN, Blackard WG. Absorption characteristic of breakfast determines insulin sensitivity and carbohydrate tolerance for lunch. Diabetes Care. 1988;11:755–60.
  9. Pai S, Ghugre PS, Udipi SA. Satiety from rice-based, wheat-based and rice-pulse combination preparations. Appetite. 2005;44:263–71.
  10. Pasman WJ, Blokdijk VM, Bertina FM, Hopman WP, Hendriks HF. Effect of two breakfasts, different in carbohydrate composition, on hunger and satiety and mood in healthy men. Int J Obes Relat Metab Disord. 2003;27:663–8.
  11. Isaksson H, Sundberg B, Åman P, Fredriksson H, Olsson J. Whole grain rye porridge breakfast improves satiety compared to refined wheat bread breakfast. Food Nutr Res. 2008; 52.
  12. Levine AS, Tallman JR, Grace MK, Parker SA, Billington CJ, Levitt MD. Effect of Breakfast Cereals on Short-term Food Intake. Am J Clin Nutr. 1989;50: 1303-7.