17 October 2013


[Editor’s note: The following story in Wired contains several interactive elements. Please click on the link below for the full experience.]
WIRED partnered with Food Network and crunched 49,733 recipes and 906,539 comments from their massive website. The result is a fascinating overview of how Americans cook.
Math Proves Bacon Is a Miracle Food
1. Comparing the Biggest Food Fads
Food is so personal and subjective that we’re always talking about it in vague and imprecise ways. But one of the many amazing things you can with big-ish data is give precise questions to answers that always seemed so subjective. Take, for instance, the question of bacon. Everything is always better with bacon, right But if so, how much And are any foods actually worse with bacon

We calculated the answer, following a simple methodology that made the most of the 906,539 ratings on foodnetwork.com. First, we searched out all the recipes that fit a certain description-—sandwiches, for example. Then, we calculated the average rating for those foods if they did not include the word “bacon.” We ran the numbers again using only recipes that did include bacon. The results were pretty great. Of all the foods we analyzed, bacon lends the most improvement to sandwiches. Many other foods also benefitted. In fact, we found that when you crunch the data for all recipes, those with bacon do in fact rate higher.
No surprises here! There are plenty of reasons why sandwiches might benefit the most; their slapped-together construction allows the bacon to stay crispy. Things get a little more dicey in salads and vegetables, which can let the bacon get soggy. The only foods that get worse with bacon Pasta and desserts. An educated guess: It’s because bacon pastas are typically finicky cream sauces that are difficult to get right. And desserts often seem to render bacon fat into a congealed mess.
The bacon experiment led us down a whole branch of questions: Could we total how much chicken there is on the entire site What about how many testicles Could we figure out how many miles of spaghetti there are Yes, yes, and yes! (The answers are in the charts above.) But one thing we really wanted to know was: What foods are most popular now, and how has food popularity waxed and waned over time We looked at the rates of comments on eight faddish foods:
Big Food Fads, Over Time

Interest in these 8 foods has waxed and waned over time.
We calculated these by first finding the total number of reviews for each food. Then, we figured out what percentage of those reviews came in each quarterly period since 2007. (That arithmetic allowed us to normalize the data—-otherwise, this thing would be a huge bacon chart and everything would look tiny.) Perhaps the most surprising thing is how much the answers conform to anecdotal evidence from pop culture. Low-carb diets and Portobello burgers were totally a mid-2000’s thing. And sure enough, their popularity was tanking by 2007. Similarly, if you live on the coasts, you’ve probably found more and more restaurants and haute grocery stores touting quinoa. The trend is very recent. Bacon, though Bacon’s always been popular, though things have accelerated ever since it’s become a full-blown meme.
Are Alton Brown’s Recipes Any Good
2.  Celebrity Chef Smackdown
Here in America, we love to turn cooking into a sport. Cook-offs are part of our culture; the entire Iron Chef franchise was built on seeing “who reigns surpreme.” The problem is, we can’t taste the food from our living rooms. But what if we were able to see stats on every celebrity chef, just as we’d see stats for starting quarterbacks
That’s exactly what we did: We calculated the performance of each chef in Food Network’s pantheon of personalities, based on the ratings users gave their recipes. And we did this across six different major food groupings. It turns out, they’re all pretty damn good at cooking. The data was very clustered, so we constrained our chart to the range between 3.95 and 4.95 out of foodnetwork.com‘s 5-star ratings. Showing up on these charts at all means that the chefs are already doing pretty good work.
In fact, one of the most surprising things about foodnetwork.com’s data is that it doesn’t seem to have a tremendous negativity bias. In other words, you simply don’t get a lot of 1-star ratings. Instead, people mostly tend to vote with their enthusiasms:

The Stars Face Off
How 14 chefs fare across 6 major foods.
Of course, we didn’t stop there. Having access to so much data allowed us to answer some kookier questions. Because why not For example, we figured out which cities had the highest percentage of their comments devoted to a certain chef—-and thus, the white-hot center of that chef’s popularity. We also figured out what chefs do best with recipes over 2-hour cook times. Turns out, it’s all men; they’re the ones who tend to cook big hulking slabs of meat that need slow-roasting. As for who cooks the most chickens It’s the women—-who presumably serve a more female-friendly audience by cooking lighter dishes.
The Only Thanksgiving Guide You Want
3. Designing Your Thanksgiving Menu
Our idea for this entire package came from a simple problem: How do you find the best recipe for whatever you’re trying to make Sure, you can go by rating. But that’s only part of the picture. What about how hard a dish is to prepare You might find two Bolognese recipes with similar ratings, but you’re more apt to make the one that takes just a couple hours of cooking time.
So, armed with foodnetwork.com’s 49,733 recipes and 906,539 ratings, we solved the problem–with a specific eye towards Thanksgiving. It’s a tough period for almost all of us: We’re cooking the most we’ll cook all year, but at the same time, we don’t cook any of those dishes regularly enough to know exactly what works. (How many times a year do you make cranberry dressing or roast a turkey)
Above is a series of charts that cut through the mass offoodnetwork.com’s hundreds of Thanksgiving recipes. We found just the highest rated. We charted see prep time on the x-axis and number of ingredients on the y-axis. Our working hypothesis for organizing the data like this was simple: Why make something complicated when you can get good results in less time, with less shopping Here’s an interactive version, for your viewing delight:
Easy Dishes for the Main Event
Thanksgiving dishes, arranged by prep time vs number of ingredients.
You can click out to any of the recipes you see listed. But the chart is telling you simply how much you have to invest in a dish, and how other dishes compare. And, if you’re gung-ho about maximizing your time, we’ve also provided a handy chart above showing exactly how to phase each of foodnetwork.com’s top dishes, so that you can get all of your cooking done in time for football.
4.  America’s Food Regions
We’re all raised with some vague intimation of America’s great food regions. And usually, we never get any greater understanding than the anecdotes about grandmama’s oyster stuffing or cousin Chet’s cornbread. But armed with Foodnetwork.com’s data, we were able to get a lot more geeky with it. We sifted through all the geo-located comments to create a slew of maps, showing where in the country people are interested in what foods.
The results hewed closely to stereotype-—while at the same time providing a finer point on the trends. Chili is hugely popular in Texas. Cornbread is less of a Deep-South thing; biscuits are more popular there. Bostonians love lobster—-as do people in Miami Beach.
About our math: We sifted Food Network‘s database for each food shown, then geolocated the comments. We divided the number of comments about each food by the total comments at that location. We then plotted those results two ways: In the maps of peak interest (above), and in the maps of regional interest (below).
The Topography of Food Interest
Charting where there’s peak interest in 6 awesome foods.
The heat map shows the intensity of food interest–expressed as a percentage of comments from any given city. To make this map topographical, we plotted the data points for all of the foods together. What you get is both a food-interest map, but also a map of regional intensity. This map blends population density with food interest, and gives you a better sense of where the most people are cooking these dishes.

<< Back