After my best chocolate chip cookie search post yielded 26 recipes in 24 hours, I knew I had too many cookie recipes to bake each and every one. So like any good geek, I averaged the recipes to make the best cookie recipe ever, or what I call a Mean Chocolate Chip Cookie. Get it? Mean? Ha ha ha.
To begin, I compared all the recipes, removing any duplicates. You’d be amazed how similar chocolate chip cookie recipes are. Then I further whittled down the list by removing those that called for non-traditional ingredients (New Hope Mills buckwheat pancake mix, almond butter) or appeared in books that I didn’t own (The King Arthur Flour Cookie Companion, Union Square Cafe Second Helpings). That left me with twelve distinct recipes, which I entered in an Excel spreadsheet (download the spreadsheet).
This experiment called for scientific precision.
Here’s where it got hard and I had to use math. I converted all the measurements to base 10 so I could enter decimals into my spreadsheet, e.g. 6 tablespoons of butter equals .75 sticks of butter. But it wasn’t enough to just average ingredients. I also needed to account for differences in the directions. Some recipes called for cold butter, others for melted. So I averaged technique as well, taking into account various oven temperatures and recommended dough chilling times.
If you’ve ever baked, you know how precise baking needs to be. The idea of averaging a recipe struck me as both amusing and insane, and I was pretty sure the resulting cookies would be terrible. After all my calculations, I baked a batch. I had to make a few tweaks, e.g. my oven didn’t have a setting for 354.17°F so I used 355°F. But I stayed true to the math as much as possible. I didn’t check on how the cookies were doing, but simply baked them for 13.04 minutes. (I got that .04 by hesitating just a moment before opening the door after my timer went off!) And what do you know?
Clockwise from top left: 1.33 eggs plus .33 egg yolk, mixing, cookie upskirt photo, dough ready to bake
These cookies were pretty damn good! I’d expected the worst. I’d expected they’d be inedible, or burnt, or floury and gooey at the same time. I had a hint they might not be too bad when I tasted the dough. But when I pulled them from the oven, I was amazed. The first bite revealed a cookie crispy around the rim, warm and chewy on the inside. A few hours later, they were firmer, but still tasty. The best chocolate chip cookies ever? I’m not sure, but I baked A Mean Chocolate Chip Cookie. And that’s enough for me.
A Mean Chocolate Chip Cookie
INGREDIENTS
2.04 cups all-purpose flour
0.79 tsp. salt
0.79 tsp. baking soda
0.805 stick unsalted butter, softened to room temperature
0.2737 stick unsalted butter, cold
0.5313 stick unsalted butter, melted
(1 US stick = 8 tablespoons = 1/4 lb.)
0.84 cups light brown sugar
0.10 cups dark brown sugar
0.54 cups white sugar
1.33 eggs
0.33 egg yolk
1.46 tsp. vanilla extract
0.17 tbsp. water
0.25 tbsp. milk
1.53 cups semi-sweet chocolate chips
DIRECTIONS
Pre-heat oven to 354.17°F, or as close as you can get.
Sift flour, salt, and baking soda together in medium bowl. Set aside.
Using a hand or stand mixer, cream butter and sugars until incorporated and smooth. Add vanilla, water, milk and eggs. Mix until all ingredients are combined. Add flour, salt, and baking soda and blend until fully incorporated. Stir in chocolate chips.
Cover and chill dough in the refrigerator for 25 minutes.
Place parchment paper on one-third of cookie sheet, drop dough by rounded tablespoons onto sheet. Some cookies will be on parchment, others off. Cook for 13.04 minutes.
Very cool!
Of course, the obvious next step is to average a cookbook…
0.07 Pork Tenderloins
0.3 Eggs
1.2 tsp water
1.14 boneless skinless chick thighs
Pan fry for 2.7 minutes, broil for 2, throw in slow cooker for 12 minutes, etc…
Without a doubt, my favorite chocolate chip cookies are from the cooks illustrated recipe. thick, chewy, rich, chocolatey, and not too sweet.
The only modification we make is to use 100% dark brown sugar, which really rounds out the flavor.
Meg, this is hilarious! A few comments…
First of all, I think to be fair you need convert everything to weights and equate the total amount of batter. The Cooks Illustrated recipe looks to be significantly larger than the Cookie Book recipe next to it, for example. Then you would average the proportions to get the mean recipe, which could be converted back to useful measures.
Second, Herve’ This, the French molecular gastronomer, talks about the sensitivity of recipes. The idea being that you define a space of all possible combinations of ingredients/techniques, figure out which combinations yield a successful result, and figure out how small a variation of the ideal recipe can be tolerated before you run into trouble. Mayonnaise is rather sensitive, for example, as are some pastries, but I wonder if cookies are fairly tolerant of modest changes.
Third, this is a case for Google Spreadsheets if I’ve ever seen one!
That King Arthur Book is very good (didn’t you used to live up here near Hanover or Norwich?) and a cookie recipe with buckwheat pancake mix is probably worth trying!
Hey, no fair! You didn’t ask “what your favorite recipe for chocolate chip cookies that appears in a book I own”! (You’re doing yourself a disservice by not trying the USC ones.)
I was just having a conversation with someone about how it’s really impossible to mess up certain combinations of ingredients. You can put them together in any way you like, and as long as you actually cook the whole mess, you can’t go wrong. You may not get what you intended, but it’ll still be good. Chocolate chip cookies fall into this category – butter, sugar, flour, eggs, chocolate… what could possibly go wrong?
First of all, I think to be fair you need convert everything to weights and equate the total amount of batter.
Yeah, I agree. I thought of that but that really was going to get complicated! Maybe once I recover my energies from this experiment I’ll give that a shot.
I think cookies, at least chocolate chip cookies, do seem to be tolerant of recipe changes, which is lucky for us!
Michael, I’m sure the King Arthur book is great, but I also wanted to be able to link to the recipes and so not having the book and not being able to link to the recipes kind of made that difficult. And I’m sure that buckwheat mix is worth trying, many of the recipes sounded amazing. Over time, I’ll probably go back and make some of them. But for the purposes of this experiment, outlier ingredients had to go.
Adam, fair point. If you want to post that USC recipe here, I’ll give it a try. I’m always game for more cookie recipes.
OK, I done it! (What are advanced technical degrees for if not for this?) Instead of averaging each ingredient separately, I calculated the weight of each ingredient in grams (thank you internets!), then used those weights to scale everything to give a recipe for exactly 1 kg of mean chocolate chip cookies. Here are the proportions (converted back to American baking units):
1.77 c flour
.66 t salt
.68 t baking soda
1.35 sticks butter
.72 c light brown sugar
.08 c dark brown sugar
.50 c white sugar
1.34 t vanilla extract
.14 T water
1.16 large eggs
.29 egg yolks
.19 T milk
1.34 c chocolate chips
That’s almost exactly the same proportions as what you got, but with a trifle less butter and a trifle more vanilla.
this is brilliant!
there’s no reason to change all the recipes so that they make the same amount of batter; the averaging takes care of that for you.
I have got to give some kudos to Harlan for using a reference to Google Spreadsheets.
Also, I have been following this for weeks now, trying out each recipe myself. I can not thank you enough for having the sense to putting these recipes in an excel document. That makes total sense.
Thanks so much Megnut!! (I’m off to make your new recipe)
Oh, I like how there is only 1/3 sheet of parchment paper 🙂
Paul,
If you don’t use a weighted average, a recipe that is much larger than the others, and that uses considerably different proportions, will disproportionately affect the final result.
I agree that Meg is brilliant, though! 🙂
Other nerdy things I’d like to do are to graph the proportions so you can see how much variation there actually is, and to use multidimensional scaling to look at how similar the recipes are to each other.
Somebody should email the Cooking for Engineers guy. He’d be very amused by this. 🙂
I love things that combine two passions! My inner math nerd and my inner chef love this recipe! 🙂
A recipe like this would be an absolute nightmare to a baker who performs bench chemistry by day. Measuring out “0.2737 stick unsalted butter, cold” with precision to the 4th significant digit using just basic kitchen tools would drive those OCD bench chemists completely mad.
The cookies certainly *look* great and I love that beaker photo. If you have a PO Box, I will bake you some of the New Hope Mills buckwheat mix cookies…
yeah, oops, i guess you’re right. sorry!
Here’s the cookie recipe with quantities multiplied by 1.26:
2-4/7 cups all-purpose flour
1 tsp. salt
1 tsp. baking soda
1 stick unsalted butter, softened to room temperature
1/3 stick unsalted butter, cold
2/3 stick unsalted butter, melted
(1 US stick = 8 tablespoons = 1/4 lb.)
1 cups light brown sugar
3/8 tsp. dark brown sugar
2/3 cups white sugar
1-2/3 eggs
2/5 egg yolk
1-5/6 tsp. vanilla extract
1/2 tsp. water
1 tsp. milk
its easier to read than decimals… hopefully it is still as good as the original!
Thanks for the recipe. Will try it this Sunday!
Thanks Alex, I especially like the 2/5 egg yolk. That was probably the trickiest part of measuring for the recipe, getting the partial egg and egg yolk measurements correct. An egg white is gooey and don’t like to separate into less than its whole.
how about metric?
USC cookies, paraphrased from the Union Square Cafe Second Helpings book, which has many fantastic recipes:
1 stick butter, soft
.5c sugar
.5c dark brown sugar
1 large egg
.5 tsp vanilla extract
1c AP flour
1.25 cups rolled oats
.25 tsp salt
.5 tsp baking powder
.5 tsp baking soda
6 oz semisweet chocolate (I’ve made it with more, or milk, or whatever – it works)
Preheat oven to 400F. Butter or parchment (or preferably silpat) two cookie sheets.Cream butter and both sugars together until “smooth and fluffy”. Add egg and vanilla and blend thoroughly (I add the vanilla to the egg first). Add everything else to the bowl except the chocolate. Mix on low to incorporate. Fold in the chocolate by hand with a rubber spatula. 12 cookies per sheet, 2tbsp each, flatten. Bake each sheet separately for 9 minutes “until the cookies have spread and puffed slightly, but have not browned”. Cool for 10 mins, then put on a rack.
Oh yeah, – these cookies also do very well rolled up in wax paper and stored in the freezer. Just cut off as many chunks as you need and lay them out on the baking sheet to come up to room temp while the oven is preheating.
Those flat-azz cookies are funny! Use shortening if you want to make great cookies. Butter just make them taste like nasty tube cookies from the store.
This is great, thanks for the roundup! I think I have a very good recipe, but there’s whole wheat flour.
Interesting that nobody mentions James Surowiecki’s book “The Wisdom of Crowds”, in particularly its opening chapter regarding Francis Galton’s experiment. For more, have a loot at wiki http://en.wikipedia.org/wiki/The_Wisdom_of_Crowds
this sounds good and yummy
Can you make a chart from the Excel sheet showing the variance for each ingredient? Would be interesting to see just how close they are – and what ingredients are most controversial – in a graphical form.
Adam, cool idea. Not sure if I’ll have time today but will see what I can do. I know that salt varies quite a bit, from 1/4 tsp to 1 tsp in each recipe. Would be interesting to see how it varies across recipes of the same yield. If anyone has time, feel free to grab the Excel file and go for it.
I found this by scanning the “Hot Now” list in del.icio.us before logging in, which I always do, cause you never know what you’ll find. I LOVE this, and have posted the link to a listserv I belong to. We’re all nerds, and many are also creatives, so they loved it! Thanks very much for doing this, and for posting the results; I look forward to following your progress!
Does it seem impractical to anyone else? I mean, I’m all about the geekery but I read the original post and have been looking forward to a kickass recipe since then…the odds that I’ll approach this one are pretty much zero.
Amy, it may be impractical, but it’s not anymore impractical than me baking 30+ batches of cookies! I just didn’t realize how many recipes I’d get, nor how much of time commitment such a project would take. The recipe is really pretty tasty, considering how it was conceived. I agree it would be nice to know the best of all I got, but that’s just not going to happen any time soon, unfortunately.
I’d like to see the final recipe in weights. Measuing all those oddball fractional cups is too tricky 🙂
I hereby declare the USC recipe the winner dy default. Romano emerges victorious from the long, drawn-out, buttery siege!
Wonderfully geeky. Maybe you should also publish the standard deviations to the recipe. I’d be interesting to see how much all the recipes varied.
Does a mean cookie mean mean flavor?
OK! I’ve uploaded Meg’s spreadsheet to a Google Docs spreadsheet, and added a couple of additional pages, converting amounts to grams and using weighted averages. There’s also a graph showing proportions.
http://spreadsheets.google.com/ccc?key=pSiojQUMOc62hM1qZUOyILQ
I’m not sure if people need to be invited in order to edit the spreadsheet, but if anyone else would like to work on it, just send me email. (I think my address should be available on the spreadsheet somewhere, but if not, post here and we’ll figure something out…)
I thought the exercise was to evaluate already known great chocolate chip cookie recipes? Greatness does not come from homogenity, so what you’ve made is an average cookie. Bah, Meg, BAH!
Southpark.Cartman.Sit(“Meg”);
Now if I can just find my .79 tsp. I know I had it yesterday…..
Jay, no one said it was a great cookie. I said it was a “mean” cookie, mean = average. And that it tasted pretty good for how it was created.
Oh, clever! Mean cookies.
OK, now I have to make some.
Just a word from a pro baker, you should convert all ingredients to weight. Volume measurements of dry ingredients are very inaccurate. For a small batch in the US, ounces would probably be best.
Then, what you *really* need to do is convert the recipes to a baker’s percentage – it’s a formula of ingredients that expresses ingredient weights as a ratio of the flour weight. What you’ll probably find is the all recipes have very similar ratios of butter, chips, salt, etc… to the flour weight. There are certain ranges for ingredients, and outside those ranges baked goods fall apart.
How about median chocolate chip cookies?
Wow. I’m going to have to congratulate you on how you came up with the name of the cookies. Maybe some Standard Deviation Oatmeal Cookies are in order at some point?
When comparing different recipes we normalized the ingredients to the total flour (sum of the differtent flour types, if more than one) contents. I used Harlans Google Spreadsheet (in gramm page) to do the averaging and then compared the weights to the ones you found:
http://spreadsheets.google.com/ccc?key=pAozGHh-p8AH9WFt_Kg5PDQ
The differences are not that big, except for the milk (10%) and the sugars (8% for brown, 6% for caster).
@harlan: The Google Spreadsheet you uploaded is read only.
Math is delicious! 😉
http://www.questionablecontent.net/view.php?comic=593
I do the same thing all the time! Though I usually don’t average the temperature or time. Sometimes I take the average, sometimes I take whatever is most common. It works out pretty well. I made a mean brioche the other day. 😛
Why remove the nearly identical recipes? It seems that to create a true mean you would need to have those in there as well. If you had a class of 20 students and asked what size TV they own and tried to determine the mean if 10 of the 20 said 32″ and all other sizes were different you wouldn’t just discard the identical sizes, they would be included for a more accurate mean. Discarding them and figuring it as if you only had 11 students in the class would be inaccurate.
Possibly, but I wasn’t trying to calculate the mean of all the recipes I received. I wanted to compare distinct recipes, and then I decided to average them.