Coke Engineers Its Orange Juice—With an Algorithm

Without further details from Coke, I believe you are absolutely correct. Oranges do not have constant flavor throughout the year, and the various species of oranges which mature at different points in the season have quite different flavors. For example, in Florida, Valencia oranges rippen after decemeber for a few months, and are probably the most delicious orange you will eat. This supply of oranges is not enough to last the whole year, so most of these oranges are squeezed, pasteurized and stored for later use. The rest of the year, you might squeeze naval orange (kinda shitty orange), early-mids, or other oranges. Even worse, oranges have varying flavor throughout the season. When Valencias first come into season, most are half ripe, or half green. Maybe only for a few weeks out of the year will you get that perfect tasting orange. And god help you if those oranges are oragnic, and half of them are covered in mold.

Rather than sell you an orange juice with varying flavor throughout the year, or even throughout the season, it is much less wasteful and much higher yield to blend the oranges together to produce some rather fine tasting juice. All of the companies receive these blends (and essentially the same blends), the only difference between the companies is in what they do with these blends. Some choose to produce a less bitter juice, or sweeter, or more pulp. etc.

Anyways, it's not clear how their algorithm is different from this advertisement.

J.Lyons and Co.[0] were doing this in 1951. LEO[1], probably the first business computer, was built to collate data from Lyons Tea Houses to calculate how many cakes to bake overnight amongst other things.

[0] http://en.wikipedia.org/wiki/J._Lyons_and_Co. [1] http://en.wikipedia.org/wiki/LEO_(computer)

is it really different than why people go to McDonalds or any other type of resturaunt? Because they expect a certain result which is why I can see Coke going to this extent. The just have many more variables to deal with than their soda products.

Perhaps they're keeping track of each atom individually.

Combinatorial explosion?

I was skeptical too, but all it takes is 60 bits, i.e. 60 "yes/no"-questions in their "black book". Doesn't sound too far-fetched. But, of course, I do hope they're not brute-forcing it.

See http://www.wolframalpha.com/input/?i=log2%28one+quintillion%....

I think bullshit is it.

The best way I could think of to generate lots of variables is a random forest classifier, which is an ensemble of (say) thousands of trees. Each tree constructs a lot of variables at each node, so maybe you'd have (stretching here) hundreds of thousands to perhaps millions of variables. But 1e6 << 1e18.

Maybe they're counting each pixel of some remote sensing image of orange groves? Bullshit again, that would have to be a 1e9 x 1e9 pixel image. I'd believe it from NGA (http://en.wikipedia.org/wiki/National_Geospatial-Intelligenc...), but not from Coke.

If there's a shred of truth to it, they're using some dodgy mechanism to increase the variable count.

BlackBook is a mixed-integer programming model implemented in OPL/CPLEX (used to be an ILOG product until IBM bought them).

So all this stuff about classifiers is off base. This isn't a predictive model -- it's a math program.

I don't know how they got that quintillion number. Could be that many "variations" instead. But the dimensions are things like time, varietals, locations, what you're gonna store/blend, that kind of thing. You end up with decision variables that are indexed in multiple dimensions and that creates the variable explosion they're alluding to.

CPLEX is awesome at handling huge amounts of decision variables so long as the problem stays mixed-integer, and that's what this is.

Is that the same as blood oranges they sell here in Florida? http://en.wikipedia.org/wiki/Blood_orange

Blood oranges are indeed great, and no doubt fresh squeezed blood oranges are wonderful.

But I have (and still occasionally) live in Italy and have been very frustrated and annoyed with Italian OJ. I have to vehemently disagree with the notion that any kind of Italian OJ, at least as sold in grocery stores in Italy, is even remotely as good as the US kind. Italian OJ is high-temperature pasteurized, depulpified, put into tetrapak cartons, and radiated. It tastes like a mixture of the orange juice you get in plastic bottles in US vending machines, plus Tang. The blood orange version strongly brings to mind Minute Maid Fruit Punch. And it's expensive!

The US OJ industry makes stuff which is far from fresh-squeezed, but it is absolutely light years ahead of European variations in taste, and I think the reason is simple: Americans drink a ton of it. It is one of the very few foods in the US which Europe simply cannot approach.

"That stuff is infinitely better than the bland crap you guys have over there."

Ohh, don't stop. Tell us more sweeping generalizations about our entire country, with its thousands of different orange juice producers (all of which you've sampled) and its vastly different terrains, climates and farming practices that all lead to the same bland crap. What do you have to say of German Engineering? Belgian Chocolate? French Wine? Italian lovers? Asians are smart and the US is fat and the Russians are alcoholics, amirite!?

Anyways, thanks for chiming in on the entirety of OJ produced in the US.

it weirds me out that not a single place in the US you can get fresh juice (save from one or two chains of fresh juice only). Where i came from, at least orange and lemonade you can get at pretty much 100% of restaurants.

yet, every glass of water comes with a thick slice of lemon dangling on the rim of the glass. Go figure.

I used to by Italian blood orange juice (it came in those soft-pak boxes like Almond/Soy milk) in the Tops supermarket in Williamsburg, Brooklyn.. So some places must import it.

I think Nassim Taleb called the idea of OJ being thought of as very healthy as a trick played on poor people.