Scaling Food

April 2016

If you were dropped at a random point in America today, nearly all the food around you would be bad for you. […] The only people who eat what humans were actually designed to eat are a few Birkenstock-wearing weirdos in Berkeley. […] If people have to choose between something that’s cheap, heavily marketed, and appealing in the short term, and something that’s expensive, obscure, and appealing in the long term, which do you think most will choose?

- Paul Graham

When people perfect a process, they gain the ability to predict the future in a small way, which is a very comforting thing. People love being able to say, with confidence, that “if I have these raw materials, this equipment, and this amount of time, I can, if I so choose, turn those into this amount of some other thing. Or I can keep the raw materials and do something else with them entirely - it’s up to me! I have agency and control over my destiny!” If they don’t say that last part verbatim and out loud they’re probably just thinking it.

People also love to make even more powerful predictions about the future, which means increasing the number of zeros on the inputs and outputs of a process, especially if the outputs can be sold for more than the cost of the inputs. So the natural thing to do with a process is to scale it up.

What does it mean to scale a process? I’ll submit that it means to change the quantities involved but to keep all of the relevant aspects the same, with “relevant” and “the same” being the hardest parts to pin down. As you increase the number of zeros, it’s easy to start focusing on the wrong aspects of the end product as the ones to keep the same.

Food is one blatant example of a process that, as it scaled, became something completely different entirely. The aspects of food that are still the same at scale are “has calories” and “is bought by people” - everything else can be changed in order to scale its production.

But “is bought by people” is a terribly misleading aspect here. Ev Williams pointed out the similarity between junk food and junk links. We’re putting something in front of people that has superficial similarities to what they want, but fails the test miserably when it comes to the more important aspects. But if it gets eaten or clicked, the conclusion is that people want it, so more gets produced. In one sense this isn’t any more evil or premeditated than a plant evolving to look appealing to bees, but I think we can hold ourselves to a higher standard as a species given, again, the whole agency and control over our destiny thing.

What are the tradeoffs that are made in the name of scaling food? Three major ones come to mind:

Shelf life. If you’re going to produce a lot of stuff in one place, it needs to stay salable for a long time before being consumed. Some food stays edible on its own without much effort at all - just keeping it at a reasonable temperature is sufficient. For another class of items, freezing them, while less than ideal, works perfectly well and doesn’t change the end product substantially. Where things slip away is when preservatives are added. They may be perfectly safe but they also allow too much leeway in terms of the rest of the contents, and make it too easy to just start adding anything else to optimize the economics instead of the inherent value of the food.

Ease of preparation. Similar to shelf life, this one takes many forms. One form is when the end product already has all the thinking done (usually requiring lots of packaging) - the food is made the opposite of scratch. Another notable form is that the raw ingredients themselves start being optimized while growing to easily slot into an existing process (making french fries for example).

Physical work required. This is the work in all aspects of production and preparation. It might seem like an obviously good thing to reduce the physical labor required to produce a unit of food, but given that the entire definition of work is shifting out from under us, there’s an opportunity here to rethink which metrics are important to minimize or maximize.

Our first attempts at scaling food got us to the point where we could feed everyone, but with subpar quality. It’s time to go back and refactor so that we scale something that looks more like the original product and rethink what we’re measuring at each step of its production.