2022-03-14 6 min read


Notes, 2022-03-14.

In discussing my recent essay on the mortar and pestle, some Members on The Prepared’s Slack observed that cooking is an act of production, and kitchens are a type of workshop: places where material goods, sometimes of startling complexity and refinement, are produced with manual labor.

I started working in food because I liked making things with my hands, and because I got to make a product that was both tangible and comprehensible (unlike, say, machine learning models, which I also work with today). But over time, I started to wonder if it really was as comprehensible as I thought. Many aspects of our food system – from the dominant yield-focused model of agriculture to Yelp ratings for restaurants – are ultimately reductive, suggesting that most people don’t find the subject as legible as I did. The result is a food system which looks like a giant fractal McNamara Fallacy, built on the delusion that a few simple metrics provide full and complete answers. We’ve engineered our way into a situation where the central problems were solved a long time ago. As the economist Jason Furman recently said about contemporary attempts at agricultural automation, “all the genius in the world that goes into mastering the soft fruit mechanization problem is going to make very small difference to an economy where the vast majority of agriculture, in things like corn, soybean and wheat, was mechanized 150 years ago.”

Now the problems ahead are precisely the ones that sprang from reductive approaches, and solving these problems might be harder. At a minimum, we’ll have to incorporate new metrics, to account for the effects we currently label “external” – greenhouse gas emissions, pollution, and waste. Other issues, like the question of how we produce food without exploiting people, might be measurable – but aren’t easily solved by technology or innovation. And yet others, like whether and how to preserve taste and pleasure in food as agriculture and the food system change, might lie beyond the reach of numbers altogether.


The most clicked link from last week's issue (~8% of opens) was a scrolly website showing emojis arranged by scale. In the Members' Slack we've been hashing out some key business decisions: starting up an in-house shop vs. outsourcing everything, and bringing teams back to the office vs. going fully remote.



  • I could write multiple issues devoted to the amazing pieces of engineering that we use to mass produce various foods, but it’s Pi day, so I’ll confine myself to some videos of pie production lines. I’m particularly fond of the pie ejector slide in this video, which transitions elegantly between continuous and discrete processing.
  • This is my favorite machine for moving pie. While you could obviously have it move the pie into your mouth, the pie cooling tower in the Table Talk factory is undeniably a more impressive destination.
  • There are plenty of smaller-scale tools to gawk at in food production too. As examples:

  • An elegant species of arbor press for smaller pie manufacturers. The process shown here is known as “blocking,” and is also used by the production line in the first video in this section. I’ve been unable to discover the etymology behind this term, nor have I come across it in any context other than the mass production of pies in Commonwealth countries.

  • The humble rotary cutter attachment plays a key role in the production of semlor. Are semlor a variety of pie? It’s not entirely clear; this is a rare instance where the Cube Rule of Food Identification leaves room for debate.

Inspection, Testing & Analysis.

  • One reason we put so much effort into managing yield and efficiency in food production is that pleasure, appropriateness, and benefit to the eater are nearly impossible to measure. Several attempts have been made to quantify pleasure, including a recent one focused on chemo-sensory pleasure, by clinical psychologists looking to study anhedonia.        

               On the other hand, methods for measuring pasta quality are very well defined and include some unexpected approaches, like infrared spectroscopy for “authenticity analysis.” It’s worth remembering, though, that “the relation/correlation between the data obtained from sensory and instrumental techniques is very variable.” And if you’re wondering which brand of pasta you should buy… “a comparison of the results of different panels is only able to distinguish major differences between samples.”
  • My new favorite testing methodology has to be the Oscar Mayer flop test, a method of grading pork bellies by the firmness of their fat. “Variations include calculating the angle between the drooping ends, measuring the distance between the 2 drooping ends, cutting a standardized portion from the belly and drooping that over the bar, adjusting values for belly thickness.”
  • A friend who keeps kosher recently commented that he’s had a much easier time finding Asian pantry ingredients now than when he started cooking a decade ago. It turns out that there’s an entire infrastructure for rabbinical inspection in China, and like all infrastructure, this has taken time to develop.



Every month, the USDA compiles its World Agricultural Supply and Demand Estimates report, universally known as the WASDE report. This is an utterly manual process. Every month, staff at the USDA’s county offices literally go into the fields and physically assess crop conditions, or call farmers in their county and gather their reports. Agricultural attachés in a hundred different countries send reports back to Washington, and conduct in-person surveys and interviews with agricultural producers in 180 countries. There’s a certain amount of opacity and expert opinion (which is to say subjectivity) involved, both in the reporting and then again in the final tallying. The room is smoke-filled in spirit if not in fact.

On the face of it, there’s nothing unusual about the US government wanting to compile global trade statistics for a commodity, product, or industry. But as far as I know, no comparable report exists for ores, steel, cars, microchips, or energy commodities. The WASDE is, I believe, unique among government reports on industry in how comprehensive and frequently updated it is.

My favorite bit of meta-reading about the WASDE is this official history of WASDE-related fraud and fraud prevention – since the report moves literally the entire agricultural commodities market, the incentive to cheat is strong. As someone told me, “they’re still talking about that thing with the venetian blinds like it was last week.” (The incident took place in 1905.)


Trees look great on LIDAR.

Read the full story

The rest of this post is for SOW Subscribers (free or paid) only. Sign up now to read the full story and get access to all subscriber-only posts.

Sign up now
Already have an account? Sign in
Great! You’ve successfully signed up.
Welcome back! You've successfully signed in.
You've successfully subscribed to Scope of Work.
Your link has expired.
Success! Check your email for magic link to sign-in.
Success! Your billing info has been updated.
Your billing was not updated.