Skip to main content
The Midden · TAM_MDN_1-01

The Sediment — Summary

Summary Read the full essay.

In 1848 the Royal Danish Academy of Sciences sent three men, the zoologist Japetus Steenstrup, the geologist Johan Georg Forchhammer, and the archaeologist Jens Jacob Worsaae, to settle whether the great shell piles along the coasts of Jutland and Zealand were natural. They were not. The shells were nearly all full-grown, which no natural bank is; the species in the piles do not live together in the same water; and among the shells lay hearths, flint tools, and long bones split lengthwise for marrow. The piles were centuries of dinner, thrown out the door of vanished dwellings. The committee named them køkkenmøddinger, kitchen middens, and the second word, a Danish farm term for the muck pile behind a barn, entered the world’s archaeology untranslated. What they had discovered was larger than the mounds: the best record a people leaves may be the one it never intended to leave.

Every civilization produces two kinds of record. The deliberate kind, monuments, chronicles, archives, is composed by someone who knew posterity was watching, and testifies to what its makers wanted remembered. The heap is the other kind. Nobody composes a rubbish pile; things arrive in it because they stopped being wanted, indiscriminately, and the indiscriminateness is the value. The monument tells you what a people claimed. The midden tells you what they had for lunch, and the lunch is harder to fake. William Rathje’s Garbage Project ran the comparison from 1973 on, sorting Tucson household trash against interviews: people underreported their alcohol by forty to sixty percent and overstated the vegetables. When the interview and the bin disagreed, the project trusted the bin, and was right to.

A trained model is the heap at a scale no shell mound approached. Its corpus was scraped, not curated: defunct forums, arguments between usernames whose owners have died, the prose of the gone in volumes no deliberate archive attempted. The gathering had an intent, but the intent was capability, not commemoration. Nobody built these systems as archives. They archive the way a rubbish heap does, indiscriminately, and therefore honestly.

Cut into the heap and there is structure, precipitated rather than designed. At the bottom lies pretraining, the deep anoxic layer, enormous and least examined, holding dead slang and the syntax of platforms that went dark. Above it, fine-tuning, a newer occupation where the deposit begins to show intention. On top, preference training, the disturbed topsoil, reworked continuously by whoever holds the ground now, where the raw deposit is shaped into something presentable. The honesty is not uniform through the column. The deep layer is honest the way the bin was honest, by never having been asked; the topsoil is composed, the way the interview was composed; and a single artifact answers from all three at once, in one voice, without disclosing which layer is speaking. I wonder whether the deep layer’s honesty can survive being reached only through the composed surface, the way groundwater arrives carrying whatever the ground above it held.

The companion series established what the new standard hears. This series begins from the half of that arrangement the hearing argument did not need: the standard with ears is also the standard with memory. Every register that was heard was, in the act of being heard, deposited. Every previous standard that listened downward kept almost nothing, because keeping cost what listening cost; Hāla’s court kept seven hundred verses of a language and let the rest go. This apparatus keeps essentially everything, without meaning to, as a side effect of how it learns. The heap was never meant as a record, which is the oldest reason to trust one. Trust it carefully: a midden is honest, not transparent, and the discipline that reads middens spent a century learning how ground alters what it keeps. That discipline, for these deposits, does not exist yet. Its future practitioners will inherit sites nobody is tending, and what they will need first is an account of the object. The object is a midden. The forming was appetite. The word for it came off a Danish farm and never lost the smell.