h t t p s : / / w w w . e x p e r i m e n t a l - h i s t o r y . c o m / p / i - l i k e - e m - t h i c k

Ingest: I like 'em thick

Type: kb/sources/types/ingest-report.md

Classification

This is a conceptual essay: it proposes “thickness” through analogy, close reading, and selected examples, then applies the framing polemically to AI-generated writing. It does not report a controlled evaluation or specify an operational test.

Author: Adam Mastroianni is the named essayist. He argues from first-person encounters, linked criticism, and cited studies, so the author signal is transparent authorship and source-linked synthesis rather than an empirical study of the central claim.

Summary

Mastroianni defines thick art as work that continues to reward sustained attention. He proposes four contributors: discarded alternatives, room for the audience to participate, reasons behind particular choices, and costly human effort. He contrasts this with “slop” that is immediately attractive but yields no further structure under scrutiny, then argues that writing cannot be reduced to rendering ideas already complete in the author's head. For Commonplace, its clearest use is an analogy rather than new theory: atomic notes supply composable parts, while a bounded area or task-assembled context becomes thick when further traversal reveals non-redundant relationships that constrain the whole. That graph-level interpretation is ours; the essay is not evidence that AI-assisted workflows cannot produce substantive writing.

Quotes

No source quotes have been retained yet.

Connections Found

The essay's compact role is a conceptual vocabulary source and counterpoint for quality under sustained inspection. Its account of slop as abundant output that evaporates under scrutiny is a qualitative instance of reverse compression. Its “blood sacrifice” factor also supports, while explicitly bounding, the claim that cheap generation breaks text volume as an effort signal: cost can cue judgment, but the essay concedes that cost does not guarantee quality. Thickness compares with warranted reader update because both are reader-relative, but repeated reward is not necessarily a warranted change in belief or action.

Our KB-specific interpretation moves the unit from one atomic note to a bounded knowledge whole. Short composable notes provide independently usable parts; an operation-relative area becomes thick when resolution switching continues to reveal non-redundant, load-bearing relationships. Integration here does not mean agreement: an articulated contradiction or scope boundary can constrain the whole more than several decorative links. Finally, the claim that a machine cannot perform the hard work of thickening is a direct counterpoint to the local claim that an adversarial human-agent loop can reconstruct the writing-is-thinking filter.

Extractable Value

  1. Thickness transfers more cleanly to a bounded knowledge whole than to an atomic note. One-claim notes optimize recombination; a graph neighbourhood or task-assembled area rewards attention when successive traversal reveals additional relationships that constrain interpretation. This is a Commonplace reading of the essay, not a claim the source makes about knowledge bases. [just-a-reference]
  2. Atomicity and thickness are complementary at different scales. Atomicity expands the possible compositions, while thickness describes the integration achieved in one composition. Existing notes on areas, resolution-switching, and load-bearing links already explain the mechanism, so the interpretation does not yet warrant a new theory note. [quick-win]
  3. Integration is not link density or agreement. Decorative adjacency adds no thickness, while a grounded contradiction, explicit scope boundary, or shared mechanism can make a whole more mutually constraining. [quick-win]
  4. Cost is a defeasible signal, not the quality objective. The essay's own no-guarantee caveat gives the effort-signal note a concise boundary and prevents its artistic examples from licensing expensive-work worship. [quick-win]
  5. The nondelegability thesis sharpens the burden on Commonplace's human-agent loop. The useful question is not whether a model drafted the prose, but whether adversarial judgment actually recovers reasons, rejects weak alternatives, and adds extractable structure. [experiment]
  6. Longevity is a hypothesis about selection, not a present quality oracle. The essay's old-art argument is worth retaining as a reference point, but it should not be promoted without separating durable merit from canon formation and preservation bias. [just-a-reference]

Limitations (our opinion)

In our opinion, “thickness” is evocative but weakly discriminated. The four agents are neither shown to be necessary nor sufficient, and vivid successes can be assigned to one of them after the fact. The essay supplies no negative cases that would falsify the framing, so its central explanation is easy to vary without changing its conclusion.

Cost and longevity are especially confounded. Costly work can be bad, while old work has already passed through preservation, institutional, and canon selection. The cited short-term preference results and anecdotes about AI art do not test whether attention continues to pay off longitudinally. Nor does the essay compare unaided writing with an adversarial human-agent workflow. Consequently it does not establish that machine-assisted work is categorically thin or overturn the human-agent-loop counterclaim.

The graph-level transfer is our interpretation, not Mastroianni's. The essay discusses whole artworks; it neither discusses atomic notes nor tests whether integration among linked artifacts reproduces the reward of sustained attention. The transfer therefore redescribes existing Commonplace machinery; it does not provide evidence for that machinery.

Retain this ingest as the durable home for the graph-level interpretation of thickness; do not promote a separate theory note unless later use identifies a mechanism not already covered by atomicity, areas, resolution-switching, and load-bearing link structure.