The Garbage Draft

In the AI Story

The concept rests on an empirical observation confirmed across decades of composing research: writers discover what they think by writing, not before writing. The conscious mind does not possess a complete, articulable thought that it then transcribes onto the page. Instead, the act of writing is the process through which vague, half-formed intuitions crystallize into specific formulations. The garbage draft is the record of this process — the visible trace of a mind discovering its own contents through the effort of articulation. Most of what appears on the page during this discovery will not survive into the final draft. But the process of producing it has deposited understanding that cannot be acquired any other way.

Edo Segal's geological metaphor in The Orange Pill — each hour of struggle depositing a thin layer of understanding, thousands of layers compounding into intuition — is the engineering equivalent of Elbow's garbage draft. The engineer who spends four hours on tedious plumbing work produces little visible output. But mixed into those four hours are ten minutes of unexpected discovery when something fails in a way that forces understanding of a connection between systems. Those ten minutes deposit the layer. The plumbing is the garbage. The unexpected insight is the fossil. When AI eliminates the plumbing, it eliminates both the tedium and the ten minutes, and the engineer does not know what she has lost until months later when architectural decisions feel less certain.

The cost of skipping the garbage draft is the cost of losing access to the felt sense — the pre-verbal, bodily, first-order awareness that guides practitioners through complex decisions. A writer who has produced thousands of pages of garbage has a felt sense for prose that no amount of second-order analysis can replicate. She knows, before she can explain, that a paragraph is wrong. She feels the wrongness as a tightness, a resistance, a subtle dissonance between what the prose says and what she meant. This capacity is not available to the writer who has only ever reviewed machine output, because the capacity develops through production, not evaluation.

The educational challenge is that garbage drafts receive low grades in conventional assessment frameworks. The draft that demonstrates first-order exploration — with its false starts, contradictions, and rough formulations — scores worse than the polished five-paragraph essay assembled from pre-approved ideas. The system punishes discovery and rewards compliance. AI has exposed this dysfunction by producing the polished artifacts faster and more competently than students ever could, revealing that the artifacts were never measuring genuine thinking. The teacher who grades questions rather than essays, the law school that evaluates through live oral argument, the medical school that assesses through direct patient interaction — these are institutions that have shifted from measuring second-order artifacts to measuring first-order capacity, from evaluating products to evaluating the thinking that garbage drafts evidence.

Origin

Elbow's concept of the garbage draft emerged alongside his freewriting practice in the early 1970s. The language was deliberately chosen to counter the Romantic ideal of the inspired first draft — the fantasy that great writing arrives complete, polished, gift-wrapped from the muse. Elbow's experience as a writer and teacher convinced him that this fantasy was not merely wrong but actively harmful, because it made writers ashamed of the messy reality of their compositional process. By naming the first draft 'garbage,' Elbow gave writers permission to produce badly — to embrace the roughness, the incoherence, the failed attempts as necessary rather than shameful. The reframing was therapeutic and developmental: once the writer accepted that garbage was the medium rather than the failure, the internal critic lost its power to paralyze.

Key Ideas

Waste is the medium of discovery. The garbage draft's function is not to produce usable text but to create the conditions under which surprising connections can surface — most of it is discarded, and that is the point.

Depth requires sediment. The geological layers of intuition are deposited through the production of material that will never be used directly — skip the sedimentation, and no fossils can be found.

AI's polished output is pre-composted. Machine-generated text arrives as finished product, eliminating the generative mess through which voice and understanding develop — the writer receives the flower without having tended the soil.

The protocol is sequence-dependent. Produce garbage first through freewriting, then bring it to the machine for refinement — reversing the order produces smooth emptiness.

Assessment must measure process, not product. Educational systems that grade artifacts reward compliance and punish discovery — measuring questions, live reasoning, or annotated garbage drafts measures the thinking that matters.