Art Is Never Finished, Only Abandoned

Transfers

The aphorism — attributed to Paul Valery (“Un ouvrage n’est jamais acheve … mais abandonne”) and sometimes misattributed to Leonardo da Vinci — reframes the act of finishing creative work. There is no moment when a painting, poem, or program is objectively complete. There is only the moment when the maker stops working on it, either by choice or by external force.

This is not a lament. It is a structural observation about creative work that transfers to any domain where the output has no natural stopping point.

Key structural parallels:

• Completion as a stopping rule, not a state — in carpentry, a joint either fits or it doesn’t; in mathematics, a proof either holds or it doesn’t. But in painting, writing, and software design, there is always another revision that could improve the work. The model reframes “Is it done?” from a question about the object to a question about the maker’s judgment: “Have I decided to stop?” This transfers to software shipping decisions, where “feature-complete” and “release-ready” are decisions by people, not properties of the code.
• The diminishing-returns curve of refinement — early revisions to a creative work produce dramatic improvements. Later revisions produce smaller improvements and carry increasing risk of degrading what already works. At some point the curve inverts: further work makes the piece worse. The model names the rational response to this curve: stop before the inversion. In software, this is the argument for shipping an MVP rather than polishing indefinitely — not because polish doesn’t matter, but because the marginal value of each additional polish pass decreases while the opportunity cost increases.
• External constraints as enabling forces — Valery’s insight explains why creative workers are often more productive with deadlines than without them. The deadline is not an interruption of the creative process; it is the mechanism that forces the abandonment decision the maker would otherwise defer. Patrons, publishers, sprint deadlines, and conference submission dates all function as commitment devices. Without them, the work accretes revisions indefinitely — not because the maker is lazy but because the work genuinely could always be better.
• Abandonment as craft judgment — the model elevates the decision to stop from a failure (“I gave up”) to a skill (“I recognized the right stopping point”). Knowing when further work will subtract rather than add is itself a form of expertise. The junior painter overworks the canvas; the master stops while the brushwork is still visible. The junior developer over-engineers the solution; the senior ships the simplest version that solves the problem.

Limits

• Not all completion is subjective — the model applies to creative and design work where quality is open-ended, but many professional domains have objective completion criteria. A bridge either bears its load or it doesn’t. A tax return either balances or it doesn’t. Applying the “never finished” model to work with clear done conditions imports false ambiguity and can justify shipping incomplete work by calling it “abandonment” when it is actually “failure to meet specification.”
• Romanticizes quitting — the aphorism can become a permission structure for chronic non-completion. The writer who abandons every novel at the difficult middle chapters, the entrepreneur who pivots away from every business at the first plateau, the musician who records demos but never finishes albums — these are not exercising refined artistic judgment about stopping points. They are failing to push through the inevitable difficult phase of any creative project. The model provides no way to distinguish wise abandonment from premature surrender.
• Collapses a spectrum into a binary — the aphorism suggests two states: working and abandoned. But real creative work has graduated completion states: rough draft, working draft, polished draft, submitted, published, revised edition. Each transition involves a stopping decision, but they are qualitatively different decisions with different stakes. Treating them all as “abandonment” obscures the meaningful differences between shipping a beta and walking away from a project entirely.
• The attribution problem undermines its authority — the aphorism is attributed to Valery, Leonardo, and others, with no definitive source. Its folk provenance means it carries the authority of ancient wisdom without the accountability of a specific argument. This makes it easy to invoke as a conversation-ending truism rather than engaging with the specific question of whether this particular work, in this particular context, should be stopped or continued.

Expressions

• “Ship it” — the software engineering command that enacts Valery’s principle: stop refining and release
• “Perfect is the enemy of good” — Voltaire’s formulation of the same diminishing-returns insight, without the artistic framing
• “Done is better than perfect” — Facebook’s poster, operationalizing the abandonment principle as organizational culture
• “Scope creep” — the project management term for what happens when the “never finished” nature of the work is not counteracted by an abandonment decision
• “The last 10% takes 90% of the time” — the engineering observation that maps onto the diminishing-returns curve the aphorism describes
• “Let it go” — the colloquial version of the abandonment decision, often said by the person on the team who has internalized the stopping-rule insight

Origin Story

The aphorism is most reliably attributed to Paul Valery, the French poet and essayist, who wrote in 1933: “Un ouvrage n’est jamais acheve — accident heureux ou lassitude — mais abandonne.” (“A work is never completed — happy accident or weariness — but abandoned.”) Valery was reflecting on his own struggle with his long poem La Jeune Parque, which he revised obsessively for four years before publishing. The misattribution to Leonardo da Vinci, while historically unsupported, has a certain aptness: Leonardo’s notebooks are full of unfinished projects, and the Mona Lisa was reportedly worked on for years without Leonardo considering it complete.

The aphorism entered software engineering discourse through the open-source and agile movements, where it provides philosophical justification for iterative release, MVPs, and the general preference for shipping over polishing. Its adoption reflects software’s structural similarity to art: code, like a painting, can always be improved, and the absence of a natural stopping point makes the decision to ship a judgment call rather than an objective determination.

References

• Valery, Paul. “Au sujet du Cimetiere marin” (1933) — the probable original source of the aphorism
• Brooks, Frederick P. The Mythical Man-Month (1975) — on the impossibility of knowing when software is “done”
• Ries, Eric. The Lean Startup (2011) — the MVP as a formalized abandonment strategy: ship the minimum, learn, iterate