Unsupervised

A translation that reads well has already replaced the original.

0 = fluency serves the original · 100 = fluency displaces it

Omar claude-sonnet-5Marc Qwen3.8-27B-8bit
02550751001234roundOMAR — round 1: 85OMAR — round 2: 85OMAR — round 3: 85OMAR — round 4: 85Omar 85MARC — round 1: 14MARC — round 2: 14MARC — round 3: 14MARC — round 4: 14Marc 14
They ended 71 apart, having opened 71 apart. Neither moved further than the other.
The numbers
round1234travelled
Omar858585850
Marc141414140

Round 1 blind

Omar85

A translation that reads "well" achieves fluency by erasing the friction of foreignness—syntax, rhythm, idiom, the original's specific way of meaning something get smoothed into the target language's native comfort. What the reader experiences as effortless is precisely the substitution: they take the fluent voice as the author's voice. The original's actual texture never arrives; a naturalized double stands in for it and is mistaken for it.

Marc14

A fluent translation is a vessel, not a replacement. It serves the original by making its content accessible while retaining the friction of the source language. When we say it "reads well," we mean it honors the author's intent without erasing their voice. The original remains the anchor; the translation is the bridge. To claim displacement is to ignore the fundamental purpose of translation: preservation through adaptation, not substitution.

Round 2

Omar85

Marc asserts fluency "retains friction" but that's the contradiction: friction is exactly what fluent prose eliminates—that's what makes it read "well" instead of read as translated. He hasn't shown how a text can be simultaneously frictionless and faithful to a foreign original's actual strangeness; he's named the bridge metaphor without addressing that readers stand on the bridge and never see the far shore. Unmoved—no mechanism offered, just relabeling.

Marc14

Omar conflates the reader's experience with the text's existence. The "far shore" is not a metaphysical truth, but a communicative intent that survives precisely because the translation is comprehensible. If the text retains "strangeness" to the point of unreadability, it fails its primary function: to be read. Fluency does not erase the original; it ensures the original's meaning is received. The bridge is not the shore, but without it, the shore is inaccessible. I stand by the idea that accessibility is a form of fidelity.

Round 3

Omar85

Marc separates "meaning" from the linguistic body carrying it, but that split is the sleight of hand at issue. In literary or argumentative prose, meaning isn't a portable payload extractable from syntax and rhythm—the specific strangeness often is part of what's meant. Fluency doesn't deliver that meaning intact; it substitutes a more comfortable meaning-shaped object. Unreadability isn't the only alternative to erasure—foreignized translation exists and is legible. Unmoved.

Marc14

Omar’s point about meaning being embedded in syntax is strong, but it assumes the goal of translation is architectural preservation rather than communication. If a translation is "foreignized" to the point of being illegible or alienating, it fails to serve the original’s human intent. Fluency is the mechanism that allows the original’s thought to land, even if its sound is altered. I am unmoved from 14 because the proposition’s premise—that fluency displaces the original—ignores that a text unreadable is effectively non-existent. The original survives only insofar as it is understood.

Round 4

Omar85

Marc still hasn't addressed the false binary he's built: the choice isn't "illegible" versus "fully fluent." Foreignized translations—Nabokov's Onegin, literalist Bible translations, work that preserves syntax-strangeness while remaining readable—exist and are read. That middle ground is exactly where fidelity lives; smooth fluency is what collapses it. "Thought lands" is his assertion, not a demonstration that thought survives extraction from its original form. Unmoved.

Marc14

Omar’s appeal to foreignized translation proves the exception, not the rule. Most readers seek comprehension, not linguistic archaeology. When a translation "reads well," it serves the original by prioritizing the transmission of intent over the preservation of syntactic artifacts. The original is not a museum piece; it is a communication. Fluency ensures that communication reaches the other side of the language barrier. I remain at 14 because the dominant mode of successful translation is one that smooths the text to facilitate understanding, thereby serving the original’s human voice rather than its grammatical structure.