MAO Translations exists to replace the old fan-translation production model with one built for scale, scrutiny, and correction. We produce complete, source-aligned Japanese-to-English translations that can be audited line by line, revised without years of institutional paralysis, and reproduced across works rather than treated as one-off miracles. We use that capacity in both directions: backward, to expose and replace weak inherited translations protected by age and familiarity; forward, to eliminate the backlog of works the volunteer model left untranslated, unfinished, or indefinitely promised.
The point is not to torment fan translators, though some will undoubtedly feel tormented. The point is to end the old production regime and establish the new one.
Fan translation has had decades to prove that human authorship, volunteer labor, and long development cycles deserve a presumption of competence. They do not. Projects spend years in development, vanish unfinished, or eventually release English that misunderstands the Japanese and then becomes untouchable because readers are grateful that anything exists. Time invested becomes a substitute for quality; familiarity, for accuracy. Once the text itself can no longer be defended, “people worked very hard on this” becomes the final appeal.
That regime has exhausted its claim to deference.
Our White Album 2 translation was the Gutenberg moment: the point at which a process previously treated as artisanal, slow, and irreducibly manual became reproducible infrastructure. It showed that an agentic translation system could process an enormous literary visual novel, preserve character relationships and work-wide recurrences, produce English that was both more accurate and better written than the incumbent patch, withstand fresh source-critical review, and audit the inherited translation from individual phrases to whole-work interpretation.
The press works. Full-scale production begins now.
The first production line runs backward through the inherited catalog.
The second production line runs forward into the untranslated backlog.
The first line: the inherited catalog
We replace it.
For every visual novel with an existing fan translation, the procedure is straightforward: translate the work independently from the Japanese, freeze the result, and only then introduce the incumbent patch for comparison.
The audit should be public and exact. Every defect receives the Japanese source, the incumbent wording, the correction, the explanation, and a permanent script reference. Local errors are only half the case. The evidence must also be assembled across the work to show what happens when hundreds of individually plausible choices repeatedly flatten register, moralize motives, erase jokes, break callbacks, or reconstruct a character as someone fundamentally different.
Once that record exists, every incumbent translation faces the same problem Todokanai TL now faces.
It may remain untouched and preserve its immaculate human provenance. Its errors remain untouched as well, publicly documented and increasingly difficult to deny.
It may correct the demonstrable mistakes. The patch becomes less wrong, but the repairs are downstream of an agentic audit that found the defects, supplied the source analysis, and directed human attention toward them. Human verification still matters; it does not erase the assistance. If the machine showed you where the body was buried, it participated in the excavation.
Even a successful line-by-line repair leaves the inherited structure intact. Characterization flattened across hundreds of scenes cannot be restored by replacing a dozen phrases. A recurring image translated five different ways does not become coherent because one occurrence was fixed. Prose that is locally defensible but cumulatively destructive does not become literary English merely because the most obvious mistranslations have been removed.
Repairing damage at that scale requires abandoning the incumbent English as an editorial foundation and translating the work again from the Japanese. At that point, the old patch is not being repaired. It is being replaced.
Any replacement will also be produced in a world where the complete agentic translation, error map, terminology record, source notes, character analysis, and literary dossiers are already public. Its translators may claim that none of this influenced their decisions, and they may be telling the truth. The prior solution will still exist, however, and clean-room independence will no longer be easy to establish after years of public exposure to it.
Once the first line reaches a work, every inherited fan translation must choose among three increasingly uncomfortable positions: remain pristinely human and publicly wrong; accept agentic assistance while preserving a defective foundation; or destroy itself and attempt a full retranslation beneath the shadow of the system that already diagnosed it.
The second line: the untranslated backlog
We get there first.
Not with announcements, teaser screenshots, or twenty-percent patches followed by six years of recruitment posts. With complete scripts, playable releases, source alignment, permanent hashes, terminology records, literary notes, and public documentation of the decisions that were genuinely difficult.
The moment such a release exists, every later human translation enters an English space that is already occupied.
If its translators read ours, their project is openly downstream of an agentic translation. If they reproduce distinctive wording or unusual solutions, readers will ask whether they borrowed them. If they arrive at the same obscure interpretations, readers will ask whether they consulted our notes. If they choose differently, the alternatives can be checked against the Japanese. If they avoid the strongest English merely because the agentic translation used it first, fear of resemblance begins damaging their prose.
A later project may claim clean-room independence, but the burden changes. It must explain how its translators, editors, testers, and beta readers avoided the complete public translation for however many years the project required, and how no feedback derived from that translation entered the process.
None of this means that an independently produced translation becomes plagiarism merely because it resembles an earlier one. The point is narrower and more durable: once a complete prior English text is universally accessible, striking resemblance becomes a permanent provenance question. We do not need to make the allegation. The audience will ask it on its own.
At sufficient scale, the two production lines close the historical interval from both directions.
Everything behind us is audited and replaced.
Everything ahead of us is already occupied.
The old translations retain uncontested human provenance, but their errors are public. Future translations may be better, but they will be produced after the agentic solution already exists. Independence will have to be demonstrated rather than assumed.
None of this prohibits human translation. Human translators may continue for pleasure, artistic conviction, or the sincere belief that they can do better. They will simply have to prove that superiority in the text. Human labor will no longer confer ownership of a work’s English readership, and years spent producing a patch will no longer function as immunity from comparison.
The new standard is complete releases, source-visible evidence, permanent accountability, rapid correction, and translations that justify themselves by what they say rather than who produced them.
White Album 2 proved that this era was technically possible.
The first line destroys inherited authority.
The second line destroys translation scarcity.
The presses are already running.