The quality gates and the critic

No episode reaches your library unchecked. While it is being made, automated gates measure everything that can be measured — the picture, the sound, the pace, the captions, flashes, units, variety and originality — and an independent critic reviews what only a reviewer can judge: craft, facts, platform policy and whether it is worth watching. An episode that falls short goes back to its writer with the exact reasons. One that still falls short after its rounds is held — kept in your library with the reasons, and not charged.

The path through the checks#

 the writer builds the episode
        │
        ▼
 gates on its texts (seconds) ──▶ the finished video measured ──▶ the critic ──▶ the 90-second edit's story
        ▲                                                                              │
        └────── a fix round, with every failure and the critic's fixes ◀──── falls short ┤
                                                                                       │
              passes ──▶ the library, ready              out of rounds ──▶ the library, held, not charged

The writer checks its own captions and units as it writes. Then the gates run in one pass — the texts first, in milliseconds, then the finished video — so a fix round gets every failure at once. The critic runs only once every other gate but the captions has passed — it is never asked to review a video that must change anyway — and the edit's story is read last.

The gates#

Each gate's name below is the one the episode's Quality report uses.

On the episode's texts#

On the finished video#

A channel in another language and script is linted in that script's own units: Japanese, Chinese and Korean in full-width characters, the Indian scripts by letter, Thai by its dictionary's word breaks. A caption style changes no limit.

On the narration and the Facebook edit#

Once everything else has passed, a reader judges whether the edit makes sense on its own, from its narration alone, out of 10. Under 8, a story editor chooses better lines and the reader judges them afresh. If no choice of lines makes sense, that is a note, never a failure: the whole video still goes to YouTube and every platform that takes it whole, and the episode is not posted to Facebook or Instagram.

When the channel chose them#

The critic#

The critic is a separate review of the finished episode. It did not write it, it has a budget of its own, and it is calibrated against the channel's style, so a deliberate look — low-poly worlds, a glass human, a plain stage for an illusion — is judged as a choice. It sees:

• a contact sheet of 12 to 24 frames, most taken at the middle of a caption so each caption is judged against its own picture, and three larger frames;
• the title, the description, every caption with its times, and every line of the narration as it is said;
• how much the picture moves in each 5 seconds, measured — what stills cannot show;
• what the channel chose: endings that loop, a caption style, the rules learnt from its own viewers' retention, its language (judged as a native viewer would, never asked for English), a product's claims.

It scores each part out of 10 against "a premium, cinematic, accurate short":

Its scale is anchored: 10 is indistinguishable from the best hand-made shorts in the niche; 9 is excellent; 8 is very good with one or two shortcomings a viewer could notice; 7 is good but clearly improvable; 6 or less looks unfinished. The overall score is the craft — visual, readability, story and facts — and entertainment is judged on its own.

An episode passes the critic when its score is 8 or more, its facts and its policy are clear, and the critic lists nothing as blocking. It blocks only for a factual error or a misleading claim, a policy problem, clearly broken rendering, captions unreadable in most frames, or a title the video does not deliver. Everything else goes into its fixes: concrete changes, the most valuable first, with what to show and when.

Entertainment is a separate bar: under 7, the episode goes back once with the critic's list of what to show and where. It never goes back a second time for that, and entertainment alone never holds an episode: if nothing else fails, the better of its two versions goes to the library, ready, and the reasons are kept as a lesson for the channel's next episodes. Its quality report still lists the entertainment shortfall under What held it, and it carries no Original & reviewed badge.

Fix rounds, the polish round and the budgets#

• Fix rounds. A failed episode goes back to the same writer — it does not start over — with every failure and the critic's fixes, and is rendered and checked again. Up to two fix rounds follow the first build.
• The polish round. An episode that passes on its first build but scores under 9, or under 8 for entertainment, gets one more round on the critic's fixes, while it has time and budget left. If the polished version scores lower or breaks something, the version from before it is kept.
• Budgets. Every episode has a budget for its model calls, set by its quality — standard or premium — and about two hours of its writer's time, most of it for the first build so the fix rounds have room. The critic's review has a budget of its own, so a writer that spent everything never leaves an episode unreviewed. When a round runs out of budget, it ends there and the episode is judged as it stands; it is never thrown away and made again from nothing.
• A render cut short. If the writer stopped before its final render finished, the episode is rendered once from its own code and checked again. If a fix round leaves no video, the version it started from comes back.

When an episode is held#

An episode that rendered but still has failures after its rounds goes to the library held:

• it is not charged — the job's whole reservation comes back, and the credit history says "Held by the quality check — not charged";
• no slot posts it, and its page says Held: with the first reason in plain words — "The critic scored it 7/10 — a short needs 8 to pass" — and its Quality report lists what held it and What the critic would change;
• the schedule goes on with the next topic at its usual pace, and a setup tries another topic in its place.

What you can do with it:

1. Leave it. It costs nothing and posts nothing.
2. Release for posting — if you are content with it as it is. The button shows the price: what it cost to make, charged once. It can then be posted, downloaded and translated like any ready episode.
3. Edit it. Edit this episode asks for a revision — the changes you describe, checked again in full; a revision that does not pass is not charged. One that passes leaves the episode held, ready for you to release at the same price. See The editor and storyboards.

A video that never finished rendering is not held: there is nothing to release. Its job is tried once more, and the failed attempt is never charged. If the second attempt fails too, the job fails, not charged, and the schedule makes the next episode at its usual pace — a failed job still counts towards the day's New episodes a day.

The Original & reviewed badge shows only on an episode whose report passed in full, with the critic's facts and policy clear.

Films#

A long-form film is checked by the same rules, measured for 16:9 and 6 to 10 minutes: 1920 × 1080 at 30 frames a second; −14 LUFS (±1) with a true peak at or below −1 dBTP; YouTube's chapter rules (from 0:00, at least three, each 10 s or longer, titles of 60 characters at most); no black but brief dips at act boundaries, no freeze of 5 s; a still stretch of 15 s fails; flashes; captions; units. The critic reviews every act and the whole film, and a film's floor is 8 for its score and 8 for entertainment. A failure names its act, and a round rebuilds only the acts that failed. Two improvement rounds are included; a film still under the floor is held, free, with a watermarked preview. Releasing a held film costs its quoted price, and an improvement round you order on it is charged even if the film stays held.

Translations, revisions, hook variants and clips#

Reading the quality report#

Credits and plans#

The checks are part of making an episode: the critic's review is part of an episode's cost, never a separate charge. A held episode, a revision that fails and the making of a film held under its floor are not charged. Every plan gets every gate and the critic; premium quality, on Pro and above, changes the writer, not the bar.

Troubleshooting#

Related#

• Units, symbols and facts — the units gate and how facts are kept honest
• Originality guard and checker
• Approvals on your phone — your own yes before an episode posts
• The quality system — the design behind it