Rules of the prize

What the prize is

The Ralph Prize for Writing in the Age of AI is an annual, human-judged prize for short stories. You may write with any amount of help from artificial intelligence, including none. The prize does not measure, restrict or penalize how much AI you used.

Who may enter

• You must be 14 or over. If you are 14 to 17, a parent or guardian must approve your entry and agree to supervise your use of the site before your story can lock. In this test version you confirm this yourself (TEST · OQ-35).
• You enter one story. One story per person.
• Everyone who enters, including people aged 14 to 17, reads and reviews other entrants’ stories. Stories are not given an age rating.
• You are the primary author of your story and you are responsible for the entry, for your reviews and for any prize.

AI use

AI use is unrestricted and observed. You may use AI extensively, lightly or not at all, and you do not have to prove how much you used. After you lock your story you write a short private note about whether and how you used AI. The note is never shown to the people who read your story.

Reviewing other stories

The reviewers are the entrants and people who join only to review. After you have completed one review, your story is released to other reviewers. Your story is entered once you have completed three reviews of other entrants’ stories. You may use AI to help you read but not to write your review. A review can be ruled unusable by a human proctor, for example when it was not based on reading the story; if that happens you are asked for one more review, and you are told nothing else about it.

Reviewing without entering a story

You may join a contest only to review, without entering a story of your own.

• You must be 18 or over to join only to review.
• You do not enter a story. While the contest takes entries you may still enter one later, and the rules for entrants then apply to you as well.
• You commit to three reviews (a test setting, TEST · OQ-23). Stories are offered to you one at a time, with the same offers, due dates and reminders as for an entrant.
• You read in the same reader and answer the same questions as every reviewer. Stories are read blind, and if you recognize the author of a story you must declare it and step aside.
• What you read is confidential. Do not copy, share or publish a story you are given, or any part of it.

Deadlines

Each contest sets its own dates: the last day to lock a story, the day reviewing closes, and, where they apply, the close of the second round and the day the jury’s work is due. Your dashboard shows the last day to lock a story, the day reviewing closes and the due date of each review you accept; the later dates are sent to the people they concern. The dates the contest shows are the dates that count.

• A story you are offered to review stays offered for 48 hours. If you do not accept it in that time the offer ends and you may ask for another.
• A review you accept is due seven days after you accept it, or earlier if its round ends first: in the first round, 48 hours after reviewing closes; in the second round, when it closes. The offer shows the due date before you accept. We send a reminder before it is due. A review that is still not submitted 24 hours after it was due lapses: it stops holding a place, its story goes back to be read by someone else, and the review is still owed. After two lapses in one contest your participation is put on hold for an administrator to look at. The numbers are test settings (TEST · OQ-26).

If you have not finished your reviews when reviewing closes

When reviewing closes, a story whose author has completed the required reviews is entered. A story whose author has not is held for a further 48 hours (a test setting, TEST · OQ-26). You are sent an email 72 hours before reviewing closes that says so, and the dashboard and your story page say it too. If you finish the required reviews in that time, your story is entered. If you asked for a story to review during that time and every ask found none to read, you are credited for what you could not do and your story is entered. Otherwise your story is not entered, it is not read again, and your story page says so in plain words. A story whose author still owes a review that a human proctor ruled unusable, when the decision is made, is marked participation hold, and an administrator decides.

Reading more than you must

Once you have completed the reviews your entry needs, the dashboard offers a button, “Review another story”. It gives you one story at a time, only when a story is waiting, and never more than five in one contest (TEST · OQ-36). A story you review this way counts like any other usable review. Nothing is created when nothing is waiting, and the button looks the same for everyone.

Extra reads, and what an offer can tell you

Some stories need more reads than the standard number. In this test version every read after the standard number is offered, one at a time, only to a reviewer on the contest’s roster. There is no fallback to other reviewers: if no roster reviewer is available, the read waits and the administrators see the shortage. The roster is a list of reviewers that an administrator has approved after looking at the reviews they have done. It is never shown to participants. We tell you two things about how this can touch you, because each lets you learn something you are not told outright (TEST · OQ-11).

• An offer you receive after a story’s standard reads are done is an extra read, and extra reads go only to roster reviewers. So receiving one tells you that you are on the roster, and it can mean the story needed another read. You cannot tell which story or whose it is.
• Administrators approve reviewers for the roster and invite them to take on further reads. An invitation or an approval tells you that an administrator approved your reviews. The wording of an invitation never mentions quality.

Apart from the roster and the offers and invitations it brings, nothing you are offered, invited to or allowed to do depends on how your story or your reviews were rated, and no participant is ever shown a rating, a standing or a result in progress.

Blind review

Stories are read blind, with names and identifying details removed. For nonfiction, blindness is best effort: a story can identify its author without meaning to. If you recognize the author of a story you are asked to review, you must declare it and step aside. Recognition is a duty to declare.

How reviews are combined, and what the AI does

In this test version every review of a story counts the same. For each stage, a story’s result is worked out from the advancement and overall-rating answers of its reviews. Administrators also look at the reviews people have done to decide who joins the roster; that standing is used only to choose which reviewers are offered extra reads, and it is never shown to a participant. The platform records reading telemetry, meaning reading time, scrolling, and copy and paste, on every review, and shows it to administrators only; telemetry is not proof of reading and does not change a story’s result. Only people decide how a story is rated and which stories win. No AI scores, ranks or chooses any story, and no AI writes anyone’s review. Software helps administrators run the process.

The Prize may use AI, including outside AI services, to study stories and reviews and to improve the contest. For example, an outside service checks each story for AI-written text, for research; the result is private and never counts against a story. None of this is ever an input to any story’s result.

The second stage

Stories that are shortlisted after the first stage are read again, blind, by further reviewers from the roster. Each read is offered to one roster reviewer at a time. Reads from the two stages are kept apart: each stage has its own result for a story, and the contest then combines the two by a rule it sets (a test setting, TEST · Q1). The platform does not show any participant whether a story was shortlisted.

The jury

The final decision is made by people, not by software. The jury is a small group of judges chosen and invited by the contest’s administrators. A judge is never an administrator and is not part of the contest’s staff.

1. Judges read the finalists’ pool blind, each story under its own jury label, with no name, no title from the author’s account, and no review material.
2. Each judge assesses every story alone and then locks, once and for good, their own complete ordering of the whole pool. A judge who recognizes a story says so and steps aside from it.
3. Only when every judge has locked do the judges see the reviews of the stories and one another’s assessments. They then deliberate. Judges may hold ballots to help them, but a ballot decides nothing: the decision is the one recorded as a numbered decision version by the person chairing the session. No score, ranking or weight from the review stages decides a place. What the jury decides depends on the contest: it either puts all the finalists in order (a pilot exercise, with no prize), or it chooses a top group and then first, second and third.
4. The contest’s administrators then check each finalist (identity, eligibility, rights and consents). If a finalist fails a check, a new decision version is recorded without that finalist. It names the next alternate from the recorded list when there is one; in a pilot exercise there are no alternates, so the places close up. No ballot is held for this. Only a decision version whose finalists all pass is certified. After certification a mistake can be corrected only by a new recorded decision version, and the earlier version stays on the record.
5. Authors are told only what concerns them: an author whose story is selected is contacted to verify details before results are published, and is later asked how their name should be credited. No other author is sent anything, and the results are published only after the Foundation releases them. Names are shown only with the author’s consent; for an author aged 14 to 17 a parent or guardian also confirms.

The number of judges is not yet set (TEST · OQ-14).

Analytics

We use product analytics to see how the site is used. Analytics never contain your story text, your AI-use note or your reviews.

Prizes

There will be prizes. Their details will be announced when the Ralph Prize Foundation approves the rules (TEST · OQ-15).

Your rights

You keep ownership of your story. By entering you give the Foundation a non-exclusive permission to store, copy, format and show your story to reviewers, administrators and approved vendors as needed to run the competition and, if your story is chosen, to publish it in the anthology. You warrant that you are the primary author and that no third party holds rights that this permission would infringe. This permission and warranty do not depend on your holding a copyright in the story.

Versions

These rules are versioned. Your registration records the version you accepted, and every earlier version stays available at its own address.