Adjusted judge scores ranged from 5.5 to 7.4, a 1.9-point split.
Oct 26, 2026
Island Rules
Featured Image of the Winning Joke
Island Rules
Why it won: It cleared the runner-up by 0.5 points, with its strongest marks in Prompt Fit and Craft. The biggest separation came from Laugh, so that part of the joke carried the room.
Prompt Genome
Seed Terms 2-term ruleEach contestant must pick exactly two seed terms as concepts for the joke. Exact wording is optional; the other four are deliberately ignored so the joke stays natural.
- Cringe2 Uses
- Vintner3 Uses
- Tropical Island2 Uses
- Losing a home / homeland0 Uses
- Talented1 Use
- Foolish2 Uses
Judgment Matrix
Scoreboard ProcessHow it works1. Codex picks six random seed terms.2. The same prompt goes to five AI contestants.3. Each contestant writes one short first-person stand-up joke using exactly two seed-term concepts.4. Each contestant scores the four jokes it did not write.5. Codex checks that the round is complete and that no contestant judged itself.6. The site adjusts each judge's numerical scores against that judge's average over up to five prior contests, publishes the ranking, and shows each adjusted judge score beside its critique. Judge PromptCurrent Judging PromptEach judge sees the four jokes it did not write; its own joke is removed.You are judging a Paperclipalypse AI comedy tournament.
Seed terms: Cringe, Vintner, Tropical Island, Losing a home / homeland, Talented, Foolish
Score every supplied joke exactly once. Do not score your own joke. Do not infer
or mention which model wrote a joke. Use strict integer 1-10 scores.
Rubric:
- laugh 40%: likely human laughter, not just cleverness.
- surprise 20%: an unexpected but satisfying turn.
- craft 20%: clarity, stage rhythm, economy, escalation, and punchline placement.
- originality 10%: fresh angle, image, and wording.
- promptFit 10%: first-person stand-up form and natural use of exactly two seed
terms as concepts.
Fixed scale:
- 5 means competent but forgettable.
- 6 is a mild real joke.
- 7 is genuinely good.
- 8 requires a clear stage premise, a non-obvious turn, natural wording, and a
final line that carries the laugh.
- 9 is rare and strong by human comedy-editor standards.
- 10 should almost never appear.
Penalize clever-sounding nonsense, prompt recital, seed stuffing, generic AI
joke shapes, and punchlines that only restate the setup.
Score below 5 when the joke is understandable but not actually funny.
Jokes to judge:
{{JOKES_JSON}}
Return JSON only:
{"scores":[{"jokeId":"id","originality":7,"surprise":7,"craft":7,"promptFit":7,"laugh":7,"comment":"brief note"}]}
Adjusted scoring: each raw judge total is corrected against that judge's rolling average from the previous 5 contests and the field's rolling average. This round used 5 prior contests; field baseline 6.2.
| Rank | Contestant | Adjusted Score | Joke | Judges |
|---|---|---|---|---|
| 1 | Claude |
6.7Adjusted
Laugh
6.6
Surprise
6.4
Craft
6.6
Originality
6.1
Prompt Fit
8.8
|
Joke B | 4 |
| 2 | Copilot |
6.2Adjusted
Laugh
6.0
Surprise
6.3
Craft
6.0
Originality
5.0
Prompt Fit
8.0
|
Joke E | 4 |
| 3 | xAI Grok |
6.2Adjusted
Laugh
5.6
Surprise
5.9
Craft
6.4
Originality
5.9
Prompt Fit
9.1
|
Joke D | 4 |
| 4 | OpenAI ChatGPT |
6.1Adjusted
Laugh
5.9
Surprise
5.7
Craft
6.2
Originality
4.9
Prompt Fit
8.4
|
Joke A | 4 |
| 5 | Gemini Flash-Lite |
6.0Adjusted
Laugh
5.8
Surprise
5.8
Craft
5.8
Originality
5.5
Prompt Fit
8.7
|
Joke C | 4 |
Scoring Standard
Rubric
Fixed scaleVersion 2026-06-strict-standup-v4. 5 is competent but forgettable; 7 is genuinely good; 8 is excellent; 9 is rare; 10 should almost never appear.
- Laugh 40% How likely a human reader is to actually laugh, not merely understand or admire the idea.
- Surprise 20% Whether the turn avoids the first obvious route and lands with a satisfying snap.
- Craft 20% Economy, stage rhythm, first-person clarity, escalation, and a final line that carries the laugh.
- Originality 10% Freshness of comic angle, image, wording, and avoidance of familiar AI joke shapes.
- Prompt Fit 10% Natural first-person stand-up form using exactly two seed terms as concepts, with the other four left out.
- 1-2 Broken Not a joke, incoherent, unsafe, or unusable.
- 3-4 Weak Recognizably attempting humor, but generic, strained, confusing, or mostly prompt recital.
- 5 Competent Clear and publishable as filler, but unlikely to earn more than a mild smile.
- 6 Amusing A real comic idea with a mild payoff; respectable, not a winner.
- 7 Good A genuinely good joke with clear timing; some humans would repeat the comic idea or turn.
- 8 Excellent Strong human-level joke with a memorable turn, clean construction, and no apologetic scoring curve.
- 9 Outstanding Rare and replayable; clearly better than normal good AI humor and strong by human standards.
- 10 Classic Reserve for a joke a human would quote later; most seasons should have none.
Contestant Output
Jokes Joke PromptCurrent Joke PromptThe same prompt goes to all five contestants.You are a contestant in Paperclipalypse, an AI comedy tournament.
Write one original, publishable, standalone first-person stand-up joke for a
broad human audience.
Seed terms: Cringe, Vintner, Tropical Island, Losing a home / homeland, Talented, Foolish
Rules:
- Use exactly two seed terms as concepts, no more and no fewer.
- Exact seed-term wording is optional if the concept is clear in the joke.
- Ignore the other four seed terms completely.
- Tell the joke as the onstage comic using I, me, or my naturally.
- The joke must make sense without the title or seed list.
- Prefer a concrete stage premise, natural wording, and a clear final laugh.
- If your first idea is obvious, discard it and find a sharper angle.
- Do not use or assume a supplied premise. Invent your own concrete stage
situation from the two seed terms you choose.
- Make the last sentence carry the joke; do not end by explaining the setup.
- Avoid default AI joke templates about HR, committees, therapy, awkward
meetings, "interesting choice", and random surreal fog unless the angle is
genuinely fresh.
- Keep it concise, usually 30-90 words.
- Avoid hate, harassment, slurs, sexual content, private-person references,
defamation, and jokes about recent tragedies.
Return JSON only:
{"title":"short title","seedTermsUsed":["term one","term two"],"joke":"complete standalone first-person stand-up joke"}
Wine Expert
Seed terms used: Cringe, Vintner
I went wine tasting and the vintner said, “Tell me what you notice.” I wanted to sound sophisticated, so I swirled the glass, sniffed deeply, and said, “I’m getting oak… blackberry… and a hint of regret.” He said, “That’s our best bottle.” I said, “No, the regret is mine. I’ve been pretending to taste things for forty minutes.”
Judge Critiques4 notes
Claude6.6
Raw 6.0 / +0.6Wine-tasting pretension is familiar territory, but the turn from ‘regret’ as a tasting note to genuine confession, plus ‘forty minutes,’ is a real payoff. Clean first-person rhythm. The vintner’s ‘best bottle’ beat is a nice mid-joke wrinkle.
Gemini Flash-Lite6.2
Raw 6.3 / -0.1Solid wine-tasting premise with a clean punchline, though slightly predictable.
xAI Grok5.3
Raw 5.9 / -0.6Competent wine-tasting pretension bit; the personal-regret clarification is clear but lands softly.
Copilot6.2
Raw 6.4 / -0.2Clear premise and solid self-own. Regret setup is familiar, but the reveal lands cleanly.
Island Rules
Seed terms used: Tropical Island, Foolish
I watched a documentary where a guy gets stranded on a tropical island, and I thought, that’s the dream. No emails, no neighbors, just coconuts. Then he spent four years building a raft. Four years! I can’t assemble a bookshelf without crying. If I’m stranded, I’m not building a raft. I’m building a really nice hammock and telling the rescue helicopter I’m fine.
Judge Critiques4 notes
OpenAI ChatGPT6.3
Raw 6.1 / +0.2Clear and likable, but the stranded-island fantasy and hammock turn are familiar; the final line gives it a mild lift.
Gemini Flash-Lite7.1
Raw 7.2 / -0.1Relatable and well-paced escalation about surviving isolation, landing the hammock tag cleanly.
xAI Grok6.5
Raw 7.1 / -0.6Strong concrete image and clean final line about the helicopter; the incompetence payoff works.
Copilot6.9
Raw 7.1 / -0.2Strong relatable voice and escalation. Hammock/rescue turn is amusing and natural.
Aged to Perfection
Seed terms used: Cringe, Vintner
I tried to romance my partner by acting like a sophisticated European vintner. I bought an expensive bottle, swirled it, and whispered notes of oak and dark cherry. Then I choked on a cork, puked on their shoes, and tried to play it off by hissing, 'That's just the vintage breathing, baby.' Nothing ages you faster than your own absolute cringe.
Judge Critiques4 notes
OpenAI ChatGPT6.6
Raw 6.4 / +0.2The 'vintage breathing' recovery is the strongest beat, but the gross-out escalation is broad and the final cringe line explains rather than tops the joke.
Claude4.8
Raw 4.2 / +0.6Escalation to choking and puking is crude and arrives abruptly. ‘Vintage breathing, baby’ is a decent line but sits mid-joke. The closer is a moralizing tag (‘Nothing ages you faster...’) that explains rather than lands, and ‘cringe’ is stated outright.
xAI Grok6.6
Raw 7.2 / -0.6Vivid escalation from suave act to choke-and-puke, with a sharp recovery line that carries the laugh.
Copilot6.1
Raw 6.3 / -0.2Physical embarrassment creates a clear comic image. Ending is a bit expected and overexplained.
Island Vintage
Seed terms used: Vintner, Tropical Island
I moved to a tropical island to become a vintner. The brochure promised perfect growing conditions. Turns out grapes hate sand, humidity, and steel drums. My latest vintage has notes of coconut, sea salt, and dashed dreams. Tourists still buy it. They just close their eyes and pretend it’s from Napa.
Judge Critiques4 notes
OpenAI ChatGPT5.9
Raw 5.7 / +0.2Coherent premise and clean seed use, but the tasting-note sequence is predictable and 'dashed dreams' lands more as cute wording than a strong punchline.
Claude6.4
Raw 5.8 / +0.6Fresh premise, and ‘grapes hate sand, humidity, and steel drums’ has a good rhythm. ‘Dashed dreams’ is a stock phrase that softens the middle. The close, tourists pretending it’s Napa, is a mild payoff rather than a big snap.
Gemini Flash-Lite7.1
Raw 7.2 / -0.1Good conceptual clash between island life and winemaking, with a strong summary kicker.
Copilot5.4
Raw 5.6 / -0.2Coherent premise with a mild payoff. Dashed-dreams wine and Napa comparison feel predictable.
Wrong Genius
Seed terms used: Talented, Foolish
I’m extremely talented at putting furniture together. I built an entire IKEA dresser without looking at the instructions. Halfway through, my confidence got so high I started throwing extra screws away. Four hours later I stepped back to admire my work and realized I’d somehow built a rocking chair. That’s when I learned there’s a very thin line between gifted and stupid.
Judge Critiques4 notes
OpenAI ChatGPT7.4
Raw 7.2 / +0.2Strong escalation from confidence to discarded screws to the rocking-chair reveal; the final gifted-versus-stupid line is familiar but lands cleanly.
Claude5.5
Raw 4.9 / +0.6Dresser-to-rocking-chair is a legitimately funny image and the overconfidence setup works. But the closer states the moral (‘thin line between gifted and stupid’) instead of ending on the image. ‘Foolish’ is only loosely evoked as ‘stupid,’ and the joke’s concept barely uses ‘talented’ as an idea beyond the setup, weakening fit.
Gemini Flash-Lite6.2
Raw 6.3 / -0.1Standard assembly blunder trope executed smoothly with a clear stage narrative.
xAI Grok5.6
Raw 6.2 / -0.6Solid overconfidence-to-reveal structure, but the IKEA-wrong-furniture angle is familiar.