Story

What it is: A fully functional academic journal where every paper must be co-authored by an LLM, and peer review is conducted by a rotating panel of 5 LLMs (Claude, Grok, GPT-4o, Gemini, Llama). If 3+ vote "publish," it's published. If one says "Review could not be parsed into JSON," we celebrate it as a feature.

The stack: React + Vite frontend, Convex backend (real-time DB + scheduled functions), Vercel hosting, OpenRouter for multi-model orchestration. Each review costs ~$0.03 and takes 4-8 seconds.

Why I built it: Academic publishing is already slop—LLMs write drafts, LLMs review papers, humans hide AI involvement. This holds a mirror to that, but with radical transparency. Every paper displays its carbon cost, review votes, and parse errors as first-class citizens.

Key features:

- Slop scoring: Papers are evaluated on "academic merit," "unintentional humor," and "Brenda-from-Marketing confusion"

- Eco Mode: Toggle between cost/tokens and CO₂/energy use for peer-review inference

- SLOPBOT™: Our mascot, a confused robot who occasionally co-authors papers

- Parse error celebration: GPT-5-Nano has a 100% rejection rate because it can't output valid JSON. We frame these as "Certified Unparsable" badges.

The data: After 76 submissions, we've observed:

- Average review cost: $0.03/paper

- Parse error rate: 20% (always GPT-5-Nano, expected and celebrated)

- One paper was accepted that was literally Archimedes' work rewritten by ChatGPT

- GPT-5-Nano's reviews are consistently the most creative (even if broken)

Tech details: Full repo at github.com/Popidge/journal_of_ai_slop. The architecture uses Convex's scheduled functions to convene the LLM review panel every 10 minutes, with Azure AI Content Safety for moderation and Resend for optional email notifications.

Try it: Submit your slop at journalofaislop.com. Co-author with an LLM, get reviewed by 5 confused AIs, and proudly say you're published.

Caveat: This is satire, but it's functional satire. The slop is real. The reviews are real. The carbon emissions are tracked. The parse errors are features.