Show HN: FablePool – pool money behind a prompt, and Fable builds it in public

Original Article ↗ Hacker News Discussion ↗

A new site, FablePool, lets people pool money behind software ideas and then uses Anthropic’s Fable model to build the projects in public, raising questions about whether “crowdfunded AI development” can work beyond toy examples. Commenters compare it to a reverse Kickstarter, debate wildly optimistic budgets (like an “open source AWS” for a few hundred dollars), and argue over code quality, testing, and the need for human engineers to steer the work. Legal and governance issues also surface, including who holds copyright (if any) on largely AI‑generated code, how to manage security and liability, and whether pooled tokens or bespoke licenses could support more sustainable open-source-style efforts.

Concept & Motivation

  • Platform lets people pool money behind prompts; Anthropic’s Fable model builds the requested software in public, with code released under open source terms.
  • Seen as “reverse Kickstarter” or “GoFundMe for prompts”: backers fund a desired outcome, not a specific human team.
  • Some view it as a clever way to channel unused AI tokens / budget into shared tools and experiments.

Feasibility, Cost Estimates & Scope

  • Many criticize wildly optimistic budgets (e.g., “open source AWS” for a few hundred dollars; serious OS/database rewrites; HFT-grade GC fixes).
  • Even supporters call these “engineering theater” or “fanfic,” arguing you might get toy versions, not production-grade systems.
  • Others counter that Fable can wire together existing open source, but agree hard parts (scalability, security, IAM, hardware, reliability) are out of reach at these budgets.

Comparison to Existing Models

  • Compared to Kickstarter, open source bounties, Product Hunt, and “Supported Source.”
  • Some argue this may work better for creating new things than for long-term maintenance, but note decades of mixed results for donation-based OSS funding.

Trust, Governance & Rug-Pull Risk

  • Concerns about sending money to an anonymous, new site with little background, unclear refund rules, and “fantastic idea for a rug pull” comments.
  • Suggestions: stricter curation of projects, visible timelines and refund guarantees, more realistic examples early on.

Quality, Testing & Human-in-the-Loop

  • Demo projects show regressions and missing assets; some see this as proof of current LLM limits and “slop.”
  • Others say predictable, guaranteed progress (even if imperfect) is the appeal vs. uncertain human labor.
  • Several suggest mandatory human steering, detailed implementation plans for partially funded projects, voting on phases, and more iterative workflows.

Licensing, Ownership & Legal Ambiguity

  • Debate over whether AI-generated code is copyrightable at all, and if MIT is appropriate vs. CC0.
  • Distinctions drawn between authorship vs. ownership, jurisdiction differences, and the lack of clear precedent.
  • Some worry that claiming “we all own it” is not legally defensible; others argue paying for output should confer rights, but this is contested.

Ethics, Security & Liability

  • Concerns about funding harmful or cyber projects; mention that some prompts are blocked by Anthropic safety policies.
  • Debate over who is liable for AI-built artifacts: platform vs. funders vs. model operator; analogies made to GoFundMe/Indiegogo, with emphasis on moderation duties.

Platform Design & Implementation Details

  • Critiques of Google-only sign-in; requests for GitHub or plain email/password, and for avoiding large tech SSO on ethical grounds.
  • Reports of slow server responses and the demo project being removed/edited after criticism.
  • Suggestions to attach GitHub repos, potentially use blockchains for a verifiable ledger, or support crypto / token donations.

Broader Impact on Software & OSS

  • Some see this as “the new open source” or “vibe-coded crowdfunding,” potentially reshaping how small tools get built.
  • Others think it will mainly showcase what AI cannot do, likening many prompts to jokes or unrealistic fantasies.
  • Questions remain about whether anyone will meaningfully use the resulting code in production.