Tarpit ideas: what they actually are and when they stop being traps

Executive overview

Founders keep applying with the same ideas repeatedly, and those ideas keep failing. This isn't because the ideas are bad — it's because they share specific characteristics that make them attractive to everyone but workable for almost no one.

A tarpit idea is not just any bad idea. It's an idea that attracts constant applicants, sounds great to friends, has been tried many times before, and has never worked — with nothing having changed to make it suddenly viable.

The key signal: if your idea gets universal praise from friends, treat that as a warning, not validation.

What makes an idea a tarpit

• Must attract many founders repeatedly — not just an idea Dalton and Michael happen to dislike
• Universally praised by friends ("I totally have that problem")
• Has been attempted many times before without success
• Nothing in the technology or market has fundamentally changed
• Often one of the first ideas people generate when brainstorming startups

The two tarpit flavors

• Utopian behavior change: the idea only works if many people simultaneously adopt a new habit or app — e.g. coordinating hangouts with all your friends via a dedicated platform
• Mercenary arbitrage: chasing a momentary opportunity to get rich fast — e.g. building an investment advisory startup because you're winning on WallStreetBets and friends want in

How AI changes the calculus

• LLMs mean classic tarpit ideas may no longer be traps — technology can unlock previously impossible solutions
• The right approach: explain specifically why new technology makes the problem solvable now
• The wrong approach: ignore that many people have already tried and failed
• "Copilot for X" ideas (lawyers, surgeons, real estate agents) are common but not tarpits — not enough attempts yet to qualify
• YC funded an AI-generated podcast company that would have been unfundable pre-LLM

How to tell if you're in a tarpit

• Ask yourself: has this been tried many times before with no success?
• Ask yourself: has anything changed that would make it work now?
• Have you talked to users? Founders who haven't talked to users and want a tarpit verdict are missing the point
• Too much positive social feedback is a red flag, not a green light
• If it feels like "why doesn't this already exist, it seems so easy" — be skeptical