I’ve Seen This Before, It Didn’t Work

Substack, September 12, 2023, by D’Arcy Coolican (a16z). Routed here by Kyle 2026-09-01 from a LinkedIn share of the same piece.

The failure mode

The phrase in the title is the most common dismissal in venture, and Coolican’s argument is that it is a pattern-recognition error dressed up as experience. Old ideas resurface because the conditions around them change — technology, distribution, demographics, cost — and matching a current company to a dead predecessor without asking what moved treats a historical outcome as a law.

His example: “Midjourney looks a lot like the 100 digital design startups that failed” — and it worked, on generative models plus Discord distribution. Others he lists: micropayments, group investment accounts, digital influencers; Chewy against Pets.com; Frank against Lending Club and Prosper.

The three questions

He does not tell founders to ignore the objection — he gives them a test:

1. Has this failed before?
2. Why did it fail?
3. Am I genuinely different?

And the line that resolves it: “Don’t ignore them, but don’t let their baggage stop you.”

The inversion worth keeping

The subtler point is that familiarity can be an asset rather than a liability — “consumers favour things they know.” An idea that has been tried has already done the work of explaining itself, so a version that fixes the actual reason for failure starts with a market that understands the pitch. That is the opposite of how “we’ve seen this before” is normally used.

Connections

• The anchor source for Pattern Recognition as a concept page — most material there is about pattern-matching as a skill; this is the case for it as a liability, which is the harder and more useful half.
• The three questions are a usable diligence filter, and they connect to Timing and to The Hyperlegibility of Venture Capital: if everyone is running the same pattern-match against the same history, the mispricing sits exactly where the pattern is wrong.