LLenny's Podcast
← All frameworks
InnovationKeith Rabois (Khosla Ventures)

The Ugly Baby Test

Seek ideas that make smart peers laugh — real alpha lives in the ugly babies everyone dismisses.

Difficulty
Moderate
Time to result
~ongoing to results
Steps
4
Confidence
85%

Great ideas start as 'ugly babies' — unattractive, unloved concepts no one wants to help. Rabois runs a deliberate contrarian filter on early-stage investments: he wants roughly half his VC friends to literally laugh at him, because he knows his competitors well and their laughter signals a genuinely contrarian bet where alpha exists. The idea must be contrarian at inception but become consensus for the company to win.

Origin

Rabois attributes the 'ugly baby' concept to Ed Catmull's book (Creativity, Inc., on Pixar) and applies it as his own investing filter; it echoes the pattern that early employees of OpenAI, Palantir, and Stripe report — ideas people laughed at.

Core principles

  • 01Every great idea starts as an ugly baby no one wants to nurture
  • 02If smart, well-informed peers laugh, that's a signal of real contrarian alpha
  • 03You must know your competitors well for their laughter to be a meaningful signal
  • 04A bet must start contrarian but become consensus — you want the inflection, not permanent obscurity
  • 05Consensus-obvious ideas have no alpha left in them

How to run it

  1. 1

    Run the laugh test

    Before committing to a contrarian bet, ask whether about half of your smart, informed peers would literally laugh at you for it. Aim for that reaction.

    Pro tip Calibrate it against people you know well and compete with directly — their laughter is only signal if you understand their judgment.

    Watch out Laughter from people who don't understand the space is noise, not signal.

  2. 2

    Protect the ugly baby

    Recognise that the best ideas look unattractive and unsupported early, and resist the instinct to kill or over-critique them prematurely.

    Pro tip Rabois notes his own parents laughed at almost all his tech jobs — dismissal by the uninformed is common around real opportunities.

  3. 3

    Require a path from contrarian to consensus

    Confirm the idea can eventually become consensus — everyone using and trusting it — because a contrarian take that never converts to consensus doesn't produce a winning company.

    Pro tip Airbnb started as a contrarian take and had to become something everyone in the world uses and trusts.

    Watch out An idea that stays contrarian forever with no inflection is not a good bet — you need the crossover.

  4. 4

    Refresh your contrarian takes

    Because good ideas get adopted and become consensus, deliberately exhaust old takes and generate new ones so you keep a supply of live contrarian bets.

    Pro tip When your old contrarian takes have all been proven right, that's the signal you need a new batch.

In the wild

Wanting VCs to laugh

When Rabois makes a seed or Series A investment, he runs an algorithm: will about half his VC friends literally laugh at him? Because he knows his competitors well, their laughter signals an ugly baby with real alpha.

The laugh test becomes a positive buy signal rather than a deterrent.

Generational companies people laughed at

Rabois notes, and Lenny's research with Terrence Rohan confirms, that early employees of OpenAI, Palantir, and Stripe describe ideas people laughed at and thought would never work.

The 'people laughed' pattern recurs across companies that became generational.

Common mistakes

Chasing ideas that already sound smart

Consensus-obvious ideas have no alpha left; if no informed peer would laugh, you're likely too late or too safe to earn outsized returns.

Mistaking permanent contrarianism for edge

An idea that never crosses into consensus doesn't build a winning company — you need the inflection from contrarian to widely adopted, not perpetual dismissal.

Is it for you?

Best for

Early-stage investors and founders deciding whether a dismissed, unconventional idea is worth backing

Not ideal for

Situations demanding near-term consensus validation or low variance, where a widely-mocked bet is genuinely too risky

From the transcript

I want half half friends who are VCs to laugh at me. Like literally laugh

1:01:00

ugly babies are the ones where there's real alpha.

1:01:30

every great idea starts as this ugly baby

1:01:00

eventually if the company's going to succeed, it has to become consensus

51:30

From the episode

Hard truths about building in the AI era

Keith Rabois (Khosla Ventures)