Why do the best startup ideas sound bad at first?

Most great startup ideas get dismissed immediately because human intuition is calibrated to the present, not to where the world is heading. The ideas that generate the biggest outcomes tend to live in an uncomfortable gap: they look like bad ideas to nearly everyone, yet they turn out to be genuinely good ones. Understanding why this happens is one of the most practically useful things a founder can internalize before choosing what to work on.

Intuition is tuned to the status quo

When you hear a startup idea for the first time, your brain instantly benchmarks it against the current world—existing behavior, current market sizes, the technology that exists today. This is exactly the wrong frame for evaluating ideas, because every transformative company was built on a world that did not yet exist when it was founded. The judgment 'that sounds bad' is really 'that doesn't fit the world I currently see,' which is a very different thing.

Paul Graham's observation in 'How to Be an Expert in a Changing World' cuts to the heart of this: ideas that look bad often look bad specifically because some shift in the world has just flipped them from genuinely bad to genuinely good. By the time consensus catches up to that shift, the opportunity window has usually closed. The founders who benefit are the ones who noticed the change early—often because they were living inside it, not observing it from the outside.

This is why domain expertise earned through genuine curiosity matters more than pattern-matching on what previous startups looked like. A founder who has spent years deeply inside a problem space has updated their internal model of the world more accurately than most. Their idea can sound crazy to observers whose model is six years stale.

The Venn diagram problem: good ideas that seem bad

Graham describes a specific logical trap in 'Black Swan Farming' that trips up even sophisticated evaluators. The best startup ideas occupy a narrow intersection: objectively good, but superficially appearing bad. The problem is that the much larger category—ideas that seem bad and actually are bad—looks identical from the outside. You cannot tell the two apart just by noticing the 'seems bad' signal.

Consider how Graham describes his first reaction to Facebook: a site for college students to waste time, targeting a niche market with no money to do something that didn't matter. That description fits a genuinely bad idea perfectly. It also, in retrospect, described one of the most valuable companies ever built. Microsoft and Apple, at founding, were describable in nearly identical terms. The surface features that make an idea sound bad—small market, frivolous use case, no obvious monetization—are often the same features that indicate the founder has found something real that others have overlooked.

For working founders, this creates a practical challenge: you cannot simply follow the feeling that an idea is good, because consensus validation means you're probably too late. But you also cannot take 'that sounds bad' as a positive signal without reasoning through why it sounds bad. The question to ask is always: does this look bad because it violates some permanent truth about human behavior, or because it violates an assumption about the present world that might already be wrong?

The best ideas often don't start as startup ideas at all

One of the clearest patterns across Apple, Yahoo, Google, and Facebook is that none of them began as deliberate attempts to start a company. They started as solutions to real problems the founders personally had—often technical scratches to an itch, side projects, or tools built for a specific community. Graham's point in 'Before the Startup' is that great ideas tend to be such statistical outliers that a founder's conscious, business-oriented mind would reject them if evaluated through that lens. The ideas that slip through are the ones that never went through that filter, because they were built before anyone stopped to ask whether they were viable companies.

This has a direct implication for how you should think about idea generation. Sitting down to brainstorm 'startup ideas' is almost guaranteed to produce mediocre ones, because you're running your imagination through a business-viability filter in real time. The better approach is to become genuinely expert in a domain that matters, work on real problems, and notice what you wish existed. The idea that emerges from that process will often sound unimpressive to others—because it's specific, it addresses a problem outsiders don't feel, and it doesn't map onto any familiar success story. That specificity is the signal, not the noise.

The example Graham gives of a developer who built voice-over-IP software just to avoid long-distance charges while calling his girlfriend in Taiwan is instructive. He wasn't optimizing for a big market. He was solving his own problem with the skills he had. The result was something that fit reality so precisely that it had genuine value—which is the actual foundation every durable company is built on.

Going through the motions is the failure mode that looks like success

There is a counterfeit version of startup idea validation that produces the opposite of what founders intend. Graham calls it 'playing house'—assembling all the visible symbols of a startup (plausible idea, funding, office, headcount) while skipping the one thing that actually determines whether the company works: building something people genuinely want. This failure mode is particularly dangerous because it feels like progress for a long time, and the feedback loop is slow.

Ideas that go through this process often sound reasonable on the surface. They're not too weird, they map onto existing categories, and they're easy to explain in a pitch. But 'sounds reasonable' and 'solves a real problem' are almost uncorrelated. A founder who is selecting ideas based on how they'll land in a room—with investors, advisors, or peers—is optimizing for the wrong signal entirely. The ideas that are easiest to defend in that context are precisely the ones that are least likely to be genuinely novel.

The antidote is not contrarianism for its own sake. It's a rigorous commitment to whether the problem is real and whether real people in your target group urgently need the solution—not whether the idea sounds impressive. That test tends to surface ideas that will strike most observers as too small, too weird, or too specific. When that happens, it is worth examining carefully whether the skepticism reflects a genuine flaw in the idea or simply the gap between where the world is and where it's going.

“Great ideas tend to be such outliers that your conscious mind would reject them as ideas for companies.”

— Paul Graham, source

The one thing to do

Before dismissing a startup idea because it sounds bad, identify exactly what assumption makes it sound bad—and test whether that assumption reflects a permanent truth or a temporary state of the world that may already be changing.

Frequently asked questions

Does 'sounds bad' mean an idea is actually good?

No. The majority of ideas that sound bad are genuinely bad. The point is that the best ideas also sound bad initially, so dismissing an idea purely on first impression is unreliable. You need to reason through why it sounds bad—whether that's because it violates a permanent truth or a temporary assumption about the world.

How should founders evaluate ideas that seem weird or too small?

Ask whether the problem is real and urgent for a specific group of people who currently lack a good solution. A small, vivid, genuinely painful problem beats a large, diffuse, mildly inconvenient one. Weirdness and apparent smallness are often signs of specificity, which is a feature not a bug.

Why do deliberately 'brainstormed' startup ideas tend to be worse?

Because the act of brainstorming for startup ideas runs every possibility through a business-viability filter in real time, which systematically prunes the outlier ideas that become big. Better ideas typically emerge from deep expertise and personal experience with a problem, not from a whiteboard session.

What's the relationship between domain expertise and idea quality?

Deep expertise gives you an accurate, up-to-date model of a specific part of the world, which lets you see problems and shifts that outsiders miss entirely. The component of startup success that is hardest to fake or acquire quickly is genuine understanding of the domain you're working in.

Sources

More playbook answers · Growth Prophet home