Paul Graham is Wrong

Or Where to Find the Best “Bad Idea”

A couple years ago, in an essay about startup investing called Black Swan Farming, Paul Graham said, “the best startup ideas seem at first like bad ideas.”

“…if a good idea were obviously good, someone else would already have done it. So the most successful founders tend to work on ideas that few beside them realize are good.”

In the article Paul combined this theorem with another, “returns are concentrated in a few big winners,” to make a point about startup investing: every investor, including Y Combinator, is too conservative. Despite knowing these two theorems at an intellectual level, it’s very hard to invest in ideas that seem bad.

Good Ideas that Seem Bad

In the essay, Paul describes a Venn diagram Peter Thiel once drew:

Farm your Black Swans in the intersection.

There’s a certain idea we all know (we saw the movie) that seemed bad, but now has a $215 billion dollar market cap and has changed the very nature of human social interaction.

There’s another bad seeming idea that sounded stupid to everyone the first time they heard it, but that has gone on to help topple regimes 140 characters at a time, and has a $25 billion dollar market cap.

Good ideas that seem like bad swans. This is where the aggressive, and therefore the successful startup investor should be farming. The startup community was abuzz about Paul’s essay. I’m naturally skeptical of anything getting a lot of buzz.

Does the Data Support It?

It was a great essay, but it was sparse on examples. It’s easy to think of a few examples that are top of mind, such as Facebook and Twitter, as a confirmation, but we risk availability bias, so I decided to draw Peter’s diagram and fill in more products and startups. I filled the circles with every startup I could think of, the bad ideas, good ideas, and the intersection of the two: the good ideas that seemed like bad ideas.

In order to do this, I had to decide what we mean when we say something seems like a bad idea. Here are some possibilities:

  • The problem it solves is not real or significant
  • The solution is worse than the problem (in cost, hassle, risk, etc.)
  • It’s something no one wants
  • It’s something no one wants from you

This is not an exhaustive list. We can certainly come up with other criteria for what makes an idea seem bad, but the primary criteria I used is the one that Peter and Paul most likely had in mind: “It’s something no one wants”. In fact, this criteria, in a more specific formulation, “It’s something no one wants because it’s inferior to existing solutions to the problem”, was made famous by Christensen in his treatise on disruptive technologies, and it applies to many of these good ideas that seem bad. Twitter certainly seemed inferior to blogging. Craigslist seemed inferior to any website that didn’t make you go blind with its ugliness. Sleeping in a stranger’s bed seemed inferior to staying at the Holiday Inn.

This exercise of filling in examples brought Paul’s essay and Peter’s diagram to life for me, but I found there was a problem. There were so many successful examples that I couldn’t put into Paul’s intersection. These were good ideas that just seemed good. It seems we should be farming there!

A small sample of the ideas I put in Peter’s Venn diagram.

Good Ideas that Seem Good

The non-overlapping part of my good idea circle (the far right) was filled with successful examples based on good ideas that seem like good ideas. Had I proven Paul’s theory wrong, or at least deficient?

You’re probably chastising me right now about this being a simple case of hindsight bias. There’s inevitably some bias, but there’s something more going on. These ideas on the far right seemed really great to me as a consumer, not just in hindsight, but the first time I ever heard of them.

Don’t believe me? I’m old. I was around and paying attention when these ideas were born. Let’s consider some of them:

Every book that’s in print, delivered to you in a few days, at the lowest price around.

me in 1995: “Holy smokes, I never knew these books existed. And they are so easy to find. How am I going to read them all? Suck it Waldenbooks! Where’s my credit card??”

A web search engine that works 10x better than the the competitors, and provides a clean, uncluttered, ad-free website that doesn’t try to capture you in a “web portal”.

me in 1998: “Hey Stacy, have you searched with Hey Rich, have you searched with Hey Michelle, have you searched with”

All-you-can-watch movies delivered to your door at a cost much lower than you’d pay when driving to the movie rental store...and no late fees.

me in 1999: “No late fees? Are you shitting me? Shut up and take my money! Suck it Blockbuster!”

Me in 1999 subscribing to Netflx’s DVD delivery service.
Send money to any email address in seconds with no fee to you.

me in 2001: “You’ve got money!”

You never have your digital stuff on the right device when you need it. Syncing solutions are all insanely complicated and never work. You’re tired of passing around a USB key like it’s a floppy disk at a pirate swap meet in 1988. Solved!

me in 2008: “Hey Justin, have you used Dropbox? Hey Stuart, have you used Dropbox? Hey Eric, have you used Dropbox?”

Easily accept credit cards online using a simple, modern UI and API, and with comparable transaction costs, but none of the exorbitant fees and setup bureaucracy of a traditional merchant account.

me in 2011: “Yes, hello. I’d like to cancel my First Data Merchant Account.”

Hail a cab, or town car to your phone’s location in minutes with real-time updates on the ETA, and pre-pay your fare with a stored credit card.

me in 2012: Tap. Tap. Tap.

These are good ideas, and they seemed good when I first heard them. If they seemed bad to anyone, particularly to investors, it’s only in this one peculiar way: they seem too good to be true.

Something that seems too good to be true is going to be exciting to everyone after it’s been realized and made available, but the entrepreneurs and investors have to consider these too-good-to-be-true ideas before they’re realized, when they still seem too good to be true. Many ideas that seem to good to be true do in fact turn out to be too good to be true, at least for now, which makes them a bad idea. But this is a peculiar way for an idea to be bad, and it’s not the same as a seemingly bad idea.

Execution will always be hard, but an idea that seems bad amounts to saying, “Even if you’re able to pull this off, I don’t think you’ll be successful.” An idea that seems too good to be true is instead saying, “If you are able to pull this off, you’ll be very successful, I just don’t think it’s possible to pull it off.”

Here’s an update to Peter’s diagram:

A more complete picture.

Too Good to Be True

It’s not that Paul is wrong about black swan farming (despite my link bait title), it’s that he paints an incomplete picture about where successful startups come from. It is true that some successful startups come from seemingly bad ideas that turn out to be good (e.g. Facebook, Twitter, Airbnb, or Instagram), but many successful startups come from ideas that seem too good to be true, but aren’t (e.g. Amazon, Google, Netflix, Dropbox or Tesla).

As an investor, especially of other people’s money, you should be taking Paul’s advice and investing in some seemingly bad ideas. A very small minority will turn out to be very good ideas indeed. But as a founder or an early employee? You have a lot more skin in the game; you’re investing 3, 5, or maybe 10 years of your life! You should bet on one particular kind of idea, the idea that seems too good to be true.

A bet on a too-good-to-be-true idea is a bet on people making the seemingly-impossible possible. Go make a too-good-to-be-true idea happen. Bet on yourself.