Being “Bad” -or- The Startup Lottery Ticket

Paul Graham continues his popular series of “How To Get Rich Quick on the Internet. And Fast!” essays with his latest entry, “Be Good”, in which he describes some startup attributes common among a sampling of successful internet businesses.

One such pattern, observes Paul, is that his selected success stories started with no real revenue plans beyond “get big and flip it”.

No subscriptions. No advertisements. No pop-ups. No interstitials.

They started more like a charity than a real business, just bringing good to their userbase with nothing asked or expected in return. At least at first.

This isn't a new position for Mr. Graham. He has long advocated the idea that you just need to worry about getting the eyeballs and you can figure out how to make money from them later. Or better still — reading between the lines — you can let the sucker you flip the thing to worry about the gritty details of how to monetize it, suffering the consequences if the userbase burns it to the ground in defiance of any revenue generating scheme.

This isn't a surprising position for Paul to support. It is entirely aligned with his micro-VC organizations' business model, which is to take young, time- and energy-rich grads, fresh out of the college mill, and bankroll them with a small investment (which they need because they won't be earning anything from their Internet baby anytime soon), and then cash in when/if they manage to flip it to a sucker that still buys into the many eyeballs model (a strict “No Returns” policy in effect.)

Given the small investment, only a percentage of YC Combinator's `fundees' need to hit the jackpot for the strategy to succeed, at least for Paul. He's playing house odds in this startup casino.

Paul provides some examples to demonstrate his position that the charity-that'll-make-you-rich approach is the winning strategy: Google and Craigslist. Incontestably successful companies, and most would be over the moon to experience a fraction of their success.

Let's take a closer look at these examples, and see how relevant they are to Paul's central theme.


Craig Newmark, the founder of Craigslist, started compiling a list of upcoming local events in 1995, publishing it to subscribers via a listserver. Later, after the list was well established and had a healthy subscriber base, he started publishing entries to a website, adding functionality to allow users to email directly to categorized lists.

Today Craigslist is an internet superstar, constantly ranking among the top 50 websites worldwide. It pulls in impressive revenue numbers through a model that Craig himself describes in this entry on Yahoo! Answers.

Craigslist is a rare survivor among thousands, tens of thousands, or perhaps hundreds of thousands of exceedingly similar lists/classified upstarts, most of them run ad-free and for free (sites were often ad free as a simple side-effect of the barriers to entry to hosting ads pre-2000. At that time it wasn't as simple as signing up for an AdSense account).

A Long Day CompleteCraigslist had the perfect, rare combination of the real-world personal connections of the founder, an ideal starting locality, userbase, and a progressive evolution that allowed it to build enough momentum that eventually the network effect took over, and you'd use craigslist because everyone uses craigslist, at least in some markets.

What You Can Learn From Craigslist: Craigslist is an extreme anomaly. Holding it up as an example of a path to follow is fundamentally akin to analyzing the number picking “strategy” of the latest lottery winner. It's also an excellent example of how corrupting, and falsely compelling, a survivorship bias can be. Countless sites have followed a close to identical path, failing miserably.


Google hit the scene during the portal craze. This was a period when every other search engine, failing to sufficiently profit off of search alone, started merging with a gangly bunch of dance partners. Excite hooking up with @Home, for example. To “leverage the synergy”, each quickly morphed into a “destination for all things” portal, fattening up their content until they featured a landing page absolutely packed from margin to margin with text, news, stock quotes, images, comics, horoscopes, etc.

Ads were almost invisible in this era of visual pollution, and were seldom the problem. Many users were using dial-up, so the content inflation wasn't just aesthetically deplorable, but it also made the search process a slow, unpleasant experience.

Sergey and Larry had been working on some algorithms for internet search during this period, and with the dot com boom peaking they managed to pull together an impressive million dollars in financing before even launching their beta website.

When they did finally get something online, they did the absolute minimal amount possible. Later they described the utter simplicity of the first version as a function of their lack of HTML knowledge: It was the best they could do, or cared to do, at the time.

They had differentiated themselves, however inadvertently, and it worked brilliantly. They had copious long-term financing (not “flip it to someone else” financing) before even launching, so they had no need to worry about making money immediately, using the website as a technology demo that would conceivably allow them to sell search technology and services to third parties and businesses.

While the quality of the search results got the Google buzz started, the dial-up bandwidth-friendly simplicity of their offering really won people over. Yet it was a simplicity that came primarily because the company only really had one product — search — and couldn't link to hundreds of other provided services.

When everyone else went heavyweight, the minimalism of Google got it a lot of attention among technology trend makers. That exposure on sites like Slashdot — amplified when the community learned that Google ran Linux — got them their next $25 million in financing, and the rest is history.

WishesWith their advertising initiatives Google took exactly the same approach, and when everyone else was using pop-up, high-bandwidth, obnoxious ads, Google zagged and had text ads. As others have adopted Google's text ad approach, Google has started adding in animated and full graphics ads to their docket.

What You Can Learn From Google: If you have an algorithm or technology that is sufficiently impressive enough to get a million dollars in financing before you've even left the drawing board, maybe you can take lessons from Google beyond “differentiating from the competition is good” (which is fairly obvious advice.)

The Cold, Hard Truth

The vast majority of “go big of go home” web ventures will fail. It isn't a meritocracy. Luck has a lot to do with it. There is little you can learn from the success stories unless you also learn from the failures, yet aside from the huge flameouts (which had to have enough success to even be notable), most failures fizzle out and disappear without a trace.

From the opposite angle, many “grow revenue from day one” websites have succeeded admirably (I just gave a local babysitter directory $39 for 3 months of lookups), albeit not without the “lottery ticket” quick payoff that a miniscule percentage of the winner-take-all players yield.

Ethics and Morals

It's a risky and disingenuous proposition to build a website, baiting a community, on one model — the “Good” and charitable model — and then switch the userbase once the founder's numbers get drawn. It's a scummy behavior to engage in, much less evangelize. It also might have the opposite effect than intended: I might not care whether a site like a social link voting site has a revenue model, but when I was considering photo sites I immediately discarded those that followed the Paul Graham business model, considering the risks too high: Either it would eventually be forced to flail about, obnoxiously trying different approaches at making it pay, or it would fold with a “Sorry We Got Bored Our Numbers Didn't Come Up” notice one fateful day. Instead I went for one with a sustainable business model, and haven't been unhappy with my decision (even if it did cost me a bit per year for all of the features).

Bits And Bytes

  • This entry was authored in emacs.
  • A quad-core is definitely your best choice, especially given the huge price drops just announced.
  • I called the whole Riya thing perfectly.
  • The standard for comments in code shouldn't be driven by the need to provide endless guideposts for incompetent programmers. If it describes something that should be obvious by the code, you're fixing the wrong problem (which can be either unclear code, or incompetent programmers, or both).
  • Most developers don't rely upon books anymore because the overwhelming majority of technical books are garbage.
  • Bits and Bytes was a brilliant educational program on TVO in the early 80s, and it is entirely responsible for beginning my love of computer hardware and software.
  • Nassim Nicholas Taleb explores survivorship and confirmation biases excellently in his books the Black Swan and Fooled by Randomness. While I was put off by his ego, and the expansion of a paragraph-worth idea into chapters, they're still great reads.