Paul Graham: Bottom-up Thinker

My first few posts were devoted to the proposition that bottom-up systems work better than people think. I argued that people systematically underestimate bottom-up systems like evolution, wikipedia, free software, the blogosphere, and the market process. Obviously, the other side of this coin is that people over-estimate the efficacy of top-down systems. I’ve hinted at this conclusion in my posts about cell phone “app stores,” but it’s a general problem that extends far beyond the iTunes store: Large bureaucracies are wasteful. Indeed, I think hardly anyone appreciates just how inefficient they really are.

Paul Graham

Paul Graham

At some level, everyone knows that large bureaucracies are inefficient. We’ve all dealt with the DMV or our cable or insurance companies. We know that large, bureaucratic organizations tend to be slow-moving and unresponsive to customers. Many of us also have friends who work for the government or large companies, and we’ve heard stories of the kind of waste that goes on the inside of these organizations. But while I’ve always known in the abstract how inefficient large companies were, it’s only recently that I appreciated how large this effect is. And the writer who most influenced my thinking on the subject is Paul Graham.

Graham is an interesting guy. During the mid-1990s, he founded one of the first web startups, a company called Viaweb that helped people make online stores. Before the end of the decade, while he was still in his 30s, he sold Viaweb to Yahoo! and walked away with a large enough fortune to finance a comfortable retirement. Since then, he’s branched out into a variety of eclectic pursuits. The two that are relevant for our purposes are: he started writing essays, and he founded a company called Y Combinator.

Silicon Valley is widely known as the leading center of innovation in America and maybe the world. Indeed, it’s a cliche (one I’ve employed myself) to suggest that we’d be better off if the rest of America worked more like Silicon Valley. But of course like any large community, the Valley isn’t monolithic. There’s a lot of innovation going on there, but there’s also a lot of variation. Some parts of the Valley are a lot more innovative than other parts.


Graham has written a number of essays highlighting just how much variation there is in the efficiency of Silicon Valley firms, and just how important size is to this variation. Two kinds of Silicon Valley entities are particular targets of Graham’s ire: large companies and venture capital firms. Graham has an interesting vantage point because his firm, Y Combinator, funds extremely early-stage startups. I think people outside the technology industry tend to use the term “venture capitalist” as a catch-all term for people who fund startups, but it actually has a more specific meaning. Venture capitalists occupy the high end of the startup-funding spectrum. They tend to be investing other peoples’ money, and they prefer to do relatively large deals. Google, for example, raised $25 million from a venture capital firm in 1999. Graham’s firm is at the extreme other end: he finds 20-somethings who often haven’t even started a company and writes them their first checks, typically on the order of $20,000. In between YC and the VCs are “angel investors.” These tend to be successful entrepreneurs investing their own money, and they often make much smaller bets than VCs. Andy Bechtolsheim was an angel investor who famously wrote the Larry Page and Sergei Brin a check for $100,000 when they were just starting Google in 1998.

Graham’s hostility to VCs flows from two factors: because they’re investing other peoples’ money, they’re extremely risk-averse, which creates a variety of problems. They tend to be unduly impressed with long resumes, since investing in someone with credentials is easier to justify if the investment fails. They’re prone to herd behavior, because, again, investing in the same way as their competitors gives them cover if their investments go sour. And they have a tendency to drag negotiations out as long as possible, in hopes of waiting until you either succeed or fail before committing their money.

The other problem with VCs is that, because their profitability tends to be proportional to the size of their fund, they push startups to take a lot more money than they need. Having too much money might not sound like a problem, but getting too much seed money too early can actually be deadly for a startup. First, once a firm has money, investors are going to expect it to “put the money to work,” which typically means hiring a bunch of people who may or may not be essential to the company’s success. And second, it dramatically raises the bar for the company’s success. Selling a company for $20 million is a great success if the founders started out with $100,000 of their own money. It’s not so great if the firm recently closed a $10 million venture capital round. So taking VC money means investors don’t have the option of cashing out quickly, they’re often forced to shoot for that billion-dollar IPO whether they want to or not.

In short, founders find angel investors easier to work with than VCs for two reasons: they’re investing their own money and they’re willing to write smaller checks. The moral of the story is that even at this small scale—companies with a handful of employees, investors putting in a few million dollars—firms experience dramatic diseconomies of scale. The larger the investment, and the greater the distance between entrepreneur and investor, the less efficiently the marginal dollar will be put to use. As we’ll see in my next post, things get even worse with big, publicly-traded companies.

This entry was posted in Uncategorized. Bookmark the permalink.

4 Responses to Paul Graham: Bottom-up Thinker

  1. To hear more from Graham, check out this interview he did for EconTalk:

  2. Why would investing other people’s money make VCs more risk-averse? Wouldn’t using your own wallet be scarier?

  3. Kevin,

    That’s a good question. I think there are two cross-cutting factors at work. On the one hand, as you suggest, people are going to be more inclined to take foolish risks with other peoples’ money, especially if they get a piece of the upside but none of the downside. But on the other hand, individual managers within a VC firm are going to be worried about their job security and reputation within the firm if they bet on losers. So you get ass-covering behavior: if a brilliant 22-year-old college dropout comes along with a clever business idea that seems like it came out of left field, a VC fund manager is going to be reluctant to say yes even if he thinks it’s a good investment. Because there’s a significant chance the firm will fail, in which case he’ll have to explain how he could have been so stupid as to fund a 22-year-old with a weird idea. In contrast, if a 35-year-old with a CS PhD comes along with an idea for yet another social networking site, an individual manager is going to be less worried, because (1) the entrepreneur has an impressive resume and (2) social networking sites are a respectable thing to invest in.

    So I’m talking about a very specific kind of risk-aversion: a tendency to overestimate candidates with impressive resumes and conventional ideas.

    This is related to another difference between VCs and angels, which is that angels tend to be successful startup founders, while VCs often have MBAs or other non-technical backgrounds. Somebody with a technical background and actual experience in a startup is going to be much better at telling the difference between a groundbreaking business idea and a merely weird one. Likewise, they’re going to be much better at judging how smart the founders are, since they’ll be able to ask hard technical questions. If you’re investing in an area you don’t know that well, you’re going to be less able to spot really breakthrough proposals and less inclined to take chances on out-of-the-box pitches.

  4. Brian Moore says:

    “Obviously, the other side of this coin is that people over-estimate the efficacy of top-down systems. I’ve hinted at this conclusion in my posts about cell phone “app stores,” but it’s a general problem that extends far beyond the iTunes store: Large bureaucracies are wasteful. Indeed, I think hardly anyone appreciates just how inefficient they really are.”

    I think a major reason this attitude exists, especially amongst engineers and scientists, is that almost everything in your life is presented as a top-down problem. Everything in k-12 education is presented this way. The school system itself is organized this way. When you’re young, your family is organized this way. It becomes much worse when you get to college (especially for those engineers and scientists) because you get presented with ever-more complex top-down problems that you must solve. And since engineers and scientists are smart people, they’re pretty good at developing solutions for this.

    And so when they become voting, policy-interested adults, they apply (wrongly in my opinion) the same methods to the problems they see the government as presented with. Just put the police here, put the right people here, ratchet the budget up there, refocus our efforts and tada, you’ll have a good solution. They do not particularly appreciate the difference between the engineering problems of their education and a policy question that involves autonomous individuals. In fact, I’ve seen the “let the people involved organize their own solutions” method derided as “lazy” — “but what if they choose the wrong solution? What then?”

Leave a Reply

Your email address will not be published.