Marc Andreessen, Act II

Any company that's going to be around for more than two years is going to go through the whole cycle. Oracle, Sun, Microsoft, IBM -- all the big names go through it every few years. If you have that perspective, and your employees expect it, you'll withstand it just fine. At Netscape, we were always suspicious, because the environment was so euphoric in 1995 and 1996. We knew we weren't that smart. But everyone has to go through it once. One of my favorite examples involved iVillage. Right at the top of iVillage's stock price, its CEO, Candice Carpenter, said some insulting things about Netscape. And I said, "The wheels of justice are going to turn on this one."

So what's a better way to keep a handle on reality -- and not get swept up in giddiness that might ultimately be self-destructive?

I keep a list of the 10 most serious threats to the company. The lawyers hate this. It's actually called "Ten reasons we're going to go out of business." It definitely focuses the mind as much as the prospect of imminent hanging. So we ask ourselves, If we end up failing, what will be the reason? And we go through the list. You're starting to build in what seems like paranoia, but it's really just clear-eyed objectivity. I always enjoyed it. In fact, working on the risk-factor section is my favorite part of drafting prospectuses. It lets you look at your company and say, Okay, what might go wrong? You have an obligation to be as complete as you can, so it liberates your mind.

Let's talk about some of the myths of doing business on the Internet. A year ago, there was an enormous amount of rhetoric about first-mover advantage. Does that still ring true, or is a different dynamic at work?

It's a tricky, tricky game. Most first movers end up lying facedown in the sand, with other people coming along and learning from their mistakes. Who knows what people are ready for in their lives? Are you willing to carry around an email pager? Are you willing to buy things from strangers via online auctions? Or spend an hour a day online? A few years ago, the answer to those questions was no. Today, the answer is yes. Being the first mover with the right approach is very important. Being the first mover with the wrong approach means you're dead.

What's your current thinking about the get-big-fast mentality? Three or four years ago, there was this belief that the opportunities were so great, you couldn't be too aggressive. What changed?

The key question is, What were the great opportunities? That strategy worked when the opportunities really were there, say for Amazon, AOL, Yahoo!, or eBay. But it wasn't true for just about everybody else. Especially in the past 24 months, people really lost sight of whether there was any real hope or prospect of a sustainable business. The truth is that some of these ideas may ultimately work -- they just came to market too soon. I wouldn't be at all surprised if there was a wave of new startups over the next 5 years that are basically reincarnations of all business-to-consumer ideas that are currently going out of business.

What have you learned about staffing a startup? How fast can you increase your head count before you regret the way you did it?

We hired very fast at Netscape. We ended up with some groups filled with supergeniuses and others that weren't. It was very dependent on who the managers were. If you hired the right manager, that particular group was going to be great. But if you happened to hire a bad manager, that whole place was going to be horrible. We call it the Rule of Crappy People. Bad managers hire very, very bad employees, because they're threatened by anybody who is anywhere near as good as they are.

But the hidden danger in startups is not hiring bad people; it's hiring good people instead of great people. Competent managers can usually screen out bad people, but they have a hard time screening out good people.

So how do you fix things?

At Loudcloud, we address that challenge head-on. We have full-time recruiters on staff, and we hired them very early. We say, "Good isn't good enough; it's got to be great." And we have a rating scale for each candidate. The trick is that people don't get hired if they are just above average. People get hired only if they are at the very top.

What's your thinking about "Internet time"? What actually takes place faster because of the Web, and what still creeps along?

Software evolution goes much, much faster. In the pre-Internet software industry, we all got used to these two- or three-year product cycles. Companies would spend two years building a piece of software in the lab, and then they'd run it through this elaborate internal testing process. On the Web, it's different. You build software, put it on your Web site, and it's deployed globally. So Yahoo! and eBay and Amazon can iterate their software environments every couple of weeks. And then the world does your beta testing. You can do market research pretty much instantaneously. When I was at Netscape, Bill Campbell used to say, "Put up a new service on the Net, and if it works, it's a product. If it doesn't, it's market research.''