Dick Costolo is the CEO of Twitter. Ben Horowitz is general partner at the VC firm Andreessen Horowitz [3]. Costolo became CEO of Twitter after the departure of its former CEO, founder Ev Williams. Andreessen Horowitz specializes in helping firms whose founders want to succeed as CEO. In November, at Fast Company'sInnovation Uncensored conference in San Francisco, the two talked about how startups succeed and fail.

Ben Horowitz: You came in as the outside CEO of Twitter. It's the biggest thing that you have run; it's giant. And as a result, you've brought in people who are radically more senior than what you had been used to. What's harder and what's easier about managing super-senior people?

Dick Costolo: The harder thing about it is that they have success biases. Our CFO, Ali Rowghani, was the CFO at Pixar. So he reported to Steve Jobs and Ed Catmull... I'm sure I tell him a few things and, you know, the back of his brain is going, "That's not what Steve would have said."

So that's a challenge. And when I want him to do something a different way, that's a discussion. With a junior person, I ask them to do something a certain way and they're like, "I got it," and they leave the room.

Horowitz: But how do you know you're right?

Costolo: Well, the great news is I don't. I don't have to be the smartest person in the room. There are a bunch of management team meetings where somebody on the team will say, "Dick, what do you think?" And I'll say, "You know, I really have no idea, so you guys are going to have to keep talking about this for awhile. If it's still not resolved, I'll make a decision. But I really don't have any idea on this particular issue. So hopefully you guys can come to some sort of agreement." You can't do that all the time.

Horowitz: That's the knock-wood-management take.

Costolo: I try to make sure, because we've got a really strong senior team now, that I don't go against the will of the team too frequently. If we are discussing some issue and I go into the room thinking, Oh, the answer is definitely A, but as we start to discuss it I discover that Ben [Grossman, head of global operations] thinks B and Ali thinks B and the platform engineering lead thinks B... Well, you don't want to too frequently say, "Okay, I hear what you guys are saying but we're gonna do A"--because then those guys start to tune out.

Bill Campbell has a great story along this line. Bill is, as you know, involved with Google and Apple and was the CEO of Claris and the CEO and chairman of Intuit [4]. In the early days of Claris, he would say, "Well, we need to do this, we need to do that. Nope, nope, you guys are wrong; we're gonna do this." And Donna Dubinsky came to him and said, "Hey, listen, if you're gonna make all the decisions, we'll go back to Apple. You clearly don't need anyone this senior." So I try to make sure that even if I go into the room disagreeing with these guys, if I start to hear from a bunch of senior people that "We really feel A and we feel about it this way," I'll set my thinking aside and say, "All right, let's go do that." And then I have to leave the room and commit to that decision. I can't say later if they were wrong, "Well, I thought Y but those guys wanted to do X." There's no better way to undermine your own leadership.

Horowitz: That looks particularly bad on a CEO. I couldn't help it! They overrode me. You don't get the privilege of being liked when you're CEO.

Costolo: Yeah. I'll tell my managers, "Look, managing by trying to be liked is the path to ruin." And they'll say, "Well, does that mean I should just tell someone, 'You can't work on project A, you idiot, how many times have I told you this?'" And I say, "No, you can be empathetic--you just can't be sympathetic."

You can say, "I understand you want to work on project A. It's just not a priority right now." You can't say, "I know you want to work on project A, Dick just really doesn't want to make it a priority." Don't sympathize with them, right? You can empathize; don't sympathize.

Horowitz: Yeah.

Costolo: Okay, I have one for you. You and Marc Andreessen and the firm talk about how you want to give the founder/CEO the ball and let them run the show. It's not going to work 100% of the time. So how do you guys think about that when it's not working? What do you do?

Horowitz: That's a really, really hard question. The big value of the founder running the company is really two things: the knowledge and the commitment. When you found a company, you have the original vision, you make all the original decisions, you know every employee, you kind of know every aspect of the product architecture and its limitations. You know the team and their limitations; you know the market and the customer feedback. That's a giant body of knowledge, which is very difficult to transfer.

So our general philosophy is, Well, we'd really like to take that person and help them learn the CEO skills to run the company. The tricky thing is, what if they can't learn the skills, or what if they don't want to learn the skills? And that's hard to tell because nobody is good at being a CEO. Nobody is actually a natural CEO. You learn the job, and the job is hard and weird and awkward and unnatural and...

Costolo: ... And psychologically damaging.

Horowitz: And psychologically damaging. You feel like you want to throw up all the time.

Costolo: I remember reading this book by Bill Walsh, the former coach of the 49ers. Jack Dorsey and I recommend books to each other once in a while; he had recommended this one to me. And [in it] Bill Walsh talked about when he first took over as coach and he is working his ass off and trying to make the 49ers successful. And they go on this road game, and the placekicker misses a last-second field goal, and they lose the game. And Walsh is on the team flight on the way back to San Francisco and he breaks down and starts crying. And the rest of the coaches are like, "Don't let any of the players come up here and see the coach."

Horowitz: That's exactly the challenge. You see the founder burst into tears. Maybe not in front of you, but you know he's crying because he's got the tracks on his face. Is that the reason to replace him? You may be replacing Bill Walsh, right? This is how it is. Sometimes the very best people take it the hardest.

For me, a lot of it comes down to when the founder loses so much confidence that he just can't even go to work; he's just raising his hand and saying take me out. Then you're in the position of "Okay, now we've gotta find somebody to replace him."

A lot of people want the guy who gets the trains to run on time; it's like, let's get somebody who's a real, professional manager--a been-there, done-that guy.

Those people don't tend to work out that well because they'll often shove the founder to the side and seek to put their own imprint on the company without the knowledge or the moral authority. What generally happens is they'll maximize so it will all be great until you need a new product--and then the company dies. We've seen this over and over again in the tech industry.

So what we look for is somebody who can mind-meld with the founder and get what is the knowledge, what is the culture, what are the things that they feel they did wrong, what are the things they did right. So the outsider will have that ongoing very, very deep connection to the history of the company.

My model for that was always Bill Campbell at Intuit. He and Scott Cook [cofounder of Intuit], you couldn't even tell where one of them stopped and the other one started. Without Cook, Bill wouldn't have been nearly as successful, and without Bill, Scott wouldn't have been nearly as successful.

So when you really understand these things, you can make the change. But it's a very, very high-risk transition.

Costolo: Yeah. You want to avoid that person who comes in and says, "Step aside, son, I'm gonna show you how to do this."

One of the things I try to do with my senior leaders who have been brought into the company, like Ali from Pixar and Chris Fry from Salesforce and Adam Messinger from Oracle--guys who have worked with Larry Ellison and Marc Benioff and Steve Jobs--I say to them all the time: What would Benioff, what would Ellison do in this case, right? Adam tells this great story: If we had two engineering leaders who disagreed about something at Oracle, we would always want to make sure we agreed about it before we took it to Larry, because if we didn't agree when we took it to Larry, he was liable to make some completely arbitrary decision that neither of us wanted. I mean, he's speaking with a little bit of hyperbole.

Horowitz: Yeah, that's actually a very effective CEO technique, management by terrorism. Either you decide or I will destroy the company.