Interview: Netscape's other communicator - Marc Andreessen

Interview: Netscape's other communicator - Marc Andreessen

It would be difficult to identify an IT industry figure who has risen to fame and fortune more quickly than Marc Andreessen (above). He was co-author of the pioneering Mosaic browser, who along with Silicon Graphics founder Jim Clark created Netscape Communications just over three years ago, now serves as Netscape's senior vice-president of technology and de facto chief communicator. Recently, on his way to give a keynote speech at NetWorld+Interop in Tokyo, Andreessen spoke at length with IDG's Don Tennant about his role in Netscape's phenomenal ascent, his view of the competitive landscape, and his best guess as to what's going to happen from hereIDG: If Jim Clark had never sent you that first e-mail to say he wanted to talk to you about starting a company, what do you suppose you'd be doing now? I presume it would not have been a "You want fries with that?" scenario.

ANDREESSEN: Working at 7-Eleven and living in a cardboard box? I don't know - I had the good fortune of getting a job in Silicon Valley after I graduated from school, so I was already out in California by then. I worked for a small company called EIT [Enterprise Integration Technologies], an early Internet-focused company which was later bought by VeriFone. So I probably would have stayed in Silicon Valley and gotten a job somewhere.

Silicon Valley has such a unique approach to starting and building and growing companies that it was a real eye-opening experience - especially with Jim. It's certainly a lot easier to start companies in Silicon Valley than it was in Illinois [where Andreesen co-wrote Mosaic, the forerunner of Navigator, at the University of Illinois' National Center for Supercomputing Applications] or other parts of the US. So I might have done that.

IDG: So you never had a plan to do a start-up around Mosaic?

ANDREESSEN: No, and there are two reasons for that. One is that in Illinois and in general outside of Silicon Valley in the US there's very little understanding of how to do that, and of course it's not something you typically think of doing. In the Valley it's the first thing you think of doing.

The problem for me was that I didn't initially see how serious this was all going to become in '94. When I started working in the Valley we had done Mosaic and the Internet was growing, but it was still considered a toy. Interactive television was going to be the future of everything. ATM was going to be the future of computer networks, and interactive television was going to be the future of consumer networks, and the Internet was sort of going to go away. But then interactive TV didn't happen, and ATM became Internet backbone technology, but was not a replacement. Then it started to become much more serious, but that wasn't at all clear earlier.

Jim's original idea was to start an interactive TV company, and we spent about three months trying to figure out how to do that, until we decided it really couldn't be done.

IDG: So when he first contacted you what did he say he wanted to do?

ANDREESSEN: It was 'let's do a software company' for sure, because he had done a hardware company [Silicon Graphics], and hardware companies are very capital-intensive and they take a long time to take off. And then we basically tried to figure out what kind of software to build. We knew we wanted to do something involving networking and consumer online services. If interactive TV was going to be the platform, then we would need to build software for that. What we did is we went out and looked at it from the standpoint of a start-up and realised that there was no market there - nothing, in fact, was happening.

Then we said the Internet is continuing to grow and all these smart people are doing all this Internet stuff, and although it's not being taken seriously right now, in a year or two if it keeps growing at this rate it will be.

IDG: Meanwhile, Mosaic belonged to whom?

ANDREESSEN: The code that we wrote was owned by the University of Illinois, which was then licensed to another company in Illinois after we actually started Netscape. So they went off and did their thing. The people involved with that made the mistake of assuming that they could take the code that we had written as undergraduate and graduate students and go sell that without really carrying it forward. And that was an error because the software market moves so fast, you have to keep moving.

IDG: What was Navigator's relationship with Mosaic?

ANDREESSEN: It was a complete rewrite. When we wrote Mosaic it was written as completely different versions for Windows, Mac and Unix - completely different code bases. It also had other problems - it required all the data to download before you could see anything. It didn't do streaming downloading, so it really only worked over high-speed networks. So we had to rewrite it to get cross-platform and to get acceptable performance over low-bandwidth networks. Those were the initial virtues of Navigator in the 1.0 version.

IDG: It's well known that you co-wrote Mosaic with Eric Bina, who also worked at NCSA and who is now a lead developer at Netscape. How did you and Eric divide tasks? Who did what?

ANDREESSEN: He did all the hard stuff. We sat down and decided what we were going to do, and we just coded intensively for three months. We basically divided it up ahead of time where he would do the layout rendering engine that would form the guts of it, and then I did the user interface around that, and some of the networking code. It was pretty much a 50-50 split. I couldn't have done it without him - he did, I think, most of the hard stuff. He's a better programmer by far - just a laser focus.

IDG: What gave you the idea to do it in the first place?

ANDREESSEN: The idea just came about because we at Illinois and a lot of other universities at that point in time - 1990, '91, '92 - took the Internet for granted. It was something that was really only used by Unix hackers. And most of the people on the Internet didn't think there was a problem with that - in fact, you wouldn't want most of the riffraff on the Net, that was the thinking at the time.

What we saw was an opportunity to unlock two things at once. One was to make it easier to use the Internet, and the other was to provide an incentive for people to create content that would be interesting. So we basically just did it in order to try to connect what had happened in the PC world with GUIs, with what was happening in the Internet world. This is something that software companies at that point - Microsoft, Lotus and others - were trying to avoid, almost consciously. They didn't like that idea - it was not a good idea because they were making tons of money selling proprietary e-mail systems, proprietary networking software. I don't know if they were consciously trying to avoid it, but unconsciously, I think they were. So it was sort of a void, and somebody was going to do it. The idea was hanging out there in mid-air.

In order for it to happen, there had to be somebody who was familiar with the Internet from the Unix hacker standpoint but really cared about having other people be able to use it. And there weren't that many of them at that point because most people didn't want other people to be able to use it. And since we were at a federal research centre we had lots of grant money and a place to do it.

IDG: OK, let's move to the competitive landscape. Whose technology do you have more respect for in the space that you're in - IBM/Lotus' or Microsoft's?

ANDREESSEN: More historical respect for Lotus'. Ten or 15 years ago Lotus had a lot of what's happening on the Internet now. The problem was, it wasn't on the Internet - it was a totally proprietary thing. And they ended up building it into this whole monolithic thing, so now they're having a lot of trouble actually delivering on opening it up. In the Internet space Microsoft has probably been more aggressive. But Microsoft's approach, as is their typical approach, is they're basically cloning what other people are doing.

IDG: IBM and Lotus sell lots of Navigator browsers with their software. But when I spoke with Lotus President Jeff Papows recently, he made the following statement: "If Netscape wants to unbundle the browser from [Communicator] Version 4 [the new Netscape client that will ship in June], which to this point they said they won't do, we'll continue to support it. But nobody at IBM or Lotus is going to ship Communicator. So they have a choice to make." Are you going to unbundle the browser from Communicator so IBM can ship it with their products?

ANDREESSEN: No comment. You know, Papows is about three levels down from the CEO. We have a good relationship with the rest of IBM and we're certainly going to try to continue that. I have no idea what the relationship with Lotus will be going forward. IBM sells a lot of our software today, so we're going to try to make sure that continues.

It's interesting why Lotus doesn't want to ship Communicator. What Communicator delivers is open e-mail and groupware - that makes Notes irrelevant. They want people to use the Notes client because then they're trapped to the Notes server. That's their strategy, that's fine. But it's certainly not what we would recommend the customers do.

IDG: Would you leave the door open to unbundling the browser so IBM can ship it with their products?

ANDREESSEN: We'll leave the door open to everything. In the long run, we'll do anything to grow the business. But we're about to launch Communicator, and the last thing we would do is announce something to upstage that. That's why the no comment.

IDG: Ray Ozzie, the creator of Lotus Notes, once told me that Netscape's software is 'architecturally unsophisticated'. What's your response to that?

ANDREESSEN: Lotus Notes has a hugely sophisticated architecture, which today is their undoing. One of the problems with Notes has been that it's too complex - the Notes development team historically has had computer science disease. They've been building the ultimate answer to everything. If you look at the Notes architecture it's extremely sophisticated and extremely monolithic, but it's overboard, it's extreme. It's got its own floating-point architecture - it does its own floating-point computations, it doesn't even use the chip for that. Weird stuff like that - it's got its own networking capability which in the age of the Internet is completely unnecessary. So their architecture is certainly too sophisticated, and that has made it extremely difficult for them to retrofit it. Our architecture is much more modular, much more adaptable, much more extensible. We've been able to add a lot more functions, a lot more capabilities - a lot more quickly.

IDG: Except your messaging products. Getting that stuff out the door has apparently been harder than you expected, given that you originally expected to get it out in the first quarter. What's been holding it up?

ANDREESSEN: We would have preferred to ship it in the first quarter instead of the second quarter. No software company ever ships software on time. Let's be frank - Microsoft has never shipped a software product on time. The difference between us and Microsoft turns out to be that when Microsoft slips, they slip in terms of years; we slip in terms of months. Exchange slipped, like, three years; we slipped, like, three months. We'll take three months. It's software - it's going through a very long and intensive beta cycle so we make sure that it works. But it will ship in June.

And we've been shipping some stuff early. We wrote Netcaster [the push-technology component within Communicator] itself in Java and we were able to deliver it sooner than we thought we would because it turned out to be faster to write that way. We didn't have to go through the huge cross-platform effort that we have to do on everything else - which is why we're going to do more of our future development in Java.

IDG: Do you have many former Microsoft developers?

ANDREESSEN: Oh, yeah - we hire people out of Microsoft quite a bit. The biggest problem in hiring people out of Microsoft is not attracting them to Netscape - it's getting them to move out of the Seattle area. So we will do a remote development centre at some point within the next 12 months, which we haven't done so far. And there's a good chance it would be in Seattle, for that reason. There's quite a pool of technical talent there.

Follow Us

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.
Show Comments