Three core members of the original Apple Macintosh team are building the next generation GUI for Linux. And reinventing the open source business model along the way.
|PHOTOS © RUSS CURTIS|
A legendary user interface design guru at Apple, Andy Hertzfeld got his first look at Linux in the summer of 1997 while visiting science fiction writer Neil Stephenson. Little did he know that three years later, he would be working late nights with some old friends from his days at Apple, developing one of the most ambitious components of the Linux OS — the Nautilus file manager. Hertzfeld’s company, Eazel, hopes that Nautilus will eventually become the platform for a new generation of easy-to-use Internet-enabled services and applications and that their GUI know-how will help bring Linux to the masses. Linux Magazine‘s Publisher Adam Goodman and Executive Editor Robert McMillan met with Hertzfeld, Eazel CEO Mike Boich and Hertzfeld’s former boss and Eazel CTO Guy “Bud” Tribble in Palo Alto recently.
Linux Magazine: Who came up with the idea for Eazel?
Andy Hertzfeld: It was mine, more or less. I got interested in free software in January of 1998 when the Netscape announcement caught my attention, and all the press surrounding that led me to Eric Raymond’s set of papers.
I had been exposed to Richard Stallman’s ideas before and I had kind of rejected them. But when I read The Cathedral and the Bazaar in January of 1998 I saw it as a potential solution to the morass that the software industry had fallen into. I was chagrined that the users hadn’t been well served by the industry over the last ten years. The balance of power had rendered them pawns. The platform vendors had all the control. Innovation had stagnated. The personal computer revolution was founded on idealism and it had turned into this quagmire of unhappy customers.
There was one company sucking profits, maybe two companies. I thought: “Boy, the GPL [the GNU General Public License] allows the common infrastructure to be shared and owned by the community. That fixes all the problems.” Potentially anyway. There are new problems created, but it could be a much firmer basis for an open, fair, and free software industry.
So as soon as I understood that, I thought: “This is something I should work on.” The next frontier for the open source software was usability. In almost every other dimension — like robustness — it had exceeded proprietary alternatives, but usability was lagging behind. This was not surprising, because it was a labor of love from a bunch of programmers who just weren’t used to thinking of the mainstream end-user. Since my forte was usability, I thought perhaps I could make a difference.
LM: What was the reason for choosing to work on the GNOME rather than the KDE windowing environment?
Hertzfeld: I think that GNOME has a better architecture when it comes to the component model, which is very important to what we’re doing with Nautilus. But beyond that, KDE seemed to be trying to clone Windows. That was unappealing to me. GNOME has a more open-ended, innovative feeling of trying to transcend Windows, of trying to do something new. Later, as we started meeting with people, I just fell in love with Miguel de Icaza [GNOME's lead developer] and some of the other GNOME developers who had just terrific personalities that I really related to.
LM: It seems that when you’re talking about usability, what you’re really talking about is having the lowest common denominator beta tester be able to somehow contribute to this open source methodology. With the projects such as Linux and Apache, the users were actually very technical people.
When you start talking about desktop software like Nautilus or Mozilla, the users are less technical. So how do you leverage open source then, when the users are not going to be able to contribute? If I’m using a browser and something goes wrong with it, I can tell you about my problem, but I can’t necessarily fix it myself.
Mike Boich: To the extent that you’re doing something that can be used by a really large number of users, it attracts a correspondingly large number of technical people. So what we’re doing is becoming a high visibility project and, as a result, we have a fairly large number of technical people interested in what we’re doing.
Bud Tribble: In the early days of any technology, you have this phase where the toolmakers are making tools for their own use. If you go back to the very early days of Silicon Valley that’s the story. What was Hewlett Packard? Hewlett Packard was engineers — electrical engineers — making tools for electrical engineers. They were very good at that because the feedback loop was very short: one person. They could ask, “Is this a good tool for an engineer? Well I’m an engineer. I can tell you that, right?”
However, as technologies mature, they go into this phase where the toolmaker is making tools not just for himself but actually for a broader set of people. That’s the area where I get excited, I think that’s what excites Andy. That’s what drove the Mac team, I think that’s a lot of what’s driving Eazel. It’s the open source Linux world expanding beyond the point where toolmakers are making tools for themselves and making tools for the broader world. And it takes a certain type of aesthetic or empathy to do that.
The very earliest musicians in human history made their own instruments. They made flutes out of bones. They manufactured them themselves and made the holes the way they wanted them to play their music. But as civilization matured, you got to the point where you had Stradivarius. Stradivarius was a violin maker who wasn’t necessarily the best violin player of his day. He’s not known for being a concert violinist, but he was making tools for a broader audience — for everyone who wanted to play violins.
I think Linux is at the point now where it needs to broaden beyond the toolmakers making tools for themselves, which is where its roots started, to the toolmakers making tools for a broader audience. In the case of Eazel it’s for the broader consumer audience.
LM: What does Eazel need to do to make tools for the rest of us?
Hertzfeld: The biggest single thing that needs to happen is end-user testing. We didn’t know this so much when we were working on the original Macintosh. Your empathy with the user is important, but where the rubber meets the road is the user sitting down in front of the application and seeing if their intuition is correct: Where do they find it easy to use? Where does it not work for them? It’s not rocket science. You need to just watch them. See where they stumble, and fix it, and repeat and repeat and repeat until it gets smooth.
LM: I’m wondering what Steve Jobs thinks of what you guys are doing.
Hertzfeld: I bet if you asked him he would say we’re doomed because Open Source is not going to work, we won’t get any revenue, we are a start-up company, and the kinds of things that need to be done to build a great user platform require hundreds if not thousands of developers. Steve has a sort of self-rationalizing way of looking at the world.
Tribble: I think he would say that this argument that the users ought to be allowed more freedom over their interface would be philosophically wrong. I don’t want to put words in his mouth. But I will tell you that I actually think that both philosophies are pretty interesting. You kind of have to pick one or the other and go with it, right?
LM: So how will Eazel make any money?
Hertzfeld: One of the feelings of the company is that selling software is an outmoded notion. When you buy software you’re really buying a service. The program does something for you every time you use it, and you’d like it to do this better over time. If you [expect software to] do that, you’re going to want a set of services that’ll help you keep your system configured and well maintained, and protect the integrity of your system management data. Some of those services we’ll provide for free and some of them will be based on user capacity. So we’re a services company.
LM: Can you give me some specifics on what that means?
Hertzfeld: Well imagine we have a software catalog — that’ll probably be free. So you can go and get a much more painless application download and install than you might get today because the shell running in your system understands what’s already there and what libraries are already there. If you wanted automatic updates or notification of updates or something like that, maybe that would be a service we’d charge money for.
LM: So you become the interface to any software purchases that might take place?
Boich: But not on an exclusive basis. Another example of a service that seems pretty straightforward is automatic backup. Your data needs to be backed up on a regular basis, but it’s kind of a pain in the neck to do. Ideally your data has to be backed up off your site. So that seems to be the type of thing that people would be willing to pay for if there’s an update site that is going to be continuously available.
LM: So which services will you actually be offering first, and which of these are planned for the future?
Hertzfeld: We are resolving some of those issues right now, but I think the software catalog is the first service.
Boich: Initially the software catalog might not have every program in the world in it, but over time we’d like to be very comprehensive.
LM: How will it work?
Tribble: Well, what the user will see is the applications they want to run and which libraries or OS distribution dependencies they have. There would be a value added service to say: “Hey. You want to run this application, but you’ve got to get up to date on your distribution or library first.”
LM: Will you support different releases of the distributions? Will I be able to use this with Red Hat 4.0?
Hertzfeld: The database on our back-end is going to try to be as comprehensive as it can be. Initially we might be limited in terms of what we can support. We might say we’re only supporting Red Hat 6.0 or later. I’m not saying what we are actually going to support, but we might make trade-offs like that.
Certainly we might not bother to support Red Hat 2.0 at first. We’ll make our choices aimed at covering the broadest range of users with the minimal effort initially. But then over time we’ll extend things.
LM: When the PC invaded corporate America, it was because it had killer apps –spreadsheet, word processing, and publishing applications that made work easier. What do you think will be the killer apps that drive Linux into corporate desktops?
Tribble: My personal opinion is that the Linux killer apps are going to be the Web-based applications. Web-centric applications that we have not even thought of yet are going to start showing up as the corporate equivalent of Napster. You’re going to find it penetrating sales and marketing because they’re just totally throwing the lever over towards the Internet type of applications.
Take Internet chat systems for example. Those are going to show up on Linux first. Look at the demographics of chat with teenagers. Linux is going to penetrate that market. It’s going to then grow into the college market then grow into the corporate market. It’s going to be the same kind of grassroots phenomenon but instead of desktop publishing and spreadsheets it’s going to be Internet based stuff.
Hertzfeld: A completely different dimension that’s worth mentioning is this: If I were a CIO and I had roughly comparable alternatives, would I rather deploy the one that cost me $50 per unit or zero dollars per unit? That’s pretty compelling too.
Tribble: I actually will differ on this with Andy having come from Sun Microsystems, where most of our customers were CIOs. CIOs, I think, will be blind-sided by this. I don’t think CIOs will be drivers. I think they’ll be blind-sided in the same way they were blind-sided by the PC.
LM: So what will people be getting for free with Nautilus then?
Tribble: We’re not sure exactly.
Boich: Nautilus is part of the GNOME project so certainly any software the user gets will be free. The software’s a great shell and a great file manager. If you never connect to the Internet or you never connect to our service, you have a much better thing than you have with GNOME today.
So we’re kind of paying our dues and even having said that, we think it would be viewed as politically incorrect or some kind of Trojan Horse if we immediately started trying to sell you only services you had to pay for. So we believe in this vision where connectivity to the Internet and services kind of flow transparently into the desktop. We see it as a good thing. So some of it is free — like the software catalog.
LM: Do you think you’ll have revenue sources other than customers paying you directly for services?
Boich: Conceivably e-commerce. As you mentioned, if we have this great mechanism for downloading and selling software, we can be a channel for software distribution.
Tribble: Three years ago at Sun, one of the things I did was to try to convince Sun that software bits were going to be free and that the only added value was charging for services associated with it. It didn’t quite take hold. It got partial hold.
Every software company in the business today is facing this dilemma of charging for the bits. They’re just throw away because the bits are replicable; they want to be free. The real business model has to move to charging for the services associated with those free bits.
You’re going to see the day when PhotoShop is a completely free set of bits. You’ll be able to download PhotoShop and run it for free but, by the way, it’ll have menus in it. Some of these menus will say “Print me up some 11 by 18 glossies of this thing and FedEx them to me.” Of course, there will then be a charge for that. Some of that money will get siphoned off to Adobe.
There’ll be word processors where there’s not only a spelling checker bit or spelling checker menu but there’s a menu that says: “Please apply editorial judgement to what I just wrote and charge me some money for that.” If it’s the editor at The New York Times or The New Yorker that does the editorial review, the charge for that might be kind of expensive.
Hertzfeld: Or you can even imagine a little icon that puts out a little request onto the network. For example, “I need an editor in the range of these parameters.”
Tribble: And people bid on it.
Hertzfeld: A price appears and you say: “Okay.”
Tribble: I strongly believe that this is the future of the software world. That the bits of the software come for free but the services associated with that are what gets charged.
Hertzfeld: That’s really a re-capitulation of The Cathedral and the Bazaar.
Tribble: Imagine PowerPoint. PowerPoint is free. It doesn’t cost you $400 anymore. PowerPoint comes for free but it’s got a menu item that has a list of all these PowerPoint companies that will come and do your slides for you. Well you can do the rough version of your slides and you can select one of those companies on the menu. It goes off to some person who makes it look pretty.
It’s a look pretty menu. “Make my PowerPoint look professional.” They’ll charge me $30 a slide to do that.
LM: What kind of services do you think will be the first to be integrated into Nautilus?
Tribble: Well I’ve got a file full of my favorite MP-3s. I want to make a CD I can play in my car. I don’t want to necessarily go through the trouble of getting a CD writer. What about if I had a menu item that just said: FedEx me a CD tomorrow with the contents of this directory formatted up as a CD I can put in my car CD player for $5 bucks, $10 bucks, $15 bucks. Choose your price.
LM: Have you guys talked to any of those kinds of service providers?
Tribble: We’re just at the start of that. Another one is this: I’ve got a file full of all the photos of my family. I want a photo album printed up or I want it in frames sent to me. Just keep on going from there.
LM: So you’re describing the killer apps we were talking about.
Tribble: But they’re really killer services.
LM: How long do you see it being until GNOME and Nautilus are as easy to use as the Mac?
Hertzfeld: I would say it’s going to take at least a couple of years. Two to three years.
Tribble: When we came out with the Mac, people would ask the question: “Can I get rid of my DOS and just use a Macintosh from day one?” Actually the answer was really “No.” I think Eazel is a story where the underpinnings of the Internet provided service together with Nautilus can carry it leaps and bounds beyond what users expect to get today from their computers.
Hertzfeld: It’ll take about two years to reach full flower and that’s probably optimistic. If you just look at the PC, it came out in August of 1981 and it was at least two years before the sales took off at all. It’ll be great to have this conversation again in a year or two and see what we think then.
Robert McMillan is executive editor of Linux Magazine. He can be reached at email@example.com.