Linux for Suits - Dialogue with Don
This issue is Don Marti's last one as Editor in Chief. I recruited Don to the magazine, and I hate to see him go. Don brought an ideal combination of know-how, commitment, integrity, insight, creativity and humor—all of which sustained him through a tough period for Linux Journal, the computer industry trade press and for the Linux community as well.
Don was a smart and tough editor. He suggested many of the topics at which I've become expert. He spiked (that's journal talk for rejected) more than a few of my pieces, always for good reasons. And he always pushed me to do better work. I wasn't always happy with that (few writers are), but I'll always be grateful.
The last time the editorial staff was together, at LinuxWorld Expo in August 2005, executive editor Jill Franklin gave me a fun assignment: interview Don. So, with the help of Steve Gillmor (impresario of the eponymous Gillmor Gang podcast, as well as a veteran producer of recordings, going back to his days with Firesign Theatre), we recorded what will surely also be a podcast, timed to come out along with this magazine.
Doc Searls: How long has it been?
Don Marti: I've been at Linux Journal since 2000, and I've been Editor in Chief since 2002.
Doc Searls: When you came along, it was right when the bubble was bursting, and you came from VA Linux, which was the largest of the bubbles.
Don Marti: Yes. I jumped off the dot-com bubble right as it was popping.
Doc Searls: [laughing] We're at LinuxWorld (Expo) now, and the whole show was on cocaine back then, in a way. I mean, it was very high; there was nothing but a weird kind of gassy optimism.
Don Marti: Cocaine plus sushi and leather pants.
Doc Searls: So, I'm interested in your perspective on what's happened with Linux over the past four years. What did we understand well in the first place? What did we never quite understand?
Don Marti: Well, Linux made a lot of big promises like every one of the technologies that touched the dot-com frenzy. Linux was better than most at delivering on them. And, in the years since the dot-com boom, I think people have had time to fill in the necessary gaps and move Linux into more and more niches. Things like logical volume management, for example. And real-time improvements in Linux, and cleaning up the desktop, and getting more hardware support—just checking off those to-do list items, one at a time.
Doc Searls: Last night we had this documentation BOF. One of the guys there said that we've reached the point when it's even possible to put Linux on a random laptop and there's a fair chance it's going to work out. A lot of the behind-the-scenes work has made that possible.
Don Marti: One of the factors that helps account for that is the consolidation in the PC hardware market. Laptops used to have more weird bastard spawn hardware in them than they do today. With the introduction of USB hardware, you have a much smaller number of actual chips that your drivers have to talk to. Of course, through the same chips you're talking to everything in three aisles of the computer store, but the driver development for supporting all that can be saner and easier for more people to have a hand in.
When Greg Kroah-Hartman did an article for us on writing a driver for a multicolor LED blinky light device that plugs in to the USB port, he got a bunch of comments on that, including one from a developer who, before the next article in the series came out, had written his own USB device driver and gotten it into the kernel tree.
Doc Searls: How much have people reading and writing in places like Linux Journal—especially Linux Journal—had an involvement with the development of Linux?
Don Marti: Greg Kroah-Hartman again is a good example of that. He's now one of the top kernel people. Both through work and his own projects, he has become responsible for more and more of the kernel. He started off writing for Linux Journal in 2002. And, as he's gotten more responsibilities in the kernel, he's also written more articles for Linux Journal. Robert Love is another good example. And outside the kernel, many, many other contributors have both code that they maintain that's on the Linux CDs you get at the store, and also articles that they've written for Linux Journal.
Doc Searls: Yeah, it's always been interesting to me what role Linux Journal and journals in general have in a development ecosystem. What do you see as the future for Linux Journal and for magazines like that? At this point, it's a tough time for publications. We seem to have sustained a complete turnover of advertisers after the dot-com bubble—and managed to stay in business. But today so much more information is available freely on the Web. And we have a two- or three-month lead time. How can we stay current?
Don Marti: On the Internet, every movement looks like a big argument, and one of the things a print publication can do is pick a side and stand by a considered opinion. So, when Linux Journal comes out against something like proprietary device drivers, or when Linux Journal comes out and says that the directory server is one of the most important pieces of software in your organization to commit to open source and open standards, then we can take a consistent position on something like that and put together a set of articles that helps people succeed if they agree with us either in whole or in part.
Doc Searls: You were involved in our Embedded Linux Journal effort. What's the story with that, and with embedded in general?
Don Marti: Embedded Linux Journal was a controlled-circulation publication. And I think the idea of sending people a paper magazine for free, and that advertisers will pay to reach them, is sort of falling apart. I don't know how many of these controlled-circulation magazines you get, but it's something where the reader doesn't have a commitment in time or money to pay attention to this thing, and it ends up being one of the last things they get to. So, when Linux Journal has readers who are willing to pay for it and subscribe to it, I think that they're more likely to read it.
Doc Searls: I'm thinking also of the activity around Embedded Linux. Two years ago I had people telling me that the telephone OS market was going to come down to Java and Symbian. Now it's pretty clear Linux is going to be the big thing there, or one of the big things there.
Don Marti: Java as an application environment is still thriving on the cell phones. When you get a Linux phone, one of the features of that is a Java virtual machine, with the ability to install and run Java applications. But Linux certainly has a huge advantage for full-featured cell phones in that it's the very first OS that most of the hardware vendors develop drivers for. So that shortens the development time for manufacturers who want to get that hardware into a phone.
Doc Searls: Most of the developers that we run into at a place like LinuxWorld, or the O'Reilly Open Source Convention, are doing applications for computers, not necessarily for phones. And phones, even if they have Linux in them, are still silos. They're still closed things to some degree. Whereas a server you can make into anything you want it to be.
Don Marti: When you get cell-phone service, they give you a phone. And free as in cell phones is not something that I think of as a bargain, because that phone is strictly controlled by the carrier, who determines what you can and can't run on it. Part of that is the carriers' need to conform to regulations. And part of that is their business model. They want you to buy applications through them, rather than being able to download and install your own.
Doc Searls: Yeah, they want to enforce behaviors. Like, if you accidentally took a picture where you have no choice to just discard it, you have to either send or save. That's what my phone wants me to do. They get money for that, I assume, or they wouldn't force me to do that. But there is a sense that there is, for me at least, a kind of a closed environment. Does it concern you that Linux is often used as the base operating system in things that are inherently closed, like a TiVo for example? I mean a TiVo is a sort of a closed environment, and TiVos run on Linux.
Don Marti: A TiVo lawyer told me that the reason they have to be strict about video extraction is that they don't want to face a lawsuit from Hollywood. So, if you download and store a TV program in digital form on your TiVo, they do everything they can to make it difficult to get those exact bits off of that drive. You can record to a VHS tape, but you can't make a digital copy. And, like most of the other consumer electronics and IT vendors, I don't think TiVo is being 100% honest about big, bad Hollywood making them do this digital rights management. I think that there's a reason why IT vendors and consumer electronics vendors want to lock in their own customers and laying it all on Hollywood is not going to fly much longer. So, I'm concerned about devices that have lock-in built in to them, whatever OS they're on.
Doc Searls: You've said some interesting things about DRM in the past. For example, that all DRM is bad. You've gotten some push-back on that, but I'd like to hear what you mean by DRM being bad.
Don Marti: Cory Doctorow made a great distinction between DRM and CA or conditional access. When you sign up for a service and they tell you, “You must log in to view this content”, and you log in and then you can read and view, or cut and paste the information, that's conditional access. When you get a piece of content and it says, “Cut and paste are disabled”, or “Print is disabled” or “Read aloud is disabled”, then that's DRM. And DRM is deliberately micromanaging or removing the value from that information. It breaks some essential economic relationships that I think ultimately the authors of that information will be concerned about.
Doc Searls: If we had Hollywood executives sitting at this table today, saying they can't imagine any way other than DRM, what would you tell them about alternatives to DRM that would get them the same or similar economic benefit? That it's worth the trade-off?
Don Marti: That's a really good question. I think that a lot of the understanding that Hollywood has built up over many years of trying to understand the Internet is based on sales pitches from vendors who are pushing DRM systems. So, when a DRM vendor goes to Hollywood and talks to them about, “We can control this, we can lock out this, this will enable you to make money”, that really shapes the understanding of somebody who isn't in the technology business and who doesn't have the technical background. So, before I start spewing business ideas, I really want to listen to what the person understood to be the case about the technology and try to understand and fill in the gaps where the gaps are.
Doc Searls: This brings us to the cartelization of things. DVDs are encrypted, in their own way, because the cartel didn't want DVDs to run on any machine other than what they controlled or where they had a relationship. DVDs will run on Windows, on a Mac, but not on a Linux machine.
Don Marti: And there were other business-model-related restrictions that were built into the DVD format. For example, region coding.
Doc Searls: I never understood why region encoding was there. I mean, it's a hassle that doesn't seem to have an upside to me.
Don Marti: Well, imagine if a studio wants to release a movie on DVD in the US, when that move has not yet had its theatrical release in Europe. So, if they did not have the region coding system, then somebody might buy the DVD in the US and take it over to Europe and watch it and interfere with what has always been a classic Hollywood business model: show it in the theaters first, then wait a while, make it unavailable at all, and then release it on VHS and now DVD. And, interestingly enough, that model is being collapsed. Before the DVD format was decrypted, it was about a year from US theatrical release to DVD release, and within the past year or two, it's come down to about half a year. Hollywood wants to be able to play with business models, change who can see what when. So I think there is tremendous appeal that the DRM vendors are offering, saying, “We can control your audience, we can control the technology so that it fits with the business model that you want to try this year.”
Doc Searls: I became familiar a few months ago with Lucene. Doug Cutting who used to work at Excite, felt that keyword search was a done science, essentially. The result is some open code that anybody could use. Now anybody can do keyword search. Lucene isn't even a full product. It's one piece of building material. Last night we talked about Struts, which is another one of those kind of things. It's been sitting out there. So, one concern that I have is that Linux, as it becomes more like a foundation stone, disappears. It turns into the building, it becomes rebar and cinder block. Does that concern you? Or is that just a natural course of things? Should we pay less attention to Linux after a certain point and to the general construction business that Linux is a part of?
Don Marti: I think there are some lessons to be drawn from the history of the projects that are older than Linux and possibly more mature, as products, than Linux. And a good example would be GCC.
GCC for a long time was considered to be a good, stable compiler, capable of doing code for almost any processor out there. And, within the past few years, with a lot of the changes in the processor architectures and optimizations you can do for processors such as the Opteron, the need for ripping up and redoing parts of GCC has popped up. And, with things like the C++ standard template library, there's pressure on GCC on the language side as well. So, GCC is a piece of software that sits between the languages and the hardware. GCC was a stable, mature project, but as languages become more complex, and the number of languages people want to code in increases, and at the same time the hardware gets capable of doing hairier and faster things, then a mature piece of structure needs to have changes happen to it.
The same thing is going to happen with Linux, as hardware advances and the OS needs to be able to support more processors or processors in unusual configurations, such as the very many processors in a newer machine, or situations when you might have some processors on one die and some processors on another die, and the OS needs to be aware of which processors are where. As the hardware changes, the OS will need to advance, and as the applications that demand services from the OS change, the OS will need to advance. So, Linux won't entirely fade into the background unless hardware stops changing and the applications stop changing the way in which they use the kernel.
Doc Searls: Since we're on GCC, I know you're one of the folks who has a deep appreciation of Richard Stallman's role. I'm wondering....We've kind of gone back and forth on calling Linux “GNU/Linux” as Richard would like us to, and just Linux. Do you have a particular feeling about that?
Don Marti: The official Linux Journal policy on it is, “Leave it the way the author wrote it.” If someone wants to make clear in his or her article that the whole system is called GNU/Linux, then we leave that stand. If the author wants to say, “The name of my system is, say, Red Hat Linux”, that doesn't have GNU in its name and so we leave that name as it stands in the original article.
Where GNU comes in as an absolutely key project is as a many-year development effort to bring together a system that lets people do what they need to do, to communicate, to get by in the world of computers. As Richard Stallman himself put it, “So that I can continue to use computers without dishonor.”
And, the idea that when you click OK on that end-user license agreement, you say, “It is OK that I won't examine this piece of information that I have downloaded. It's OK that I agree not to change it or understand how it works, or explain it to someone else how it works.” I've come to understand that I don't believe that. And, I've come to a lot of that understanding through what Richard has written about the subject.
Doc Searls: To me what's so interesting about Linux and about the Free Software movement—and to the understanding of computing and software that goes back to the earliest days of independent computing—is what Richard was saying about the nature of software in the first place: that it was inherently free and wanted to be free more or less the way the wood and the pine tree wants to be free. He wasn't just talking about the economic uses of it; he was talking about the nature of the thing itself. And the feeling I have is that this is still not fully understood. Is that your sense as well?
Don Marti: Well, my sense of software is that it's something that is both speech and a device, depending on how you define it. When you talk about software as speech, many good things tend to flow from that. When you use software as a device you can get into great benefits and also fairly scary issues. So, the challenge is to apply the best of what our culture has developed for the real world to the world of software.
On both sides of the software freedom debate, people try to make analogies comparing software to real-world items. So when Bob Young says, “You wouldn't buy a car with the hood welded shut”, he's trying to make an analogy to a real-world object. When someone on the restrictive side of the debate says, “Well, you wouldn't walk into a store and walk out with a copy of the CD”, this person is also trying to make an analogy to a real-world item. It's a huge issue to understand the best of what we value about real-world goods and translate those values to the software world and the on-line world.
Doc Searls: As you know, I've been fascinated by the parallels between the construction industry and computing in general, including the software industry. In construction there is a very mature understanding of how things work together. Now, we've been sitting in this building. I'm sure this floor is a synthetic material and there is clearly some kind of sedimentary rock that's a surface over there, and behind you there is the huge corpse of a trunk of what appears to be a eucalyptus tree. It's not structural; it just graces the place as an architectural element. There's steel and terrazzo over here. So one of the things that fascinates me about construction is that it's full of open source. I mean, there are no secrets to making terrazzo. Yet there's still what we call intellectual property in construction. But none of it is in position to take control over everything else. I'm looking at a door over there. It's probably a standard door, but the latch on it may have some patents in it, and it may have a lock in it and that lock may have some patents as well. But you can replace that lock, right? And, I'm wondering if you can see a path toward that. I don't think we're at that point in software yet, where we have that same sense of modularity. Do you see us getting toward something like that in software? What might Linux have to say about that, being something like a natural material?
Don Marti: So far, the proprietary software vendors have really dropped the ball. On the free software side, Richard Stallman with the GPL has come out with a normative statement of a code of conduct for software developers and users. When someone releases software under the GPL, or chooses software under the GPL, the person is agreeing to those norms. If you want to talk about proprietary software becoming part of a mature market, or becoming a part of the useful structure, then there has to be some norm other than “all your base are belong to us”.
When you look at Larry Ellison's licenses saying, “Thou shalt not publish benchmarks and you have to click on this to agree to that”, that's not compatible with building a useful structure out of multiple materials or under multiple licenses. That's a trailer-park landlord's idea of city planning. So, really, when the proprietary software license writers decide to put as much thought in their licenses as Richard Stallman and Eben Moglen and the rest of the free software side have put into theirs, then we have some potential for that kind of innovation and growth. Until that happens, I think those who want to treat software as a mature product and a responsible market are not going to have much choice except for the free software side. So show me a responsible, innovation-compatible and integration-compatible proprietary software license and we'll see what happens.
Doc Searls: In looking back over your five years or so with Linux Journal, what great articles or achievements stand out for you?
Don Marti: I'm very proud that we did our 2.6 kernel preview very early in the 2.6 cycle, when it was still 2.5 development. That was when we let people who were doing Linux deployments and applications know, “Look, here's the great stuff coming along in the kernel.” That issue [May 2003] with Robert Love wearing headphones and the headline, “Are You Ready to Rock?”, that was the right issue at the right time to give 2.6 testing a nice kick. And, one article that I was so happy about that I had the authors do another version of essentially the same idea, was Craig Swanson and Matt Lung's “OpenLDAP Everywhere” [December 2002 and July 2005]. That company brought together the complete directory of services for all their clients, both Microsoft Windows and Linux, authenticating against it, sharing address books, using the file server and the intranet servers in a very compatible and customer-directed way. So, we, Doc, you and I talked about this, and came up with the idea of DIY-IT—largely influenced by a small company.
Doc Searls: I get a lot of credit for that, but that really came from you. There's the notion of smart companies using Linux to make themselves smarter. That was an assignment that really became my mission with the magazine. The observation that everything that happens with Linux starts with smart individuals doing smart stuff, usually without big vendor assistance. I'm not knocking big vendors at all, it's just that DIY-IT acknowledges that they're part of the ecology, not the origin of the ecology.
Don Marti: And when the vendor says, “there is no market for that yet”, that's something the customers should hear as “your competitors aren't doing that yet”. I think the next step, beyond DIY, is entrepreneurial IT. Where can you take those building blocks that are becoming large enough, stable enough, functional enough that you can get a lot of business value with very little integration work and staff time? How can you take those things and as an IT department create business value?
Doc Searls: I need to wrap this up by saying that I've been around Linux Journal from the beginning—and this is not a knock at any editors—but as far as I am concerned, you're the best editor we've ever had and it's been an honor to work with you.
Don Marti: Thank you.
Doc Searls is Senior Editor of Linux Journal.