InformationWeek Stories by InformationWeek

InformationWeek Stories by InformationWeek http://www.informationweek.comInformationWeeken-usCopyright 2012, UBM LLC.2004-03-29T14:00:00ZTorvalds: Open Source Keeps People Honest The father of Linux talks with InformationWeek about data-center Linux, trust, and SCO Group's lawsuits. http://www.informationweek.com/news/18600136?cid=SBX_iwk_related_news_Executive_insights/interviews_global_cio<img src="http://i.cmpnet.com/infoweek/982/CV_linus_bc.jpg" hspace="0" vspace="0" border="0" align="right"> <img src="http://i.cmpnet.com/infoweek/spacer.gif" width="10" height="96" hspace="0" vspace="0" align="right">   Linus Torvalds, the "father" of Linux, is an expert in, and exponent of, open-source development. InformationWeek recently exchanged E-mail with him. InformationWeek: Can you give us a quick update on your work at the Open Source Development Labs? Where do things stand in the effort to develop "data-center Linux"? Torvalds: My work personally is very simple: My sole responsibility is literally to "maintain the Linux kernel." We were very careful indeed to make it clear that different OSDL projects do not actually in any way impact my maintenance, and as such I'm totally independent of other OSDL projects, like the data-center Linux project. Same goes for Andrew [Morton, who maintains the Linux kernel]. (Aside: OSDL doesn't do just the big "data center" kind of things. Sure, "data-center Linux" is one project, but so is "carrier-grade Linux" and also "desktop Linux"). That was to make sure that people can still see me and Andrew as "neutral" developers, and there wouldn't be any worries about the agendas of big companies involved with OSDL. That said, obviously a lot of the OSDL projects are useful to me (and would be useful to me regardless of whether I worked there or not, since the results are public). So the testing farm with scalability testing, compile statistics, etc., [does] end up influencing me simply by existing. And these things are obviously literally part of the data-center work. InformationWeek: Linux and Apache are examples of open-source platforms that have become firmly established in business IT environments. What do you see as the next wave of open-source platforms to earn the trust and widespread use of IT professionals? Torvalds: I think it's pretty clear that the open-source projects are "building out" on top of existing infrastructure. Right now the databases are obviously starting to come of age, and much of the usage there builds on top of successes of open source in the kernel and Web space. But on a different note, I think it's interesting how a lot of the infrastructure is building up in other areas, too, especially the desktop. So while things like MySQL are obviously setting up in a big way, I'm still interested in the desktop. InformationWeek: JBoss and MySQL are the names of open-source platforms and of companies that help develop and then, for a fee, support those platforms. So you see any potential problems with this model? For example, how open is a team of developers whose leadership and key contributors are all inside one for-profit company? Don't open source projects that go this route begin to look less open and more like commercial software companies? Torvalds: I have to admit that that is one particular set of problems I personally have always tried very hard to avoid, but on the other hand, I also suspect that, especially in markets that are pretty focused on commercial needs anyway (and things like databases certainly would fit that), it may just be inevitable and possibly the best model to keep in touch with the needs of your customers. And the open-source aspect is still rather important in one major way: It keeps people honest. With an open-source license, if you start doing the nasty things that commercial software companies are so well-known for (looking out for No. 1 rather than trying to really help the customer), somebody else just comes along and captures the market. And that is one really important part of open source: no technical barriers to market entry, and the fact that you can trust the process, even if you might not implicitly trust the developer. So, while I personally have always opted for trying to be in a position where people really have no reason to distrust my motivation and actions, in the end I actually think that the real trust comes from the fact that it doesn't matter if people trust me (or any other open source developer) or not. Because if we are shown to not be trustworthy, somebody else can always replace us--so you don't have to be able to trust us. (I harp on trust, because I think that's pretty much the most important ingredient in any relationship, whether it is commercial or social. The trust that you won't be back-stabbed is something we're all looking for, isn't it?) InformationWeek: We presume that you, too, are now a salaried open-source developer. Is that accurate? And if so, has that influenced your perspective on these things? Torvalds: It's accurate, and yes, the concerns that situation raises is why it took me over a decade to get into this position. And even then I chose to try to make it as obvious as I possibly could that I'm still an independent agent ("stubborn as hell," as most bosses would say). The nonprofit position of OSDL helps--it clearly has strong commercial interests, but at least it's one step removed from the most direct influences. And I already mentioned my contract. In the end, most people tend to trust me just because they've seen the track record. And even when they don't, we get back to the previous answer about the process being more important than any particular person. InformationWeek: You've been an outspoken critic of SCO Group's threats and lawsuits against the Linux user and developer community. Legal arguments aside, has the open-source community learned anything from this whole episode about the need for more checks and balances in how open-source code gets vetted? Torvalds: I think our "vetting" has been pretty good all along, but there has certainly been some discussion on explicit documentation of that situation, so we can possibly go back and point to it more easily. The open development model already makes it pretty well traceable. We've been very successful indeed in tracking down the sources of various pieces of the kernel as SCO has been doing their PR thing, and I'm happy with just how quickly we've been able to totally debunk every single silly claim SCO has had. But we might make that existing implicit accountability even more explicit. I actually think that the most important part of the SCO lawsuits has been an appreciation of how the commercial and technical interests work together. Now, as you're probably aware, most Linux developers (including me) have always been pretty enthusiastic about having commercial entities involved and doing the parts that a lot of technical people just can't get interested in (i.e., marketing, customer support, etc.). But there [have] always been the "hard-core" tech people who just found the company involvement distasteful. And I think the SCO case has made it clear that it's been very useful to have big commercial companies involved, because they do not just [do] marketing and customer support, but they also have lawyers and are able to fight the FUD on that side. So I think we've seen how well the symbiosis between commercial interests and the technical interests of open source can really work. And that, I think, is fundamentally important. Also, groklaw.net has obviously shown how the open-source ideals end up working in the legal arena, too, and I think that has been very useful and made a few people sit up and notice. InformationWeek: Is there anything about the open-source development model, or its intersection with the world of commercial software and services, that is worrisome to you? Or that needs to change in some way? Torvalds: I'm not the worrying type. Worriers never get anything done, they just worry about what can go wrong. What gets closest for me is obviously the issue of software patents. It's only tangentially about open source (the biggest news lately was obviously the browser embedding patent war), and much more about broken legal processes, but since I definitely don't worry about the technology, "broken legal processes" is, I guess, the thing to look at. Photo of Linus Torvalds by Timothy Archibold <CENTER>Return to the story: <A HREF="http://www.linuxpipeline.com/trends/18600091">Open Source, Part 2</A></CENTER> 2003-12-01T09:00:00ZQ&A: Gates On Linux InteroperabilityIn an InformationWeek interview, Microsoft's chief software architect says Microsoft is already doing plenty to promote interoperability between UNIX/Linux and Windows.http://www.informationweek.com/news/16401118?cid=SBX_iwk_related_news_Executive_insights/interviews_global_cioMindful of the past and optimistic about the future, Bill Gates opened the Comdex trade show in Las Vegas this year with an overview of Microsoft's efforts to deal with some of today's most-pressing IT problems. In his Nov. 16 speech, Microsoft's chairman and chief software architect talked about anti-spam, security, and search software, and then something new: seamless computing. This excerpt, from the larger InformationWeek Q&A with Bill Gates, titled <a href="http://www.informationweek.com/story/showArticle.jhtml?articleID=16400867&pgno=1" target="_blank">Bill Gates Talks Seamless Computing, Security, And Linux</a>, focuses on the parts of the conversation that addressed Linux and open-source interoperability: InformationWeek: I'd like to switch the discussion to open source and Linux. InformationWeek recently ran a story that questioned whether Microsoft is doing enough to support compatibility between Linux and Windows. Do you feel Microsoft is doing more than people realize? Will there be more? Gates: Tell me any area we're not doing enough. I mean seriously, what area do you think we could do more in? We run UNIX shell scripts. We run UNIX libraries. We've got this whole group in India that we UNIX a group of customers that decided to go from expensive UNIX to Windows, and they wanted to take their shell scripts and their programs and run them. No rewriting, no changing, just compile and run. So we took on that challenge, and we have lots of customer case studies where we made that attractive. We've had UNIX as a primary element in the data center for most of the history of Microsoft. So, supporting NFS, supporting all of these things related to UNIX, we're extremely good at that. InformationWeek: Services For UNIX has been part of your product family for years now. Understood. But will you go beyond what you've done, to make it work more closely with the open-source community in some way? Gates: I don't know what you mean. We take Solaris, UNIX applications, and allow them to recompile and run on top of Windows. Name a UNIX protocol we support every one of those UNIX protocols. If there's something more you think can be done, tell me what it is. I mean, this is UNIX, it's not something new and different. InformationWeek: One thing that comes to mind, it may be tangentially related, is the Microsoft Communications Protocol Program, the licensing of communications protocols for interoperability. Some people seem to think that by now there should be more licensees. Gates: Name a corporate customer who has some interoperability thing they want from us, because we have interoperability. At every level, we're building interoperability. And so you've got to map it back to some customer [issue]. Believe me, if people thought there was some customer demand for using our protocols, they would license our protocols. The fact is, there are lots of these things out there that create the connections. InformationWeek: It's not just an interoperability issue. There are other things like Microsoft's management tools being able to manage a heterogeneous environment for example, Windows and UNIX/Linux at the same time. That is something, according to our research, that your customers would like to see. Gates: Management tools from us? It's not like there's a shortage of people who do that.