You are currently browsing the archives for November, 2006
The open sourcing of Java by Sun has struck me. It seems the open-sourcing of existing proprietary software by vendors generally happens as a competitive move when it becomes clear the vendor’s product is falling behind the market leader(s). I’m not convinced this is always true – it’s a hypothesis that seems to hold true, but I’m looking for counter-examples.
Tons of start-ups are going open source right out of the gates, as they should. Open source is a lower cost/higher quality development process, and provides a more compelling value proposition for start-ups looking to break into the market. But under what circumstances should software already developed and marketed in a closed source proprietary IP regime be open sourced?
There is an incentive to open source a legacy closed source product when it becomes clear that it’s not going to capture the number 1 or 2 spot in the market. Open sourcing can potentially turn a fading product into a de facto standard, building a community advocating for its adoption, lowering the maintenance/extension development costs, and creating a rising revenue stream instead of a steadily declining one. The vendor has nothing to lose, since at some point continued proprietary development isn’t justified by the future revenues.
On the other hand, a company with a product that is a leader in its category has little incentive to open source. Most of the development costs are behind it. Granted, open source peer review would improve the quality of the software, but in return the developer has to give up all future revenues for sale of its proprietary software asset while opening the door to competition for the support revenues. It just doesn’t make economic sense regardless of the virtues of open source in the abstract.
Like I said, I’m not convinced this is always true, but some obvious examples come to mind, Java being the most immediate. Java was losing ground to Ajax, .NET/Mono et al. In Sun’s universally applauded move to open source it, Java has a shot at turning this around and becoming much more ubiquitous. Same with OpenSolaris, or going way back, Star Office. Not to pick on Sun, the open sourcing of Netscape as Mozilla fits the same pattern. The same with my employer, Novell, when we open sourced AppArmor to compete with Red Hat’s SELinux.
The reason this is of interest to me is the suggestion that Novell should just open source everything we own. If it were just Novell resisting this move, it would be a fair critique. But it seems to be universal that proprietary products that are viable in the market are not open sourced. Think of another open source advocate, IBM, who actively participates in open source projects but hasn’t open sourced Websphere, Notes or Tivoli, and isn’t about to as long as they are succeeding in their markets.
All this of course begs the question regarding NetWare, which unlike other Novell products, is in decline. I am not privy to any of Novell’s thinking regarding NetWare, but it’s a fair question to ask why it’s still closed source. But the rest of Novell’s product portfolio can’t be compared to NetWare, Java or Netscape.
The bottom line for me is that open source will succeed because of its economics, not because of the altruistic virtue of software vendors. The flip side of this is that it’s not realistic to expect software companies to open source proprietary code when it runs counter to their economic interests.
In the past week I’ve had two big surprises: first, the announcement of my employer Novell’s deal with Microsoft, and second, the strong criticism of the deal by many leading lights in the open source community. I suppose I shouldn’t have been surprised on the second point, although I do think the Novell execs did the right thing. Herewith, some thoughts…as always, these are entirely personal, do not represent the views of my employer, yada yada yada…
- Some in the community seem to fear that, somehow, some way, Microsoft will crush open source through some combination of legal enforcement of IP, buying off companies like Novell, and scorched earth competitive business practices. This fear is well-founded. But I just think OSS in general, and GNU/Linux in particular, is unstoppable and inevitable at this point. The survival of open source doesn’t depend on altruism or Microsoft behaving nicely, but good ol’ fashioned economics. It’s here to say…have faith.
- The vocal leaders within the OS community are not okay with a mixed-source business model such as Novell is pursuing. For them, it is axiomatic that any kind of proprietary IP is bad. But for the rest of us, it’s not so black-and-white. Ultimately, open source is a business model, and one that has a lot of advantages, but that still leaves room for proprietary software where it makes sense. Right or wrong, we still live in a world with IP protection laws.
- It is way too early to have a good read on how best to monetize open source. Red Hat has been profitable, but I have to wonder how much of that is due to its rapid growth. Will its business model still work in a maturing market? Is a mixed proprietary/open source better, or will its inherent contradictions doom it? Which of the open source start-ups out there will ultimately succeed? Let’s not attack experimentation regarding business models this early in the open source industry’s lifecycle. We need to see what works. After all, the Internet is still surprising us…just look at YouTube.
- The community always seems to assume that the developers and the users of open source are one and the same, but they aren’t, or not only. There are a bunch of corporate customers out there whose needs are not identical with the community’s. But without serving those customers, open source will never go mainstream. We need to think about their needs, not just the community’s.
- I understand that the open source community is always on guard for anyone appearing to take advantage of what is, after all, a public good held in common — the source code — for their own personal gain. At the risk of sounding defensive, though, Novell has been contributing massively to the open source community, so we are hardly free-riding or taking advantage of anyone.
- Lastly, I’m getting really tired of people throwing around the accusation that so-and-so “doesn’t get open source”. C’mon, it’s not some esoteric body of knowledge requiring years of apprenticeship to arrive at some mystical understanding. We’ve been living it for years now. “Not getting open source” is a rhetorical device to shut down discussion. (Now I know I’m sounding defensive.)
So that’s enough for now…let the flaming begin, but at least give a moment’s thought to these issues before cutting me to shreds.
Oracle has made their Linux move, offering to support Red Hat’s RHEL at half the price of Red Hat. For Novell’s official reaction, see here. For my unofficial musings, read on.
Larry Ellison understands an aspect of the open source software industry that perhaps Red Hat missed. In July Ellison said:
[T]he interesting thing about open source is that the intellectual property is available to all of us. So what that means is that any company can take the Red Hat Linux and use it at no cost, so long as they’re willing to support themselves.
Well, that actually includes us. We could take the Red Hat Linux, as long as we’re willing to support it. In fact, we can redistribute it to others and provide support. So why would we buy Red Hat Linux, when we can just take it for nothing? [...] It’s like Fort Knox, except everyone has the keys and takes whatever gold they want whenever they want it.
No one can say Ellison didn’t warn us what he was about to do.
To wrap a little analysis around Oracle’s strategy here, think about a typical value chain for software, whether open or closed. We could define it as:
Design/Development => Test/Packaging => Distribution => Marketing/Sales => Services/Support
In the proprietary world, the company that does the design and development controls the source, making it very difficult to support the product without the developer’s cooperation. This means that proprietary software companies either provide the support themselves, or control the support providers through a partner program. Since it’s tough to generate revenue without the consent of the copyright-holder, the software developer controls the entire value chain.
But in the open source world, this isn’t the case. Anyone can package, test and distribute Linux without having done any development. (Likewise, Linux developers don’t have to work for a distribution.) Anyone can provide support for a distribution without bothering to do any Linux development, packaging or distribution. The value chain can be entirely networked, split up among different players throughout. Vendors that have assumed they’ve got a lock on support like they did in the proprietary world have just received a wake-up call: they can be disintermediated just like Red Hat. We always say that open source ends vendor lock-in – now for Red Hat, this rooster has come home to roost.
To prevent this disintermediation from happening to them, open source vendors need to treat support as an independent business that must live or die based on the quality of their support, not on the basis of exclusive access to source code and developers. There are some economies of scale in the support business, and Oracle may be betting on their size to carry the day against Red Hat. But of course Red Hat can fight back by providing world-class support equal to or better than Oracle’s.
What gets really interesting is how this will shake out in the future. If Oracle succeeds at earning revenues from support without the cost of developing Linux and packaging a distribution, then why wouldn’t others follow suit, supporting not just Red Hat, but Suse, Ubuntu and the rest? Then wouldn’t companies like Red Hat and Novell stop developing Linux and open source, since it’s not necessary to spend money there to earn revenue from support?
I don’t think that will happen for two reasons. First, a good support organization requires very skilled and knowledgeable engineers, and it’s hard to develop that skill base without participating in the development of the software to start with. Secondly, open source isn’t just about the Linux kernel – it’s about Linux as a platform for entire software stacks, some of which will be open source, some proprietary, and some sold as software-as-a-service. Improvements will be made to Linux to better support software further up the stack, which is where the software industry will really be making their money.