Starting about the time that Bill Gates wrote his infamous Letter to Hobbyists, the commercial software industry has sought to control and restrict access to source code. Before that time, code wasn’t explicitly free, but it was often freely exchanged. The rise of the commercial software industry put an end to that.
When the modern open source software movement was formalized by Bruce Perens and Eric S. Raymond, the commercial software establishment pretty much ignored it. There was no way that useful software could be created for free. Then along came the Linux kernel, the GNU operating system and applications like the Apache web server, and suddenly open source software was not only useful, its adoption started growing phenomenally.
Since it is hard to say software isn’t useful when millions use it, the commercial software industry changed its tactics. A campaign of Fear, Uncertainty and Doubt was started. Can you trust software made by a bunch of anonymous hippies? Who will support it? Who can you hold accountable?
In response came companies like Red Hat, who said “hey, I’ll support it, and I’ll give you better, more responsive service than you get from the commercial software guys.” Slowly, the FUD argument started to fade.
Now I’ve seen the next front on the war for open source. Commercial software companies are attacking the term itself. They are trying to say that commercial software and open source are actually the same thing, even though there is a huge difference between companies that garner most of their income from the support of software and those that earn most of their revenue from the sale of proprietary software licenses.
Words are important. One of my favorite philosophers, the late George Carlin, based much of his work on the examination of how words are used to control people. Take the invasion of Iraq by the United States. Following on the heels of the 9/11 terrorist attacks in New York, the US government sought to justify it by associating Iraq, even though not a single terrorist charged in the attacks was from Iraq and studies showed no link between Saddam Hussein and those terrorists. Yet in 2007 a Newsweek poll showed that 41% of Americans thought that Iraq was responsible, which was actually an increase of 5% from September 2004.
Now it is not the purpose of this post to start a debate about the war, but I wanted to demonstrate that if you say something enough times, even if it is false, people start to believe it. The commercial software companies know this.
For example, let me pick on Matt Asay (I could probably pick on Dave Rosenberg but I don’t read his blog). On December 22nd he ran a blog post with the paragraph:
Five years from now, I’m not even sure what it will mean to talk about “open source” and “commercial software” as if they are two separate and distinct things.
See, Matt works for an open core company that makes their money from selling commercial software licenses on top of a core piece of software that is published under an open source license [Note: see comments below – after researching it, it seems that Alfresco is not “open core” but neither is Alfresco Enterprise “open source”]. To drive value to his company he has to make the argument that while open source is good, it can’t produce value unless someone pays for it, thus there must be a commercial software component. I disagree.
He follows this up on Christmas Day with a post about an InfoWorld article on the future of open source:
Dave Rosenberg writes that 2009 will be the year when open source becomes paid software, but I think we’re already there. We’ve been there for at least two years, in fact. We just didn’t know it.
Once again the association that open source and commercial (paid) software are one and the same.
Now I have no doubt that commercial software companies will have to become more open. They’ll have to provide better and more free APIs and they will have to work hard to build communities around their products, but that doesn’t make them open source.
Finally, the next day Asay follows up with a very paternalistic post on the struggles that the data portability field is having on defining what is “open”. I say “paternalistic” because he comes across as if the whole topic is boring and beneath consideration.
See, we in open source have been through this (attribution/badgeware debate, anyone?), and we resolved it by throwing up our hands in despair and moving on.
Oddly enough, that was probably the right thing to do, as the only people that really care about such things are the vendors involved. Customers don’t care
I claim that customers don’t care because they don’t understand. It’s posts like Matt’s that really blur the lines between open source and commercial software. They didn’t care about Linux when no one used Linux, but suddenly less than a decade later Linux is doing well. Now as open source moves up the stack it’s the same situation. Once true free and open source software becomes a viable alternative it will cause customers to care.
But it’s comments like this that make the process take longer. I’ve helped build a business around OpenNMS, which remains 100% open source software, and as I try to explain the value to potential customers I can no longer rely on “it’s open source” to mean what it used to mean. We still get replies like “yes, it’s open source, so how much is the enterprise version?” It’s “free food” all over again.
Now some of my detractors will say that I just make up terms to suit me, and that my understanding of “open source” is not valid. I get mine from The Open Source Definition by the Open Software Initiative. If anyone says that it is not valid, I’d love to hear the reasons why. What I love about it is that it starts off with “Open source doesn’t just mean access to the source code” (emphasis mine). The commercial/open core/hybrid/shareware folks would love for people to believe that’s all it means.
I can’t say that I blame them. I’ve seen the power of open source in action and if I ran a commercial software company it would be in the best interests of my shareholders to leverage anything I could, including even the most tenuous association with it. But likewise it is in my best interests to point out how wrong they are.
I’m not going to have any effect on those companies, and I realize this. Heck, Matt has his bully pulpit on cnet and my three readers get to visit my rants on an old Dell server with donated bandwidth. But who I really want to reach are those that might consider buying these companies. As Matt says the clients don’t care about open source so the investors shouldn’t either. They need to judge the value of a commercial software company against other commercial software companies.
And they need to keep in mind that projects like OpenNMS are growing stronger every day. While our open core competition might have prettier interfaces and more features, we’re catching up. We’re also focusing directly on the needs of our community, and not the buzz-word du jour. How much value does a piece of commercial software have when we might be able to replace it in six months? Customers might not care about truly open source software in large numbers now, but I’m willing to bet they will. I’m wondering who’s betting they won’t?
Hi Tarus,
An example for the confusion is your reference to Matt Asay’s company as “open core”. I don’t think Alfresco is open core as you define it. They have an enterprise version, but it is not something on top of a core. It has the same feature set and code is 100% GPL. http://www.alfresco.com/products/networks/compare/
This is just an example that it’ll be increasingly difficult (already is) to categorize companies/products as open source or open core or any other category as commercial companies try various different models. It’s a continuum rather than binary (stating what it is not what it should be)
I agree that words are important and understand your desire to keep open source as a clean definition. I get the distinct feeling that you’re rowing against the current.
Heh, I believe that “100% GPL” is a bit misleading.
If you look at the Alfresco Enterprise License:
http://www.alfresco.com/products/ecm/enttrial/register/terms.txt
you can see that it is *very* non-open, with lines such as:
2.1.2 CLIENT agrees not to cause or permit the reverse engineering,
disassembly or decompilation of the Program(s).
but I can see from the website how you could be confused. This is a very novel use of the term “open source”.
As for “rowing against the current” I’ve been doing things the hard way most of my life if I believe they are the right things to do.
The enterprise license is not an open source license, but they also have the “Labs” version that is just GPL, plain open source, no?
The question is what is the difference between Labs and Enterprise versions? The “open core” implies the Enterprise version has additional features that are not available in the open source version. I could find no indication this is the case. If I’m not missing something from a feature perspective they seem to be same. If this is the case, this does not fit into “open core” model as I understand it.
Their differentiation for Enterprise seems to be the support, release cycles, etc. rather than features. If anything this seems to be closer to the dual license approach of MySql rather than open core, no?
My point is that business models are diverging, there are no clear cut lines. You want Hyperic, Zenoss etc. to use the term open core and not open source because they keep some features out of the open source version. Are you saying Alfresco should also not use open source term because they offer an enterprise license, even though same feature set is available via GPL as well (unlike open core approach)?
It is my understanding from Matt’s many blog postings that there is a difference in code between the paid and free versions of Alfresco. I could be wrong (it happens) and if so then you are right, they would not be “open core”.
Note: after some research, I’d have to say that Alfresco is not “open core” but neither is Alfresco Enterprise “open source”. They are something rather new, which is interesting. Check out this thread:
http://forums.alfresco.com/en/viewtopic.php?f=2&t=13576&start=15
and this blog post:
http://nheylen.wordpress.com/2008/08/04/alfresco-open-source-or-not/
I have absolutely zero problem with an open source project releasing the same code under two different licenses.
It just seems a little unclear in the case of Alfresco since the enterprise license is so closed. For example, one can buy Red Hat Enterprise Linux, take out all of the trademarked art, recompile it and offer it as CentOS. The Alfresco license prohibits that. Since this violates many of the points in the Open Source Definition (1 through 3 immediately jump to mind) it doesn’t seem like it can be called open source. Perhaps the Labs version can but the Enterprise version can not.
I’m promoting the use of the term “open core” software for those commercial software companies that have an open source component, but that make their revenues mainly from the sales of closed software licenses. I believe that most enterprise commercial software will fail into this category in the next few years. It is less emotionally loaded than crippleware, shareware or hybrid.
Open core companies owe it to their shareholders to maximize profit, which means they should be making decisions to force more people to buy their closed version (i.e. with Hyperic you get “down” notices with the open version, but “up” notices are only available with the paid version). That’s what commercial software companies do.
But those decisions aren’t necessarily what is best for the community, and thus the community has little option outside of forking it, and who really wants to do that? By comparison, companies that publish 100% of the code under an open license can reap the full benefits of an open source community and truly deserve to be able to use the term “open source”.
You do not seem to get the difference between open source and free software and among different types of software revenue streams. You have your software industry history timeline mixed up a bit as well.
You say:
“Starting about the time that Bill Gates wrote his infamous Letter to Hobbyists, the commercial software industry has sought to control and restrict access to source code. Before that time, code wasn’t explicitly free, but it was often freely exchanged. The rise of the commercial software industry put an end to that.”
Actually the “rise of commercial software” predates Gates’ memo, if not his birth. The industry actually does not care much about restricting access to source code; most of it just wants to be paid for intellectual property it develops. The fact that Stallman does not want to be paid is his choice just as Gates’ position that he wants to be paid is his choice. Even Stallman grants that. Why is Gates’ letter infamous?
You then write:
“When the modern open source software movement was formalized by Bruce Perens and Eric S. Raymond, the commercial software establishment pretty much ignored it. There was no way that useful software could be created for free. Then along came the Linux kernel, the GNU operating system and applications like the Apache web server, and suddenly open source software was not only useful, its adoption started growing phenomenally.”
But that formalization of OSI by Perens and Raymond was in 1998. Prior to 1998, the “commercial software establishment” in the form of IBM (IBM), HP (HPQ) and other major software and IT systems suppliers had already embraced Linux and funded Apache web server development.
Then you say “In response came companies like Red Hat (RHAT)…” but in fact Red Hat had been founded five years earlier, and the software publisher that acquired Red Hat in 1994 (but retained the brand name) goes back even further.
Finally you say:
“… there is a huge difference between companies that garner most of their income from the support of software and those that earn most of their revenue from the sale of proprietary software licenses.”
It’s completely unclear what this huge difference is since there is no such dichotomy in the software market. Most software companies, such as Oracle (ORCL) and SAP (SAP), recognize about 33% of their revenue from licenses and 67% from maintenance and professional services. IBM receives relatively little of its revenue (as a percentage of its total revenue) from “the sale of proprietary software licenses;” even the revenue of IBM’s software group is highly oriented to mainframe software (and therefore comes from selling the system, even though the software is unbundled.) Microsoft receives a high percentage of its revenue from product OEMs such as HP, Lenovo and Dell (DELL) and increasingly from advertising, consumer services and other forms of software monetization.
What is even more unclear is what these various revenue-flow mixes of various software companies have to do with open source. Apparently you feels that your company is different than IBM, Microsoft, Oracle and SAP because it only services open source software and they all service any kind of software.
So you give away the razors to sell the razor blades like King Gillette. You’re still a commerical software company.
— Dennis Byron
Byron:
I do actually understand the difference between free software and open source, which is why I chose the OSI definition in my post and not Stallman’s four freedoms.
As far as my timeline goes, I didn’t mean to imply that my post was a complete history of software, but it is a history of software from my experience. I got my first computer, a TRS-80, when I was 12 in 1978 (2 years after Gates’ letter). Back then most of the software I came across was through magazines, where the code was published freely. While not technically “open” or “free” I was able to easily modify, say, the basic format of a text adventure game which I would then share with my friends as a new game. But shortly after that (around the advent of the first IBM PC) software became much more commercial *in my world*. Of course I realize that the whole of the software industry predates this by decades, but it wasn’t until Gates came on the scene that it became more of an industry unto itself and not tied to hardware. If you disagree with me then I’d love to hear why IBM didn’t insist on owning the software it licensed from Microsoft. I can’t believe they fully understood the ramifications of that decision that cost them billions (grin).
Next item: I state that Perens and esr “formalized” the open software movement. I didn’t say they started it, but they did give it a definition. Of course there was open and free software before that, but I’m sticking by the OSI definition as *the* definition of open source. I would love to know of a better one. The purpose of this post was that there are those who would like to get rid of that definition or blur it to the point where it becomes meaningless.
Finally, as to my quote “… there is a huge difference between companies that garner most of their income from the support of software and those that earn most of their revenue from the sale of proprietary software licenses.” I was describing companies that call themselves “open source”. I don’t believe that Oracle, IBM, SAP, etc. refer to themselves as open source companies, although I expect most commercial software companies to have an open source component within the next few years.
The point of this post was that the open source development model has been shown to be very powerful. It migrates the power from the owner of the copyright to the end user. Those benefits, in flexibility, usefulness and cost have caught the attention of the decision makers in many industries.
So now parts of the commercial software market respond by labeling themselves as “open source” although they still control the code in much the same way as they always have. They release a small “core” piece under an open source license, but their business model is to use that core in such a way as to drive people to buy the commercial “enterprise” version.
OpenNMS doesn’t do this. Fully 100% of the code is available under an OSI approved license. The community (mainly the Order of the Green Polo) has the final say in all development decisions. Any feature good enough for OpenNMS is an “enterprise” feature.
So where are the razor blades? If I buy a Gillette razor I can’t use blade from Schick. There is no “per node” pricing of OpenNMS software. There is no required maintenance contract to get bug fixes. Any commercial services client of The OpenNMS Group is free to walk away at any time without losing access to the latest, greatest, most stable release of the software.
So I’m not sure what you point was, but I do appreciate the input and I hope I’ve cleared up a few of your questions of fact.