Lou's Views: Penguins vs the Dismal ScienceDec 04, 2000, 14:38 (6 Talkback[s])
(Other stories by Lou Grinzo)
By Lou Grinzo, LinuxProgramming.com Lou's First Law of Computers: A sufficiently large economic incentive beats a royal flush.
Dennis Powell and Kevin Reichard recently talked in LinuxPlanet articles ( here and here) about the difficulty that some software companies are having (or anticipating) in making a profit in an open source-dominated market. If you hold the situation at arm's length, and consider all the factors involved, this really isn't a surprise, although it seems it's not as clear to some people around here as it was to Dennis and Kevin. For those who've yet to see those articles, Dennis was directly addressing Linux distro companies and the economics of selling free software, while Kevin focused on the recent decision by Adobe to cancel their port of FrameMaker to Linux.
First, a disclaimer and a warning, of sorts: While I don't own stock in any of the companies Dennis, Kevin, and I mention, I do have a degree in economics, a field which will forever be known as "the dismal science", thanks to that irrepressible party animal Thomas Malthaus. Holding that degree means I'm legally entitled to reverse my position on any economic issue, with zero advance notice, by invoking the magic phrase, "on the other hand." (The tendency of economists to do this reputedly caused US President Harry Truman to ask someone to find him a one-handed economist.)
The basic problem for the Linux distro companies, despite the wishes to the contrary of many in our community, is that software does indeed want to be free, as in beer. There's nothing special about software, nor is this a special trait of its being intellectual property instead of a physical good, like a car or a bowling ball. Economists have long observed that in a sufficiently competitive environment the price of any product is driven down by the market to the cost to produce each unit, or in econ-lingo, until marginal revenue equals marginal cost. It doesn't matter if it's software, hardware, or underwear. As for what "sufficiently competitive" means, consider what those shrink wrapped boxes from Red Hat, SuSE, Mandrake, et al., are competing with: Not just each other and the other OS's on the market, but most important of all, they're competing against the free version of themselves. Linux newbies quickly figure out that the free copy isn't sorta kinda almost like the shrink wrapped copy, but it's the same exact product, at least in terms of the bit pattern on the CD(s), even if it lacks other goodies like support.
But these identical copies aren't just free beer, they're free speech, which is why we see mail order companies that can legally sell $2 Linux CD's. It also explains why they can still make money at that absurdly low price--the bits they put on the CD's are free (beer), so their only cost is that of duplication, order processing and fulfillment, running a web site, and normal business administration. Most important, they're not paying a flock of developers at an annual, fully-loaded, labor cost (not salary) of $100,000 to $200,000 each to develop the software they're selling. That keeps their expenses and the prices we pay low. Economics doesn't get much simpler than that.
The story for the distro companies is conspicuously less rosy, to say the least. For starters, they do have to pay that flock of programmers, and they also have to pay the support people, bear the promotion cost to build brand awareness and loyalty, etc. But since the CD's in the package can be freely copied, they have a gaping hole in their revenue stream. I haven't seen the numbers, but I'm willing to bet that the vast majority of their users don't pay them a cent for the software and don't pay for support--they buy a $2 CD or they get a free copy from a friend or the Net, and they subsist on free online help or they buy a book or two.
Aside from the relatively small number of people who buy the shrink wrapped package, the only time the distro companies receive revenue is when someone pays them for support or subscribes to some other service, or they hire the company at consulting rates to do custom development work. (That kind of consulting was supposedly a solid revenue stream for Cygnus before they were consumed by Red Hat, in fact.) This means the companies must fund the entire outfit, including developers, support techs, management, advertising--the whole ball of wax--from the revenue stream generated by things only a part of their customers will pay for. Whether this explains why Red Hat's stock is trading at under $7 a share as of the morning of December 4th, can be debated endlessly, I suppose, but it sure looks to me like Wall Street has figured out that even if the Emperor Penguin isn't completely naked, he's still wearing far less than a full Armani suite.
The inescapable conclusion is that the Linux distribution companies will never grow very much, even if they do turn a profit, since there's just not enough revenue there to fuel a big company. The open source model eviscerates their revenue stream, making the outlook for these companies pretty grim.
On the other hand (hey, you were warned) aren't we in a wholesale movement away from "bits for bucks" to a services and subscriptions model, anyway? I believe we are, and that's why I think the two most fascinating developments in the industry are Eazel and their subscription services, and Microsoft.NET, MS's attempt to go distributed with Windows. The big unknown in all this is how much people will be to pay for which subscription services. I've spoken to clients who see nothing wrong with this shift, provided they get something in return, such as greatly simplified system management. (Trust me on this one, all you would-be software tycoons: Figure out a way to insulate Windows users from DLL Hell and Linux users from Shared Library Misery, and you'll be both wealthy and worshipped beyond your wildest imagination.)
A key main reason there's so much uncertainty in the industry right now is simply because we're in between equilibrium states. The prior state saw the closed source software/bits for bucks model dominate, with free support, later moving almost entirely to paid support. In that state we had software-only companies of all sizes making a profit, from one-person shareware shops to outfits like our cousins up in Redmond. It was and largely still is an economically viable business model, and it fueled a lot of companies and some spectacular investment portfolios.
The next equilibrium point will be more oriented toward open source, and I suspect that in the long run this will all but eliminate the software-only companies. There will be a lot of consolidation as companies merge and acquire each other, and also evolution, as companies try to stay independent and convert themselves into service companies. I won't guess how far this new equilibrium point will be from the old one, in terms of either time or degree of change from the last point; I don't think anyone can tell for sure while we're in the middle of the transition.
The really interesting thing is that the further we go along this path, the more software will be produced by either the stereotypical open source project, staffed by people scratching an itch and not getting paid for their work, or by large companies, like IBM, HP, Compaq, Intel, and others, that have a financial incentive to spend big money on software they can give away. Don't think for a nanosecond that these companies want to spend horrendous amounts of money developing software they can give us because we're nice or lovable; it's simply in their best interest to do this to promote more pragmatic endeavors, like making money from hardware sales. Seen in this light, the only thing surprising about IBM's stunning, wall-to-wall Linux commitment is that it's not even bigger. IBM, thanks to the variety of hardware platforms it sells and supports, probably has more to gain from Linux's long-term success and arrival as a unifying platform than any other single company.
The good news is that there are no surprises here. Drive the market price of a good to almost nothing, and companies either leave the segment entirely or find another product or service to sell, since there's no way for them to make a profit. It's not about right or wrong any more than is the speed of light or the way you react to a loved one's voice; it's just economics, pure, simple, and sometimes a little dismal.
Lou Grinzo has been a professional programmer and technical writer since 1980. He's written for numerous computer publications, including Dr. Dobb's Journal, Software Development, EXE, Linux Magazine, PC World, and Computer Shopper, and Windows Magazine (hey, a geek's gotta eat). His dream is to win the lottery, retire, and spend his days creating the perfect programming language, which he'll then use to write the perfect operating system. His non-computer hobbies include juggling and woodworking. He lives in the wilds of upstate NY with his wife, Liz.