Open Source: On why not
[April 11, 2000: This article seems to have pressed some hot buttons, based on the hate mail I've received regarding it. What's interesting is that every single such letter I've received has accused me of holding a position with which I don't actually agree. Either I didn't express myself well, or a lot of people aren't reading it very carefully. So I'd like to make some things clear up front: I am not arguing that open source is wrong. I am not arguing that there is no place for open source. My position is that there is a place for open source but also that there is a place for commercially-developed proprietary source. My argument is not with people who think that some things should be open source (in fact, I agree with them); rather, my argument is with the people who think that everything must be open source and that there should be no proprietary commercially-developed software at all. I am attempting to argue that it is acceptable for some things to not be released as open source. Furthermore: I think that Linux is a wonderful thing and I am very glad that it exists. However, I think that the companies trying to build businesses out of it are on a fool's errand. While I think nearly all of them will fail in that enterprise, I don't think this will particularly harm Linux overall, nor do I wish Linux to be harmed. I am quite confident that irrespective of the commercial prospects of these companies (which I think is dismal), Linux will continue to exist, to spread and to be enhanced, and I am very glad that this is the case.]
I'm beginning to get a bit tired of people asking everyone and their dog to release the source for every product they can think of.
Somehow people have gotten this idea that this is the wave of the future, and that each and every commercial software vendor needs to get on the bandwagon.
But they're looking at it from the wrong point of view. Open Source has significant drawbacks for commercial software vendors, because at the lowest level it virtually guarantees easy piracy of the product.
I'm not saying that no-one should release anything Open Source. There's definitely a place for open source. What I'm saying is that there is not and should not be any obligation to do so.
I think the problem is that most of the people making these suggestions have a very concrete understanding of their own situation and of the benefits for them in an Open Source release of something that they use or would like to use, but only a very vague understanding of the actual situation and effects this would have on the producer of the software.
The Open Source advocate knows how expensive the product is, and that's real money from his pocket flowing out.
But it seems as if the conceptual model ends at the point where the money is received at the corporation, as if it somehow vanishes into a black hole at that point. Or perhaps they have this image of it flowing into the wallet of some fat-cat corporate executive (or worse, fat-cat stock holders, who "of course" are scum).
But that's not the case. Those corporations have real expenses and that money is used to pay them. Without that income, the corporation can't exist. (And virtually all of these corporations pay no stock dividends.)
People are talking about Red Hat, for instance, as a big Open Source success story. They went public, their stock went through the roof, the founders are now wealthy beyond the dreams of avarice, and they now have a market cap above $13 billion dollars (as of 2/10/2000).
But now that they're publicly traded, we can examine their books in the cold hard light of their SEC filings. Here's the most recent one, as this is written (updated 6/28/2000):
(update 11/19/2000: http://www.sec.gov/Archives/edgar/data/1087423/000095016800002190/0000950168-00-002190.txt for the quarter ending 8/31/2000)
Someday, somehow, someway, this corporation is actually going to have to make a profit. A big one. Continuously. For a long time.
It sure hasn't happened yet.
In the year ending 2/29/2000, they had gross revenue of $42.43 million and losses of $43.91 million. They actually spent more than twice as much as they took in as revenue. (11/19/2000: in the most recent quarter revenue was $18.4 million and expenses were $32 million. They're still spending money nearly twice as fast as they're making it.)
Their fundamental problem is that they're trying to make revenue by selling something which they are obligated to give for free to anyone who asks for it. This costs them money (to operate the FTP server and pay for the network connection to it) without giving them any compensating revenue. In the mean time, while they have been developing additional software which they include in their distributions, they're obligated to give all of that away, too. And at least one competitor is now selling, at cut-rate, what amounts to a clone of Red Hat Linux including all their custom code. And there's nothing, nothing at all, that Red Hat can do about it.
Their solution to their revenue bind is to try to sell services along side, and to make the bulk of their revenue from that. So far it isn't working. It remains to be seen whether they can ever turn around. (11/19/2000: Since then they've tried other things. None of them have worked yet.)
Red Hat's market capitalization is preposterously high: It's more than a thousand times yearly gross revenue, which is absurd for a company which has never turned a profit and shows no sign of doing so any time soon. (written 2/10/2000; as of 6/28/2000 due to the meltdown of RHAT stock price, it is now "merely" a hundred times yearly gross revenue, which is still preposterous.) (11/19/2000: Based on stock price as of 11/17/2000 and 4 times the revenue for the most recent quarter, its market cap is now a much more reasonable 21.5 times yearly gross revenue. But that's because the RHAT stock price completely collapsed, not because their revenue has substantially increased. More on that later.)
By comparison, Intel's market cap is 32 times revenue. Qualcomm's market cap is 24 times revenue. Sun's market cap is about 20 times revenue. IBM's market cap is about 3 times revenue. Apple's market cap is less than 2 times revenue. And they're all profitable, though Apple's been struggling. (All numbers as of 2/10/2000).
For all the enthusiasm about open source, the fact is that no-one has yet figured out how to make a profit out of it. No corporation which has based its business model on open source is in the black. That's not to say that it can't be done; we don't know yet. But it's a bit disheartening that every one of them is bleeding cash like a stuck pig.
And until at least one of them makes a consistent profit, you're not going to see a big buy-in to the concept from the traditional software developers which have been using the traditional software-source-is-proprietary business model.
Yes, I know that IBM has released JFS to Open Source. Yes, I know that id released the source to Quake. Yes, I know that Netscape released the source to Communicator. Yes, I know all about Corel.
But you don't see IBM releasing the source to AIX or ViaVoice, or id releasing the source to Quake III Arena. And Netscape hasn't released their server code. What all of them are giving away (which is what it amounts to) is things they no longer care about. In the case of id, it was more an acknowledgement of reality than anything else, because the source to Quake had been stolen and was out in the world already. In the case of Netscape, they went open source on Communicator about the time they gave up on trying to make any money selling browsers.
You haven't seen any corporation give away the source to one of their crown jewels yet, unless they were legally obligated to do so (as is the case with Red Hat). And you're not going to anytime soon.
As to Corel, it has been bleeding money for years, as Michael Cowpland has frantically floundered around trying one thing after another to try to turn the company around. Embracing open source is his latest miracle cure for Corel's chronic losses. It's an act of desperation. It remains to be seen whether it will work any better than the last five things he tried. (Cowpland's job is on the line; he's facing a stockholder revolt, and has been perilously close to being fired for mismanaging the company.) (11/19/2000: Well, Cowpland is out on his ear and it's about damned time. The company was literally within days of bankrupcy when a white knight rode in and infused a huge amount of cash. Much to the chagrin of Linux users everywhere, the white knight was Microsoft. And now there's reason to believe Corel will abandon Linux and concentrate on supporting Microsoft's ".NET" initiative instead. As a result, Corel may actually survive after all.)
Corporations don't live on good will. They need money to operate. Any time they take a successful product and release the source to it, they have to figure that the income stream from it will immediately plummet as people begin to create their own copies for free (whether licensed or not). It won't necessarily drop to zero, but it will definitely dropfootnote.
Let's take a hypothetical small developer with a couple of hundred employees, which is privately held by one or two individuals. When you buy a product from them, that money doesn't flow into the owner's pocket. It's not a matter of the difference between a new Ferrari in his garage twice each year instead of only a new Porsche only once per year.
That kind of mental model makes it easy to think of him as a greedy money-grubbing bastard who can easily afford to give away things you want but don't want to pay for. But that's not reality.
That money doesn't go to the owner; most of it goes to pay for the operating expenses of the corporation, and the single biggest operating expense is the payroll. (The second biggest is usually taxes.) If you ask the owner to give away a product he's now selling, it means he has to select at least one of his employees and sacrifice that employee on the altar of Open Source by laying him off.
Our corporation owner also works there. He knows the names of every single person working for him. He sees them in meetings. He talks to them in the halls. He visits their booths and sees the pictures of their spouse and children. He gets to meet those families at the corporate picnic. The employees are his family; they're his friends; he cares about them; and he knows that the salaries they earn support their lives and their own families. To reduce his revenue so as to prove that he "Gets It™" about Open Source would require him to seriously harm someone he cares about and feels responsibility for. There's a price for everything, and that's the price you're asking him to pay so that he can prove that he "Gets It™".
How can you be so heartless as to ask him to do this? Does it make it any better just because you don't know the name of the person who's going to get laid off and because you don't have to be the one to tell that employee the bad news and escort him to the door?
So unless you can show how this corporation will get offsetting revenue which compensates him for the revenue lost by giving away a product he now sells, and I mean immediate revenue in large quantities, don't bother asking him to open source anything.
"Good Will" doesn't appear on the books. "PR" doesn't pay the salary of the unfortunate employee who's going to have to be laid off. "Getting It" isn't legal tender to render to the IRS for corporate taxes. Salaries and taxes require lots of money, and they require that money right now. So don't bother with an explanation of how it will increase their revenue some day, if the winds blow right, maybe.
Show a financial case right now why releasing the source to one of his products will increase his revenue overall (or at the very least break even) right now, or don't bother asking.
Think like a capitalist. It's the money, stupid.
If you want him to open the source to something, you better explain to him how he can do that without destroying the life of someone he cares deeply about: one of his employees, who is also his friend, who would have to be laid off if corporate revenues decline.
If the cost of going open source is laying off people, then the price is too high.
The exact same thing applies to a large corporation like IBM. It's larger, but it's still made up of people. The revenue they take in still goes primarily to payroll. "Why doesn't IBM release the source to OS/2?" I've heard again and again. Because if IBM does so, it will cost them a lot of moneyfootnote and reduce their revenues, and that's a shitty way to do business. It's generally not popular with stockholders to increase your investment in some product so as to deliberately reduce your revenue from that product. But more important is that if they do that, someone's got to be laid off.
Every time someone suggests that IBM should do this, I ask what benefit it would have for IBM. And ever time, no-one comes up with a direct source of revenue which would offset the loss in sales and the increase in engineering expenses. Instead they talk about nebulous concepts like PR and customer good will and getting it.
That's horseshit. What that really means is "I want them to release it and I don't really have a good reason why it would be good for them. They should do it because I want them to, irrespective of whether it helps or harms them."
Good will and PR and getting it don't appear on the corporate spreadsheet, and you're not going to get Lou Gerstner's attention until you can demonstrate a source of revenue -- dollars, bucks, moolah, simoleons, cash -- created by doing this.
That's not because Gerstner doesn't "Get It™", but because he has a legal obligation to his stockholders to operate his company in a profitable fashion (not to mention a moral responsibility to his employees to do so, so that he can keep employing them).
"Revenue" is not a dirty word, and neither is "Profit". There's nothing immoral about selling software. Doing so employs people. There is no obligation to put anything into Open Source.
If someone wants to do that voluntarily, more power to them. But don't demand, or even suggest, that someone do so unless you can show them how it benefits them directly, immediately, personally.
Eric Raymond (a major proponent of open source) said it best:
"Either open source is a net win for both producers and consumers on pure self-interest grounds or it is not. If it is, you cannot lose; if it is not you cannot (and should not) win."
Update, June 28, 2000: Since that was written, Red Hat stock has dropped from 87.063 to 27.9375. That's a drop of over two thirds.. RHAT hit its peak on December 9, 1999 with a closing price of $143.125. Today's closing price represents a drop of over 80%.
Meanwhile, VA Linux has dropped from a peak of $242.875 on its sixth day of trading (12/17/1999), to a close today of $42.875; a decline of more than 80%.
Caldera closed its first day of trading (3/21/2000) at $29.438 and closed today at $13, below the actual offer price of $18 and a decline of 55%.
Apparently investors are beginning to realize that there's no economic there there when it comes to Linux.
Meanwhile, Corel has announced yet another losing quarter and though they were planning to release a major product for the Linux market in April 2000, nonetheless they are projecting those losses to continue for at least two more quarters. There's no sign that Corel's embrace of Linux is making any difference to its financial outlook, which continues to be extremely bleak. Their attempt to merge with, and loot, Inprise in search of cash fell through. A white knight gave them $20 million (God knows why) but at the rate they've been burning money that wasn't even enough to keep them alive through one quarter. (It also wasn't enough to pay the costs involved in the huge layoff they announced.)
Corel peaked at $39.250 on 12/09/1999, but closed trading today at $4.0313, a decline of nearly 90%.
The Microsoft antitrust legal decision was announced on Monday, April 3. If anything was going to help Linux stock prices, you’d think this would be it. Here’s how our Linux stars in the following week:
Not one of them rose in the week following the announcement of that decision. Apparently not even that can stop the melt-down of the Linux stocks.
Update, 11/19/2000: The market has finally realized the actual situation of the companies involved in trying to make a commercial success from Linux, and their stock prices have completely collapsed. There's a good reason why, and no better summary of it can be found than this quote from the VA Linux 10-K dated 10/26/2000:
In other words: "We're losing lots of money; we expect to continue losing lots of money; we don't know how we're going to stop losing lots of money but if we don't figure out something fairly soon, we're dead meat, because the VCs are no longer calling us on the phone and our stock price has collapsed so we have no way of raising more money to cover our losses."
The market for "raising additional capital" for speculative stocks has vanished now. The VCs no longer pour money into anyone who mutters the magic words "high tech"; they have to actually show a viable business plan on how they're eventually going to be profitable -- and these guys don't have one. And it's not just them; it's the lot. None of them know how they'll become profitable.
In the mean time, their stock is a complete disaster, and there's no hope of raising substantial amounts of money with new stock offerings.
The market is now pricing their stocks at something like what they should be, and it's down from the heady days of a year ago. Down a lot.
* Caldera first trading day 3/21/2000 closing price was 29.438.
Columns labelled 5/00 and 8/00 contain the highest price achieved by the stock respectively during the months of May and August 2000. 11/17/00 is the closing price on that day. All prices are in US$. Change% is from the first column to the closing price on 11/17/2000.
The tech stocks haven't done all that well in the last year, but even by the standards of the tech stocks, this is really stinking. And even in those periods in which the tech stocks have again risen, these stocks have continued to fall. (Comparing best price in 12/99 to 11/17/2000, Intel is down 56%, Qualcomm is down 50%, Sun is up 9%, IBM is down 14% and Apple is down 68%. Apple, however, is a special case; it lost 52% of its value on one day (9/29/2000) after a very, very bad surprise announcement about the prospects of the company. And even then it's done better than any of the Linux stocks cited above.
There's a good reason why these Linux companies have done so badly on the market: almost all these companies are going to fail because they're never going to earn a profit. And ultimately, all successful companies must be profitable.
[And these failures will have no effect on the future of Liinux or the open source movement -- just wanted to toss that in so I don't get more hate mail.]
Update 5/14/2002: Well, the dot-com bubble burst and everything went into the toilet, and it's now understood that a lot of these companies were made of smoke and mirrors and fast talk and hand waving. Let's update those stock prices one more time, shall we?
Cobalt Networks was acquired by Sun. VA is no longer in the Linux business; it's a shadow of its former self and trying to survive by selling (as non-open-source) the software used in SourceForge. It's still losing money and probably won't be around much longer.
Red Hat is also still losing money. They had one quarter where they claimed a pro forma profit, but their GAAP result (as filed with the SEC) was a huge loss, and they are continuing to burn cash. At their current burn rate they'll be dead within one more year.
There is still no example of a company which has embraced open source as its central business plan which has actually become profitable. I still won't say it's impossible for that to occur, but it's looking increasingly unlikely.
I'm aware that philosophically "open source" doesn't mean "free", at least in theory. But in the real world, as soon as you release the source to something, nothing can prevent a determined individual from creating their own working version of it without paying you anything, and nothing can prevent him from spreading it around to other people who aren't willing to do that work for themselves. Irrespective of the philosophy of the open source movement, as a practical matter there will always be a drop in revenue from sales if the source to some product is released.
IBM releasing the source to OS/2 would not be free for them. IBM doesn't own it all. It contains source which belongs to other corporations, and also source written by IBM which is based on intellectual property (IP) of other corporations. Some of that IP is covered by non-disclosure agreements and most of it is covered by royalty agreements. Before IBM could release even part of the source of OS/2, it would be necessary to make a very thorough pass through it to identify those files which IBM owns 100%, and those which contain either source or IP belonging to other corporations. Probably they'd have to do it twice, with two different teams, just to make sure they didn't make a mistake -- for the lawsuits which would result if they screw up would be most impressive. The process of evaluating several gigabytes of source this way would be long, time consuming and very expensive.
This page has been viewed 2474 times since 20010726.