Thursday, September 23, 2010

Red Hat's financials are an argument for software patents

Having founded and run the NoSoftwarePatents campaign, one of the hardest things for me to do is to make concessions to the proponents of the patentability of software. But during my campaign I was also known among activists for thinking outside of the box and for trying hard to separate reality from ideology, which is never easy.

In that spirit, I can't help but admit that Red Hat's financials can serve as a fairly strong argument for software patents.

Before I explain why, I want to stress that I'm aware of the sensitivities surrounding this question. Instinctively, most FOSS advocates want to see Red Hat do well in its daily business, and on the stock market (NASDAQ:RHT). But in my opinion, Red Hat sets a bad example.

It's disconcerting that the world's largest open-source-only company pursues a questionable business model that doesn't really stimulate more investment in open source, and that may indeed lead political decision-makers to the conclusion that software patents are needed to protect innovation against unsustainable business models. By "unsustainable" I mean that certain models may work well for one company, or for a few, but could never work for the knowledge economy at large.

Innovation and employment are political goals -- revenues and profits are merely incentives

Thinking back to the debate we had in Europe over software patents and innovation, I remember that the fewest politicians we talked to cared about whether software patents were "just" or inhowfar software was different from other fields of technology. What the decision-makers really cared about were the purposes intellectual property rights serve.

In connection with patents, the two key political objectives that tower above all other considerations are innovation and employment. Politicians realize that innovation has value all by itself, and they know it's a question of competitiveness. But they also want to create and sustain employment, especially in terms of high-quality jobs. That's even more important to them than collecting taxes on a company's sales and profits, although that is, of course, also desirable from a political point of view. As far as innovation and employment are concerned, profitability is just a means to an end: it can serve as an incentive to invest in innovation.

One can argue that the reason why most of the industrialized world has software patents is unrelated to those overarching considerations: patent offices simply grant them, and the courts uphold them. That's true but ignores the fact that if lawmakers wanted, they could do away with those patents. They could pass legislation to that effect. In practical terms, that would mean doing away with large parts of the entire patent system, but legislators could certainly do so if they deemed it good policy. However, they just don't.

In the EU we were able to get a legislative proposal rejected that would have resulted in even more widespread enforcement, but we didn't have any chance to get a truly restrictive law passed. We scored a victory that was only a defensive one. In the US, the Supreme Court recently made it clear in its Bilski decision that judges "should not read into the patent laws limitations and conditions which the legislature has not expressed." That's a clear message: if you want to get rid of software patents, or any other category of patents, talk to Congress, not to the judges.

No one has so far made a convincing case that lawmakers should abolish large parts of the patent system. That experiment just won't happen, at least not anytime soon. The benefits of such a bold move are purely speculative. We can be totally convinced of our views, but we can't prove that this is a better recipe for the economy at large. The risk of abolition doing enormous harm to the economy is far too huge for any political body to take its chances.

Red Hat's business model does more harm than good

A few months ago, open source journalist Glyn Moody asked Red Hat CEO Jim Whitehurst whether his company could reach a turnover of $5 billion, and why it was taking so long. Here's how Glyn reported on the answer he received:

[Whitehurst] said that he did think that Red Hat could get to $5 billion in due course, but that this entailed “replacing $50 billion of revenue” currently enjoyed by other computer companies. What he meant was that to attain that $5 billion of revenue Red Hat would have to displace software that currently costs $50 billion.

This thinking along the lines of a 1-to-10 destruction ratio is familiar to me. MySQL AB, an open source company with which I became involved early on, described its impact the same way.

I actually have doubts whether Red Hat will ever reach that revenue level, and even if it hypothetically did, its destruction ratio won't necessarily stay the same. But rather than getting distracted by investment considerations, let's just take Whitehurst's statement at face value for the sake of the patent argument and do a simple extrapolation.

Running the numbers

The three largest proprietary software companies (leaving the more diversified IBM Corporation out for the sake of simplicity) are Microsoft, Oracle and SAP. Let's look at their contribution to innovation and then compare it to Red Hat and its destructive vision.

Last quarter, Microsoft had R&D expenditures of $2.3 billion (18%) on revenues of $12.9 billion. In its last quarter before acquiring Sun, Oracle's R&D budget amounted to $0.7 billion (12%) on revenues of $5.9 billion; and in its most recent quarter, SAP spent $0.4 billion on R&D, about 14% of its revenues of $2.9 billion. So the weighted average of those three players' R&D expenditures relative to revenues is in the 15% to 16% range.

Assuming that Red Hat indeed achieved the destruction of $50 billion of proprietary software revenues, this would kill an annual R&D budget of roughly $7.5 to $8.0 billion. In exchange, Red Hat would spend about $0.9 to $1.0 billion. That's a very open-source-friendly estimate because Red Hat's R&D percentage would likely go down if the company grew so much, and because there are other Linux players such as Canonical whose contribution to innovation is so minimal it has already drawn criticism from within the FOSS community.

At any rate, the net balance would be a loss of $6.5 to $7.0 billion in R&D, and of all of the high-quality jobs this relates to.

Politically, that would simply be undesirable. Of course, one can argue that consumers and companies outside the IT industry would seemingly "save" costs. But it wouldn't really work that way because with so much less innovation, those IT users wouldn't be able to become as productive as they could with continuing investment in R&D at previous levels.

Conclusions for innovation policy

I explained further above that the political perspective on this is focused on what works for the economy at large. Replacing $50 billion of proprietary software revenues with $5 billion of Red Hat revenues would be theoretically fine if Red Hat's business model were scalable and could serve as a role model for many other companies. But it's a uniquely parasitic model that can't be replicated. The only company for which it works on such a scale is Red Hat itself.

Venture capitalists used to be much more enthusiastic about open source years ago. But since Sun acquired MySQL and Red Hat acquired JBoss, there haven't really been any significant "exits" (IPOs or trade sales) of open source companies. Venture investment in open source startups has cooled off and, compared to previous levels, slowed down to a trickle.

So if I were a political decision-maker concerned with innovation policy, Red Hat would clearly not be a company for me to support. Instead, I would view its financials as an indication that proprietary software developers may very well have a point when requesting strong legal protection for the fruits of their R&D efforts.

From a pragmatic if not utilitarian perspective, it doesn't really matter if there are a few "trolls" taking advantage of the system, or if there's a problem with trivial patents, as long as patent protection favors a sustainable approach to innovation while patent abolition would simply be grist to the mills of a company like Red Hat.

The need for reasonable positions

I just wanted to draw attention to the fact that Red Hat's model is the antithesis of economically sustainable innovation. The proponents of software patentability could use it to their advantage when discussing innovation policy with lawmakers.

In that light, I can't share some people's enthusiasm about Red Hat being the flagship open-source-only company. A more IP-based, R&D-centric company would reflect far more favorably on open source as a whole than a company that just monetizes technologies of which it developed only a very small part (and which it didn't originally create). To foster innovation, the greatest rewards must go to innovators, not to mere monetizers.

The way I have always argued for open source when talking to politicians was that it's needed as a key competitive force to exert pressure on the incumbents. When I argued against Oracle's acquisition of MySQL (as part of Sun) last year, I wrote a position paper that stressed the importance of open source to disruptive innovation, and throughout the merger control process I defended MySQL's GPL-based but IP-centric business model. But it's much easier to argue that open source should be allowed to compete and spur innovation than to make radical demands along the lines of Red Hat's destructive vision. The political response to the latter might be more -- and stronger -- software patents...

If you'd like to be updated on patent issues affecting free software and open source, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents.