Friday, 2 September 2011

Commoditised software

SCO v Novell

On Tuesday the 10th Circuit of the United States (federal) Court of Appeals gave judgment in the long running case of the SCO Group, Inc v Novell, Inc.

Some readers with an IT background may remember Novell, who were the first company to offer a decently priced and relatively usable networking system for small and medium sized businesses, with their NetWare product. Since the heady days of dominating this market sector, times have been harder for them (others have caught up), but they have been staking their future of late on commoditised software, in particular that offered by open source endeavours such as linux. IBM have also been investing heavily in this new sector.

This is not a small market to tap into. Practically all the automated trading systems of the large financial institutions are now running on linux, and in combination with the Apache web server it occupies a sizable chunk of the web server market, where it offers real competition to windows server products. Mega-sized web operations requiring high scalability and reliability now use linux as a first choice: if you use Google, Facebook or Wikipedia, or indeed read this blog, the services have been running on linux.


Linux is an extraordinary venture. It began in 1991 as the hobby of a certain Linus Torvalds who was trying to write an open re-implementation of the POSIX standard. POSIX is the standard, published by the International Standards Organisation, for unix-like operating system interfaces (I come back to unix again below). The defining feature of this venture was that Torvalds decided to release his computer code ("source code" in computer-speak) for this reimplementation using the General Public License (colloquially known as the GPL). The GPL basically provides that anyone can use and modify the code covered by it as much as they like, and freely transfer it to anyone they like, provided that if they make modifications to it they make the source code to the modifications freely available also. It is a copyright license from the author of the source code which in effect makes a bargain saying "you can use my code without charge, provided that you make freely available also any improvements to it which you develop: you don't have to use my code, but if you do, that is the basis on which you can do so".

This turned out to be a winning formula. Other individuals interested in writing a POSIX re-implementation joined in. Because it was freely available, a number of universities started including it in their courses on computer operating systems. More people started contributing, and it snowballed. When it became a fully usable server environment, a number of companies (including IBM) looking for an alternative to Microsoft Windows, which was at the time gathering to itself something of a monopoly, became involved. And so on it went.


Now back to unix. This was originally a proprietary operating system developed by Bell/AT&T (with a different flavour developed later at the University of Berkeley) in the late 1960s and during the 1970s. Over time, the interface and specification for this operating system became an international standard, POSIX, as later supplemented by the Single Unix Standard (SUS). POSIX is published by ISO and the rather larger SUS standard by the X/Open Group. Although the specification for the unix operating system is standardised and open, most of the implementations of it originally were not. There were and are a number of proprietary implementations of the standard, including those of Sun Microsystems, now Oracle (Solaris), Hewlett Packard (HPUX) and others. Some parts of these implementations were licensed from AT&T and other parts were separately developed by the companies concerned and therefore owned by them. A non-proprietary offering, deriving from the University of Berkeley source code which they subsequently made publicly available with any AT&T code removed, has been obtainable for free for a number of years, now known as BSD; but at that time at least it did not have the breadth of implementation (nor the commercial back-up) of the commercial offerings.

The important legal point here is that rights over copyright are intellectual property rights owned by the author, or a person to whom the author has assigned the rights, preventing copying. In relation to standards such as POSIX and SUS there is no legal preclusion of a re-implementation of the standard by freshly written source code, provided that any fresh re-writing is not done by copying from the original. What are called "clean room" re-implementations, under which the author of the re-implementation is forbidden from seeing the original, are perfectly legal from the copyright point of view.

Novell bought AT&T's unix from AT&T in 1993. They sold some of it on to the Santa Cruz Operation in 1995 (what this "some of it" comprised, later formed the subject of the litigation in SCO v Novell). In 2001 Santa Cruz sold to Caldera Inc whatever it was they had acquired from Novell. In 2003 Caldera Inc changed their name to "SCO" (presumably to make it resemble Santa Cruz Operation), and sued IBM alleging, amongst other things, that IBM had breached SCO's copyright in AT&T unix by copying some of AT&T unix into linux. At the same time, SCO brought test cases against some other companies using linux, claiming that linux infringed their AT&T copyrights by virtue of IBM's contributions and in a number of other ways.

Novell were worried about this. Like IBM, they were starting to get heavily involved in providing linux services to customers who would pay for support and for certain "add-ons" which Novell were providing. They stated that the proceedings against IBM, so far as based on copyright, and the other test cases, were bound to fail because Novell still owned the copyrights which were alleged to be infringed. Novell claimed that the 1995 deal did not in fact transfer to Santa Cruz the copyrights to the AT&T code, and instead only granted to them a right commercially to exploit the code and to add their own improvements. If true, this would kill all the copyright-based claims by SCO stone dead. SCO accordingly sued Novell on this.

SCO's allegations against linux were always going to be difficult to succeed on: SCO never publicly identified the code which they said had been copied, nor did they explain why linux was not an independent fresh creation which, as explained above, would not be constrained by copyright law. In addition, when still calling themselves Caldera, they had themselves marketed a linux product under the GPL so making publicly available the source code which in their subsequent litigation they claimed was still proprietary.

But none of this could get to court if they could not prove ownership of the AT&T copyrights alleged to be infringed, through the 1995 sale to Santa Cruz. They failed to do so. After a tortuous series of events and proceedings, which included SCO filing for bankruptcy protection1, in a federal jury trial in March 2010 the jury decided that Novell still owned the copyrights in question. SCO appealed to the federal Court of Appeals, and in their judgment on Tuesday the Court of Appeals decided that the verdict was to stand.

Commoditised software

The last legal road block to widespread commercial adoption of linux is now to all intents and purposes at an end.

What has happened is that high-grade operating system deployment has now become as commoditised as, say, water or indeed air. Companies such as IBM and Novell charge not for the operating system product itself, but for services they provide on top of the product. Their business model is now very different from that of, say, Microsoft, who derive a large part of their income from sales of the Windows operating system itself and their Microsoft Office products.

What we are likely to see in the future is a market dominated by both linux and Windows, but in differing sectors. Linux will be ubiquitous at the large scale end. Windows will likely hold on to a decent share of the small and medium sized business sector and more particularly the desktop computer, where Microsoft Office reigns and where individual users want the convenience of Microsoft's large range of consumer desktop offerings.

A modest proportion of the smaller business sector will likely also be retained by the other free POSIX implementation mentioned above, namely BSD. (It is also to be noted that BSD forms the basis of Apple's OS X operating system as used on Macs and some other of its devices. The distinguishing feature of BSD is that its licence does not require modifications and improvements to be offered back to the public domain, as in the case of linux, which makes it attractive to specialist hardware vendors such as Apple but unattractive to the service-based business models of IBM and Novell.)

The big battle is now on for mobile devices, namely tablets and mobile phones. Windows has become seriously squeezed in this area by linux-based offerings such as Google's Android, and we will have to see how this all pans out.

In addition, although copyright issues concerning linux are now dealt with, the US patent system allows patents to be issued in respect of software, whereas most of the rest of the world, including the UK, does not: in the EU, under the European Patent Convention "computer programs ... as such" are not patentable except insofar as they solve a "technical problem" (as opposed to a business problem) in a non-obvious way. Patent litigation concerning software will keep lawyers, in the US at least, well employed for some years to come, and the full extent of what is and is not covered by the European Patent Convention is also likely to be explored in future litigation. I may well come back to the last of these in respect of the recently initiated Spotify litigation.


1 The way in which the bankruptcy court allowed SCO to use its creditors' money to proceed with its litigation ambitions is one of the more notable features of this case. Whether this is a particular feature of US bankruptcy law (which is a federal matter) rather than the particular predilection of the bankruptcy judge concerned, Judge Gross, is a matter I leave to whose who know more about US bankruptcy practice.

No comments: