Open Source and Open Standards: Do we really need both in Australian government?
In a recently published interview in the IT trade press, Victoria's new CTO, Tony Aitkenhead, made the claim that what governments needed to pursue in order to improve software value was open standards rather than open source software. Is this view correct? What benefits do each of these bring to government computing? How do they differ? Could governments in Australia benefit from open standards more than they would from open source? Should open standards be mandated for the procurement of computer software? Let's see if we can answer some of these questions.
The Open Road
First, what is an open standard? Many long, complicated definitions exist, but let's stick to something more comprehensible. To help us delineate some attributes of open standards, we ask the following questions: can you procure a solution, in functionally identical and interoperable form, from numerous, competitive suppliers? Can your staff migrate their knowledge sets from one supplier's products to another? Can you minimize (to an acceptable level) the retooling necessary (of infrastructure, data and applications) in any such migration? If we answer yes to all these questions, then we most likely have an open standard in place, delivering viable open systems solutions.
Let's now make this more tangible by way of example: cars are an open system. You can train your staff to use (i.e. drive) a car sold by one vendor and they will most likely be able to use a car from any other vendor. Additionally, any infrastructure you have built to work with one vendor's cars (roads, traffic signals, mechanics, petrol stations, L-plate and P-plate training and certification) will work with other vendors' cars.
Open standards in the automotive space allow for competition between suppliers, which is good for consumers. This multi-sourcing of automobiles means that you can go to numerous vendors and acquire essentially the same functional product; vendors have to compete on quality, price and service. This is because the costs to you of switching from one vendor to another are fairly low, so vendors have no choice but to provide real quality value-add, merely to compete. And by multi-sourcing, we do not mean different car yards supplying the same brand of car; we mean cars sourced from different, highly competitive vehicle manufacturers.
Another example is commodity computer hardware. In fact the name itself is a giveaway. Most computers are now commodities and they have become so through enormous competitive pressures, brought about by essentially interchangeable parts, sourced from disparate suppliers, all vying for your business. Even in the CPU sector, once considered to be the stranglehold of Intel, we are seeing intense competition from AMD and VIA on one hand, and on the other, super-efficient, scalable and increasingly mass-market PowerPC-based CPUs from IBM and Motorola. This strong competition in hardware prices has resulted in an almost 1000-fold increase in price-performance characteristics in this market sector over the past decade.
Contrastingly, there has hardly been any price-performance improvement in mass-market proprietary software during that same period. The reason for this is simple: the market for platforms and productivity applications is generally not based on open standards, which would facilitate fiercely competitive suppliers of interoperable and interchangeable software, resulting in price-performance growth characteristics similar to the hardware industry.
As it stands, the commercial off-the-shelf software market is controlled by a dwindling number of powerful vendors, effectively wielding non-standards-based proprietary lock-in strategies, which maximize their profits for minimal ongoing effort. The end result has been negligible or negative price and feature improvements over the past decade.
Back to the Source
Open source is certainly not the same as open standards. Open source software has attributes that can be distilled into two very distinct areas. The first is a software improvement process that is premised on the availability of source code to whosoever wants it, along with the freedom to extend that code in whatever manner one chooses. The second is total and absolutely irrevocable freedom associated with the use of this software. By this we mean there are no restrictions placed on you as a user at all; you can install and run as many copies, on as many systems, with as many CPUs, for as many users, as you care to.
While access to the source appears to be merely a boon for developers, the associated redistribution freedom is actually of greater importance for consumers, by virtue of the economic benefits it accrues. It is often difficult to see this at first, because until now, users of software have never actually had any software freedoms at all, merely restrictions. However, freedom in software redistribution actually opens up competitiveness in the software marketplace like nothing else has ever done in the IT industry's history. As an example, there are several hundred vendors of Linux, a healthy, competitive ecosystem in fact, all vying for some or all of the burgeoning Linux platform market, with interoperable and interchangeable products; think Free as in Enterprise.
And freedom is the attribute of open source that is never highlighted by opponents of the software. Yet it is the attribute that allows business and government agencies to regain control over their IT infrastructure and applications. It also frees up the market for competing vendors to vie for after-sales government business, as no one vendor has a product advantage over their competitors, because none of them "own" the open source software in question.
One Linux vendor can sell a solution to a government agency, and another vendor will have just as much expertise (and access to the source code) to provide ongoing support of that platform, because the great majority of the software components are interchangeable between different vendors' products. The upshot is not only strong competition for initial sales, but also for ongoing support, maintenance and security fixes, thereby reducing costs to government agencies.
This situation flows naturally as one of the outcomes of the GNU General Public Licence (GPL), a licence that has now been upheld in court and shown to be viable. A licence that is growing in stature in the global software marketplace, particularly amongst government and public sector agencies, due to the benefits it brings to them.
Open source software establishes a neutral playing field onto which all vendors can step and compete, openly and fairly. They compete on value, quality, security processes and after sales support. Many of these procurement decision factors are thrown out the window when a dominant monopoly vendor has a stranglehold on the government market. That vendor does not really have to deliver on value, on quality, on security processes or on being attentive to government client needs, because the market does not force them to. It merely has to appear to do so, which it can do by spending sufficiently on marketing.
Such vendors are not evil, merely playing the capitalist game as it is supposed to be played: to their benefit. Strong -competition is the only process that delivers the benefits of the capitalist marketplace for government consumers, and should therefore be fostered and encouraged at every opportunity by government procurement officers, as it is in the government's best long-term interest to do so.
When software is released as open source one cannot assume that it will be based on open standards. And open standards-based software is not automatically available only in an open source implementation. However, due to various economic and philosophical factors, much open source software is indeed based on open standards and for the same reasons much proprietary closed source software is not.
It is simple to see why. Unless they have another agenda in play, most vendors know they can make more money from their clients if they can corral them onto software that is not based on open standards, because the subsequent high costs of switching to alternatives enables far greater scope in price hikes than an open market allows.
Sometimes the market is already established and based on open standards, so this corralling tactic is not available. Other times, a competing vendor has already obtained a strong lead in the marketplace with a non-standards-based product. So its opponent will release a product and then try to establish it as the new open standard, in the hope that this might gain industry support and therefore enough momentum so that it can overtake its rival. In such circumstances, a vendor can also acquire, for the day at least, the aura of being a supporter of such standards, to be heralded at later dates when strategically pertinent: "See, we do support open standards! (When it suits us ...)"
So, in a sense, probably all major vendors can claim that they support open standards in some part of their software product spectrum. In reality, no major vendor from the traditional proprietary software space ensures that most of their products are standards compliant. Open source vendors and developer consortia are far more likely to achieve this goal because they are not in the game of corralling users towards scenarios where there is a high cost of switching to other vendors. The licences of open source just do not allow this, so the impetus or temptation is removed. Therefore, these vendors might as well just support open standards, and most well-managed open source products do.
One thing that may not be immediately obvious when reviewing the aims and attributes of open source and open standards is that open standards, by themselves, mean very little in the Realpolitik of the software industry. There are plenty of open standards, backed by largish vendors, that go nowhere. One example is the International Standards Organization's Open Systems Interconnect (OSI) network protocol suite, designed to be a superset of the functions provided by the equally open, yet at the time, less vendor-backed Internet TCP/IP suite of protocols. It is interesting to note that the OSI suite became a white elephant, going the way of the dodo almost 10 years ago, while the Internet, even though it did not have government and large vendor backing, soared.
What made TCP/IP fly and OSI sink? Simple: the former had a complete, ready implementation of all the important software components, released as open source. The coupling of open standards with a viable open source reference implementation was the critical factor of success and future hegemony. Open source is what can make the difference to the success of open standards.
So, knowing all this, if we were to make a recommendation at a procurement policy level within government - something which would increase the leverage that government has over vendors, greatly reduce the termination costs of migrating away from specific vendors and technologies, and increase the likelihood that data and documents will be readable in decades to come - we would indeed suggest a policy of mandating open standards and multi-sourcing. In this new regime, open source software could then compete openly and fairly on a level playing field.
Of course the reality is that for many government agencies, utterly locked in to vendor-specific proprietary communication and messaging protocols, data and document formats, application APIs and particular applications' look and feel, it will be a major effort to introduce an open standards procurement policy.
Thankfully, it is an effort that is worth making.
Con Zymaris is CEO of IT services company Cybersource. He welcomes feedback at firstname.lastname@example.org