The Brampton Factor: Software costs under pressure

Hardware prices may have fallen over the years but software costs have always headed in the opposite direction. Now experts are saying that may be about to change. Martin Brampton has his doubts.

The Gartner Group assures us software costs will fall over the next decade. The analyst group says buyers will find ways of cutting costs much in the same way they have done with hardware and services.

Factors such as business process outsourcing (BPO), software as a service (SaaS), service-oriented architectures (SOA), along with the rise of the emerging economies, are cited. Well, Gartner may be right. But I'm not so sure.

One obvious flaw in the parallel Gartner draws with hardware is the humble but essential PC. It has certainly fallen in price, especially when you take account of the vast increase in hardware power.

Yet the one component of the PC that has gone up in price is the software - the operating system. So the mere possibility of hardware cost reductions is certainly not enough to bring down software prices. Market power is at least as important.

Another potential weakness in the Gartner proposition is the strong likelihood the 'Chinese Effect', which has brought down the cost of many manufactured goods, is about to go into reverse. Should we really be relying on emerging economies such as Brazil, China and India to produce cheap software and bring down prices for us, just when their own prices look like taking off?

Not only are wage costs rising sharply in many of those countries, the value of their currencies is also rising in relation to the dollar and will very likely also do so against the pound.

While it is certainly possible that China will become a factor in the creation of software, there are also quite a few constraints. Unlike India, skill in the English language is much less widespread in China. At least as important is the fact China's own rapidly growing industries have a huge need for IT skills. Even with the phenomenal numbers of qualified people coming out of the Chinese education system, a great number of them will be needed by the indigenous economy.

Another factor cited by Gartner is open source. This seems an obvious way to bring down costs but it is worth examining this idea in more detail. The reason open source software is normally free is frequently overlooked. It has little to do with the cost of development, which is normally a minor factor in the total price of a software product.

The primary saving comes from not having to pay for the packaging, marketing, selling and support of software that make up such a huge share of the price. Open source software avoids many of those costs and is, in that sense, a much more efficient distribution model.

Although there is significant volunteer activity in open source, major products usually have a commercial backer. In some cases, open source products are proprietary products that have been turned into open source by the likes of IBM.

The reasoning is simple. In any competitive market, it is difficult to make money selling software. Many products have huge write-offs somewhere in their history. A more profitable route is often to give away the software and to make money providing expert services.

But making assumptions about cost savings from altered distribution models raises the question of whether this is universally applicable. Not every creator of software is convinced they can make money by giving away their work.

While IBM believes that in some circumstances it can make the open source model a commercial success, other companies such as Microsoft look extremely unlikely to attempt it. The danger is that the service revenues the software generates end up accruing to someone else.

There are links between ideas such as SOA and open source. SOA looks very much like the old established notion of software components linked with the even older adage that there are no products, only services. And the similarity?

Much open source software is excellent but fragmentary - just like the best components. There are numerous first-class open source classes in a variety of languages to carry out specialised operations. While they are a valuable resource, it takes time and skill to piece them together into useful applications.

Predictions of a lively marketplace in components have been around for a long time and not a great deal has happened. Complete applications are complex and the meaning attached to data can be subtle. This makes it difficult to design and use effective interfaces.

Although SOA units are intended to be larger than traditional classes, the larger they are, the more complex the interface may have to be. All these ideas are based on sound principles of software architecture and so will have some kind of success. But progress in practice usually lags far behind the optimistic and oft-repeated predictions of revolutions in software production.

Perhaps SaaS can deliver cost savings? Again, progress with application service providers has been slow since this was first touted. And powerful vendors have not allowed any financial benefits from this model to deprive them of revenues.

For example, Microsoft moved quickly to ensure that thin client deployments of a Windows desktop involved payment of much the same software licence fees as traditional desktop implementation. So new modes of operation seem to demonstrate the balance of power between vendor and buyer rather than to change it.

What of BPO? Certainly there will be cases where it results in shifts in the use of software and it will have financial effects. But the history of traditional outsourcing shows the benefits are often much harder to realise than expected - and that things sometimes go badly wrong. Oh, and whatever happened to that old chestnut, gaining competitive advantage through IT?

All told, it seems unlikely these changes will greatly affect how much is spent on IT. Historically, what has been remarkable is people's ability to find new applications for computers. Provided the operating system doesn't eat up all the gains from hardware development, there should still be plenty more ideas to keep us all busy and spending money.