Packaged software, an accident of history
Everyone takes it for granted that the natural order of business is to sell software just as though it were a physical product, shipping it out as a manufactured item and charging a one-time perpetual license fee. In contrast, selling software as a continuously updated service, on a pay-as-you-go subscription, seems like an anomaly. But the accepted status quo is in fact merely a quirk of history, brought about by government action.
It was a ruling in an anti-trust suit against IBM in 1969 that gave birth to the independent software industry, as Rubicon Consulting's Michael Mace relates in a fascinating blog essay published in May. Independent software vendors (ISVs) are so-called precisely because they were defined by their autonomy from the vendors of computing hardware platforms such as IBM. And since the platforms were physical products, the newly independent software vendors adopted the same product metaphor.
In other words, the genesis of the packaged software model and its one-time perpetual license fee was a government-engineered fork in the history of software. Or as Ryan Martens, founder and CTO of Rally Software, put it when he and I were chatting about this when we met in November last year, the concept of software that everyone takes for granted today was designed by an IBM committee at the behest of Washington DC.
That hardware product metaphor also gave rise to the whole waterfall methodology of defining and building complete software products and then launching them into production as finished artefacts, Martens pointed out. Hardware, of course, has to be built like that, but software doesn't, as agile development practices (such as those supported by Rally's application lifecycle management software) allow a more iterative approach. When combined with a SaaS delivery model, the feedback loop to find out how customers are reacting to new functionality for a product like Rally's own shrinks to weeks or days. That compares to Martens' experiences in an earlier role developing licensed platform software at BEA, where it took months or even years before software he'd developed had been deployed by customers and the feedback started trickling in. The release cycle then was eighteen months or more. At Rally, it's eight weeks.
That 1969 anti-trust ruling has a lot to answer for. At the time, it was important to separate software from the underlying platforms so that competition and innovation could flourish, but there's nothing to recommend the continuation of the dysfunctional business model that happened to arise at the time. It introduces unnecessary additional costs and delays into the software value chain, and is inherently slow to respond to customer needs.
Other models are more appropriate to today's connected world. Unencumbered by the computer industry's history, mobile telephony companies have adopted a different route. According to an article in today's FT, Services hold key to Nokia's future, specifically "software-based services." Pay-as-you-go services and frequently updated functionality are the norm for the emerging generation of smart phones (although the mobile carriers do have some legacy issues of their own that are frustrating for users).
Now that computing has evolved into a Web-hosted model, users no longer need to be locked into a given platform to be able to deploy applications, and thus there's nothing anti-competitive about a software vendor also owning and operating the underlying server infrastructure. That frees applications to be delivered on a pay-as-you-go or subscription basis and kept constantly updated by the vendor in an iterative, collaborative relationship with customers. The packaged software era can finally be consigned to history, a temporary anamoly in the evolution of computerized business automation.