The Software Paradox

While it was published over a year ago, RedMonk analyst Stephen O'Grady’s The Software Paradox neatly sums up what many practitioners already know: software revenues, especially when sold with a one time upfront cost, are in decline, likely irreversibly. This is happening at a time when the aggregate value of software has never been greater, hence the paradox.

Stephen points toward many causes including the rise of Open Source and cloud computing, but software also suffers the same problems as other digital media. No matter how expensive it is to create, the cost of reproduction and distribution is approaching free. Pure software vendors are destined to fight the same battle the music industry has for years as distribution costs have gone to zero.

Early in my career I worked for companies which generated their revenues with shrink wrap, pay up front, business models. Even in the 90s it was clear how unsustainable this business model was. Features were added for the sake of features, with the hope that customers would shell out $200/seat for an upgrade. This was never in the best interest of customers or developers.

Bug fixes were cost prohibitive, so software companies had the pressure of creating a “gold master” which required the software to be nearly bug free upon delivery. For customers, to get bug fixes they were forced to buy new versions of the software which often contained superfluous features which were intended to sell the release (for instance flat buttons). Often customers didn’t upgrade (Windows XP) and developers supported software well beyond its shelf life. I believe the subscription model of licensing a fluid product like software, regardless if it is hosted or locally installed, is beneficial both to the producers and the users, and it is surprising companies such as Microsoft have taken so long to adopt it.

With that said, subscription model alone may still not be enough. For most of my career I worked for a business which combined software with data aggregation business into a subscription package. This provided for a very predictable business. While we produced very little data ourselves, the data aggregation service was worth the price of admission for many of our customers.

But ultimately the business died a slow death. As a former member of the management team, I think am at liberty to say we had our problems, but ultimately we failed to adapt to the rise of “full stack” competitors who controlled not only the software stack, but the data as well. It was difficult to come to agreement that data processing was a worthy investment when it appeared that almost all of the profit was in software subscriptions. We thought software would create a wide enough moat, but we were wrong.

It also worthy to note even Open Source proponents are starting to admit that producing Open Source software as a primary business requires awkward business models that are difficult to scale. Stephen quotes an essay from Cloudera founder Mike Olson:

“..it’s pretty hard to build successful, standalone open source company. Notably, no support or services-only business model has ever made the cut..”

Olson is one of the pioneers who attempted to productize Open Source software, first with Sleepycat and now Cloudera. If he doesn’t think the services model (which has been proposed as viable business model for Open Source software products for as long as I can remember) works, he’s probably right.

Stephen does propose some solutions (collect and sell data) for software vendors, but I think it is time to question the viability of standalone commercial software as a product. If the future of pure software distribution is Open Source, which has been proved difficult to monetize, to what end are software developers working? Are they serving themselves or those who package and otherwise leverage their work?