Open source software might be free to download, but it ends up costing as much as traditional software because of the complexities of supporting it, or inflexible licensing structures. Or so the argument goes.
You would expect to hear this from Steve Ballmer, Larry Ellison or any of the other grand old men of proprietary software. It is more surprising when you hear such thoughts being aired by the head of corporate services at Canonical, the commercial operation behind Ubuntu-flavoured Linux.
Neil Levine is not suggesting open source software has no commercial advantage over closed platforms. Far from it. He is talking about what it will take to get open source to achieve its full potential, bringing the cost savings and business transformation he feels so passionately it can. And that means acknowledging the reasons it has not always done so.
Levine says that the problem is that open source software ended up looking too much like the proprietary software it went in to replace. Or more specifically, owning and running it felt just the same.
"Charging for software is not a very innovative model but it is one that many open source vendors also use," he says. "Even with some of the better known open-source companies, you are ultimately paying a per-machine licence fee to use the software, however obfuscated it may be."
"Open source is boring... it hasn’t been the game-changing phenomenon people were hoping for."
The trick to getting value from an open source deployment is to pay only for the services you need. Paying for every machine is daft, Levine argues. You might need support for your transactional and business critical systems, for example, but not for your development servers because your technical staff will not need help.
How did we end up here? Open source was meant to break down barriers and drive companies like Microsoft and Oracle out of business. The flexibility and speed of development that open source makes possible should have left the big, licence-fee-driven software firms slowly fossilising like so many dinosaurs, its proponents argued. But this has not happened.
Laurent Lachal, open source research director at analyst firm Ovum, thinks open source has won some battles, but the war has pretty much been called off.
"Open source is boring now," he told us. "It used to be exciting but it hasn't been the game-changing phenomenon people were hoping for. Or were afraid of, depending on your stance."
The technical arguments for OSS have been persuasive. Linux is the backbone of many of the world's stock exchanges, a conclusive victory all on its own. But the desktop environment is largely unchanged and still dominated by Windows.
"Open source is just part of the landscape, woven into a lot of software. You could say it has taken root and flowered, but has not displaced the native flora," Lachal says.
For Levine this is a consequence of the reasons put forward for switching to open source. "So much selling had to be done around the technicalities. Open source could claim to be more secure, to have fewer bugs and a faster development path, but ultimately it was me-too software. It had to compete directly: Red Hat against Solaris, Apache against IIS."
No one thought about changing the business model at the same time because that would have been too much change for companies to take on all at once.
But for open source be the disruptive force it promised to be, the business model has to change. To supplant Lachal's "native flora", open-source vendors have to redefine the product and get back to their roots.
"The bits must be free. The value comes in the services and support we sell alongside the software. The cost of the operating system must not be a barrier to adoption," Levine says. "As an industry, we are still having the conversation about how much it costs to run a particular software package. Instead, we should be looking at how the software supports what a company is trying to do. Does it help them to innovate or does it get in the way? Perhaps we should coin a new phrase. It is not about the total cost of ownership, but about the total cost of innovation."
The two are linked, he adds, but to innovate a company has to consider the expense and speed of developing an idea.
"If you need $X to get 10 servers running whatever platform at $Y per machine, the cost could be so high that a great idea costs six figures just to get to the prototype stage. The cost of the software is a friction," says Levine.
Ubuntu is free so Canonical's customers pay for the services and support they need. Generally that is for the servers involved in production. "Those allocated to testing and development won't have any support overheads," Levine says.
He attributes Ubuntu's success in the cloud to its accessible pricing structure. Someone who has an idea, he says, doesn't need to go to the chief executive for a budget. Instead, they can get up and running with Ubuntu, then bring in the services contract once the idea has proved itself.
Levine agrees with Lachal that open source software hasn't fulfilled its potential, but he believes its influence and importance are at a tipping point.
"I think the next few years are going to see that change. Some things are only available from open-source vendors, and a lot of that innovation is being built on Ubuntu," he says. ®