While financial analysts work out what the Google/eBay deal means, another group will be working on an even harder problem. The agreement will bind together Google Talk and Skype, the two companies' voice over IP (VoIP) systems, leaving the engineers from both sides to cope with their mutual incompatibilities. While Google Talk is largely based on open standards — although far from fully compliant — Skype is as closed as a Hebridean pub on the Sabbath. Can it work with an open standard system without becoming open itself? Logic says no, and nobody's saying how.
This highlights yet another danger of ignoring open standards — the way that a closed system boxes in not just the users but the provider itself. In an industry delineated by big deals and rapidly changing alliances, the ability to share technologies is paramount. By deciding to exclude others, a company is making that job hard if not impossible: not only must a closed system be re-engineered to interoperate with others, its associated business model needs changing too. Closed systems are always driven by business decisions: this always complicates the technical process of providing a service and imposes arbitrary limits on future options.
It is particularly ironic that VoIP — which combines two of the most universal and open systems on the planet, telephony and the Internet Protocol — has spawned quite so many artifical barriers to open connectivity. That is a hangover from the days when telephone companies expected monopolistic powers through divine fiat. The Web is a properly open system: you can choose your client, server software or hosting company, operating system, ISP and DNS registrar, mixing them as you wish. Try doing that with most commercial VoIP systems.
We're 15 years into the Web revolution, and all the buzz continues to stick to open ideas. That's not because of hippy idealism, it's because in a time of rapid change only open systems can hope to adapt fast enough to survive. Any company that finds itself dependent on closed systems should ask itself why it's setting artificial limits to who it can work with and how it will evolve. There may be valid reasons: at least examine the down side. An open approach is good for the market, good for the users — and good for the future of the enterprise.