Skype: Not Open Enough

A couple months back, I attended a dinner organized by Lee Dryburgh (founder of the popular eComm Emerging Communications Conference) and Thomas Howe. The event was a rousing success with many pillars of the VoIP community in attendance. Soon afterwards, the mailing list that had originally been used to organize the dinner venue became a forum for a discussion about whether the success of Skype has obviated the usefullness of the SIP protocol as an enabler of innovation.

Lee's argument in favor of the proposition had three parts. To paraphrase:

(1) Skype has done a better job as a multi-modal client (voice, video, IM and file transfer) than any applications built using the SIP protocol;

(2) the SIP protocol was never designed for, and is therefore ill-suited for, the next generation of communications applications in general (and social networking in particular), and;

(3) the emerging communications industry will stagnate if it continues to view SIP as an acceptable protocol for innovation.

Those falling in the opposing camp (myself included) argued that while the SIP protocol has indeed been overhyped, with unrealistic expectations left unmet, Skype could never displace it because it is not "open."

In response to an article by Jon Arnold mentioning Skype as a walled garden, Lee proceeded to compare the openess of Skype to the "Skype-clone" Gizmo5 (which he treats as an approximate stand-in for the SIP protocol). The crux of his position is that since SIP-based Gizmo5 is no more "open" than Skype, the arguments denouncing Skype based on its lack of openess are nullified.

In support of the contention that Gizmo is not really open, he states that:

With Gizmo5 you cannot build a true peer because the source code is not available. Its method of establishing the P2P overlay is proprietary and so without the source code you cannot build a true peer (it does not use the open P2P SIP standard). Therefore using the same yard stick used to denounce Skype as closed, Gizmo5 is also a "closed network."

And, in defending the accusation that Skype is closed because it does not interoperate with SIP, and will not entertain peering relationships with other VoIP services:

Maybe [Gizmo5] thinks that it's open because it supports the SIP URI names/address space? But I think that would be a weak argument for two main reasons. Firstly because Skype supports not only the private Skype namespace (i.e. a Skype ID) but also the E.164 namespace (i.e. telephone numbers). There is little consumer demand for the additional support of the SIP URI space ... You can call between Skype and Gizmo using the E.164 namespace. I see no reason Skype should have to support the SIP URI namespace to help bolster a competitor! But again, this argument completely lacks vision of the long term evolution in communications, sticking to telephony calls over IP (yawn) being the future.

In the interest of furthering the debate, I feel compelled to weigh in with my two cents.

Industry watchers started proclaiming that 'SIP is dead' in 2005, largely because no one entity had been able to build a game-changing, consumer focused service around it. What was true then is even more true today, with many of the pioneering SIP oriented business ventures either dead or on life support. So, I can understand Lee's attempts to upset the apple cart. It can only be healthy to question the foundation ideas on which billions of dollars have been invested and countless man-hours have been spent. Maybe it is time to ask whether the relative paucity of successful communications applications can be blamed on the SIP protocol itself.

However, while I agree with its spirit, I disaree with the underpinnings of Lee's argument. Openess fosters innovation. But Skype is not open in the same sense that Gizmo5 is open, and far from being open in the sense that the SIP protocol is open. To argue that it is undermines what might otherwise be worthwhile intellectual excercise.

Remember that both Skype and Gizmo5 generate virtually all of their revenue by charging users to connect a PC software application to the PSTN. Though it may be a boring twentieth-century endeavor, for both Skype and Gizmo5, staying in business means selling call-in and call-out credits. The current state of the communications industry is such that consumers are extremely reluctant to pay for anything other than PSTN access.

Against this backdrop, the term "open" can be used in several ways:

Open Source means that the intellectual property used to create software is freely available for anyone to use and/or change. The Asterisk PBX is the best known example of open source communications software. There are many open source SIP software implementations. Neither Skype nor Gizmo5 is open source in this sense.
Open Standard refers to a format or specification that has been agreed upon by concensus, is publicly available for reference, and can only be changed by a governing body through an established process. The SIP protocol (IETF RFC 3261) is an open standard in this sense. The Skype protocol, was created by Skype and is known and controlled only by Skype. There may be some truth to belief held by many that it is a "better" protocol than SIP. But an open standard it is not.
Open Platform refers primarily to a business model where the owner of a network (be it a physical network like Verizon, or a social network such as Facebook) allows third parties to access its users for the purpose of pursuing commercial activities. Openess in this regard can vary greatly depending on the level of access that the network owner grants to third parties. Skype is an open platform to the extent that it publishes an API that allows third parties to create complemetary products/services and conduct commerce with Skype subscribers. Gizmo is an open platform to the extent that it's users are free to call and be called from other VoIP networks. It makes no sense to talk about SIP in terms of an open platform, since it is a protocol.

To claim that Gizmo5 is not open because the source code is not 'open source' misses the point. Because Gizmo5 is built using an 'open standard' and operates as an 'open platform' it is not necessary for third parties to have access to the underlying source code to interact with a Gizmo5 user. If you wish to contact a Gizmo5 user, all you need to know is his SIP address (URI). And a Gizmo5 user wishing to call a PSTN number, simply needs to know the SIP URI of any available SIP gateway. Importantly, Gizmo5 realizes no revenue from these calls to and from outside networks. That a Gizmo5 user can connect with outside networks is made possible because Gizmo5 uses the same open standard SIP protocol that is in common use by other networks as well. That is the power of open standards at work.

Further, claiming that Skype is open because it is possible to reach a Skype user by calling a Skype-In telephone number (ie. the e.164 namespace) is like saying that Verizon is open because anybody can call a Verizon subscriber's cell phone. If viewed in this context alone, I don't know anyone who would argue that Skype is open in any of the three senses enumerated above. The fact remains that the only way for a Skype user to interact with a non-Skype user is by paying for Skype credit.

Just to be clear, I am an enthusiastic fan of Skype. It has single-handedly altered the business of communications, and I believe that its business model is exactly what it needs to be. Having amassed a huge user base, it is a fortunate beneficiary of the inverse network effect theory (which dictates that it should resist opening its network except in the most limited and controlled ways). It is doing this currently with the Skype API, which permits certain applications control of the Skype PC client software to extend Skype functionality (but always in a way where calls to or from the PSTN require Skype credits).

I am also an admirer of Lee, who I first met at the dinner. Nobody in recent memory has done more to coalesce the emerging communications industry around a coherent theme. As a result, eComm is the place where visionaries turn up. And, his attempts to expand the industry's collective consciousness beyond the current status quo is laudable.

However, his argument falls short when it comes to SIP. I like to think of myself as having a pragmatic approach to new ideas. And despite being a proponent of SIP through Solegy and opensourcesip.org, I would be among the first to jump on the bandwagon if a better alternative were to come along. But, as exemplified in Jon Arnold's current list of VoIP companies to watch, the SIP protocol is pretty much the only game in town. And for me, it remains the most useful tool available to build new ways to communicate. The existence of Skype does nothing to diminish its allure.